# **VISUAL DYSFUNCTION IN VISUAL DYSFUNCTION IN**

**SCHIZOPHRENIA: A VIEW INTO THE MECHANISMS OF MADNESS? SCHIZOPHRENIA: A VIEW INTO THE MECHANISMS OF MADNESS? VISUAL DYSFUNCTION IN SCHIZOPHRENIA: VIEW THE MECHANISMS OF Topic Editors**

**Topic Editors Steven Silverstein, Brian P. Keane, Randolph Blake, Anne Giersch, Michael Green and Szabolcs Kéri Topic Editors Steven Silverstein, Brian P. Keane, Randolph Blake, Anne Giersch, Michael Green and Szabolcs Kéri Steven Silverstein, Brian P. Keane, Randolph Blake, Michael and Szabolcs** 

#### *FRONTIERS COPYRIGHT STATEMENT*

© Copyright 2007-2015 Frontiers Media SA. All rights reserved.

All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use.

**ISSN** 1664-8714 **ISBN** 978-2-88919-515-2 **DOI** 10.3389/978-2-88919-515-2

# *ABOUT FRONTIERS*

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

# *FRONTIERS JOURNAL SERIES*

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing.

All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

# *DEDICATION TO QUALITY*

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view.

By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

# *WHAT ARE FRONTIERS RESEARCH TOPICS?*

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area!

Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# **VISUAL DYSFUNCTION IN SCHIZOPHRENIA: A VIEW INTO THE MECHANISMS OF MADNESS?**

Topic Editors:

**Steven Silverstein,** The State University of New Jersey, USA **Brian P. Keane,** The State University of New Jersey, USA **Randolph Blake,** Vanderbilt University, USA **Anne Giersch,** Institut National de la Santé et de la Recherche Médicale, France **Michael Green,** University of California, USA **Szabolcs Kéri,** University of Szeged, Hungary

The owner of this image is Steven M. Silverstein.

Research on visual perception in schizophrenia has a long history. However, it is only recently that it has been included in mainstream efforts to understand the cognitive neuroscience of the disorder and to assist with biomarker and treatment development (e.g., the NIMH CNTRICS and RDoC initiatives). Advances in our understanding of visual disturbances in schizophrenia can tell us about both specific computational and neurobiological abnormalities, and about the widespread computational and neurobiological abnormalities in the illness, of which visual disturbances constitute well-studied, replicable, low-level examples. Importantly, far from being a passive sensory registration process, visual perception is active, inferential, and hypothesis-generating, and therefore can provide excellent examples of breakdowns in general brain functions in schizophrenia.

Despite progress made in understanding visual processing disturbances in schizophrenia, many challenges exist and many unexplored areas are in need of examination. For example, the directional relationships between perceptual and cognitive disturbances (e.g., in attention, memory, executive function, predictive coding) remain unclear in many cases, as do links with symptoms, including visual hallucinations. The effect of specific visual disturbances on multisensory integration in schizophrenia has also not been explored. In addition,

few studies of vision in schizophrenia have used naturalistic stimuli, including real-world objects, and almost no studies have examined processing during interaction with objects or visual exploration, which can provide important data on functioning of the perception for action pathway. Relatedly, studies of visual processing in schizophrenia have also not been conducted within contexts that include emotional stimulation and the presence of reinforcers – characteristics of many real-world situations - and the consequences of this are likely to be an incomplete view of how and when perception is abnormal in the condition. An additional important area involves treatment of visual disturbances in schizophrenia. Two major questions regarding this are: 1) can visual processing be improved in cases where it is impaired (and by what types of interventions affecting which cognitive and neurobiological mechanisms)? and 2) what are the clinical and functional benefits of improving specific visual functions in people with schizophrenia? Other important and understudied questions concern: 1) the extent to which indices of visual functioning can serve as biomarkers such as predictors of relapse, treatment response, and/or recovery; 2) the potential role of visual functioning in diagnosing and predicting illness; 3) the extent to which some visual perception disturbances are diagnostically specific to schizophrenia; and 4) the extent to which visual disturbances are truly manifestations of disease, as opposed to aspects of normal variation that, in combination with disease, serves to modify the clinical presentation. This Frontiers Research Topic explores some of these, and other issues facing this exciting interface between vision science and schizophrenia research. We include papers that span the entire range of different Frontiers paper types, including those that are data driven (using psychophysics, electroencephalography, neuroimaging, computational and animal models, and other methods), reviews, hypotheses, theories, opinion, methods, areas of impact, and historical perspectives.

# Table of Contents

# *07 Vision in Schizophrenia: Why it Matters* Steven Silverstein, Brian P. Keane, Randolph Blake, Anne Giersch, Michael Green and Szabolcs Kéri *10 Schizophrenia Spectrum Participants Have Reduced Visual Contrast Sensitivity to Chromatic (Red/Green) and Luminance (Light/Dark) Stimuli: New Insights into Information Processing, Visual Channel Function, and Antipsychotic Effects* Kristin S. Cadenhead, Karen Dobkins, Jessica McGovern and Kathleen Shafer *18 Low Spatial Frequency Bias in Schizophrenia is not Face Specific: When the Integration of Coarse and Fine Information Fails* Vincent Laprevote, Aude Oliva, Anne-Sophie Ternois, Raymund Schwan, Pierre Thomas and Muriel Boucart *24 Neuropsychological Functions and Visual Contrast Sensitivity in Schizophrenia: The Potential Impact of Comorbid Posttraumatic Stress Disorder (PTSD)* Ibolya Halász, Einat Levy-Gigi, Oguz Kelemen, György Benedek and Szabolcs Kéri *30 Visual Surround Suppression in Schizophrenia* Marc S. Tibber, Elaine J. Anderson, Tracy Bobin, Elena Antonova, Alice Seabright, Bernice Wright, Patricia Carlin, Sukhwinder S. Shergill and Steven C. Dakin *43 Visual Context Processing in Bipolar Disorder: A Comparison with Schizophrenia* Eunice Yang, Duje Tadin, Davis M. Glasser, Sang Wook Hong, Randolph Blake and Sohee Park *55 Effects of Short-Term Inpatient Treatment on Sensitivity to a Size Contrast Illusion in First-Episode Psychosis and Multiple-Episode Schizophrenia* Steven M. Silverstein, Brian P. Keane, Yushi Wang, Deepthi Mikkilineni, Danielle Paterno, Thomas V. Papathomas and Keith Feigenson *66 Object Substitution Masking in Schizophrenia: An Event-Related Potential Analysis* Jonathan K. Wynn, Kristopher I. Mathis, Judith Ford, Bruno G. Breitmeyer and Michael F. Green *75 Schizophrenia and Visual Backward Masking: A General Deficit of Target Enhancement*

Michael H. Herzog, Maya Roinishvili, Eka Chkonia and Andreas Brand

*84 Sex, Symptom, and Premorbid Social Functioning Associated with Perceptual Organization Dysfunction in Schizophrenia*

Jamie Joseph, Grace Bae and Steven M. Silverstein

# *93 An Event-Related Potential Examination of Contour Integration Deficits in Schizophrenia*

Pamela D. Butler, Ilana Y. Abeles, Steven M. Silverstein, Elisa C. Dias, Nicole G. Weiskopf, Daniel J. Calderone and Pejman Sehatpour


Heng-Ru May Tan, Luiz Lana and Peter J. Uhlhaas

*135 Are Patients with Schizophrenia Impaired in Processing Non-Emotional Features of Human Faces?*

Hayley Darke, Joel S. Peterman, Sohee Park, Suresh Sundram and Olivia Carter


Jejoong Kim, Daniel Norton, Ryan McBain, Dost Ongur and Yue Chen


Justine M. Y. Spencer, Allison B. Sekuler, Patrick J. Bennett and Bruce K. Christensen


Anne Giersch, Laurence Lalanne, Mitsouko van Assche and Mark A. Elliott

*231 Negative Correlation between Leftward Bias in Line Bisection and Schizotypal Features in Healthy Subjects*

Michele Ribolsi, Giulia Lisi, Giorgio Di Lorenzo, Giuseppe Rociola, Cinzia Niolu and Alberto Siracusano

*240 The Relation between Cognitive-Perceptual Schizotypal Traits and the Ebbinghaus Size-Illusion is Mediated by Judgment Time*

Paola Bressan and Peter Kramer


# Vision in schizophrenia: why it matters

#### *Steven Silverstein1 \*, Brian P. Keane1, Randolph Blake2, Anne Giersch3, Michael Green4 and Szabolcs Kéri <sup>5</sup>*

*<sup>1</sup> Department of Psychiatry, Robert Wood Johnson Medical School, and University Behavioral Health Care, Rutgers, The State University of New Jersey, Piscataway, NJ, USA*

*<sup>3</sup> Department of Psychiatry, University of Strasbourg, Strasbourg, France*

*<sup>4</sup> Department of Psychiatry and Biobehavioral Sciences, University of California at Los Angeles, Los Angeles, CA, USA*

*<sup>5</sup> Department of Physiology, University of Szeged, Szeged, Hungary*

*\*Correspondence: silvers1@ubhc.rutgers.edu*

#### *Edited by:*

*Xavier Noel, Fonds de la Recherche Scientifique - FNRS, Belgium*

*Reviewed by: Wolfgang Tschacher, Universität Bern, Switzerland*

#### **Keywords: schizophrenia, vision, perception, risk, cognition, blindness, brain**

Visual processing impairments are now well established in schizophrenia, including abnormalities in: contrast sensitivity (Kiss et al., 2010; Kelemen et al., 2013); excitatory and inhibitory functions such as those involved in forward and backward masking (Green et al., 2011) and surround suppression (Dakin et al., 2005); perceptual organization (Silverstein and Keane, 2011a); facial emotion recognition (Turetsky et al., 2007) and motion processing (Chen, 2011). There has been little work on color processing to date, but clinical reports indicate frequent descriptions of increased intensity of, or change in colors, in addition to changes in brightness contrast (Vollmer-Larsen et al., 2007). Of etiological relevance, visual distortions (which occur in over 60% of patients) have the highest sensitivity for conversion to a psychotic disorder among all basic symptoms (Klosterkotter et al., 2001). In addition, visual impairments in children of parents with schizophrenia predict later development of the disorder (Schiffman et al., 2006), and visual abnormalities in children in the general population are more strongly associated with the later development of schizophrenia than any other form of sensory impairment (Schubert et al., 2005). Finally, seemingly subtle visual impairments contribute to poorer real-world functioning (Rassovsky et al., 2011; Green et al., 2012). In short, visual changes (e.g., distortions, hallucinations) are common, and they have etiological, pathophysiological, and functional significance. In some cases, they can be viewed as models of impaired neural circuitry that can inform our understanding of the same connectivity problems occurring at larger scales, such as in the frontal lobe, or involving connections between brain regions (Phillips and Silverstein, 2003).

Given this, and the fact that vision is the most studied and best understood function in neuroscience, why is vision such an understudied area in schizophrenia research? (Silverstein and Keane, 2011b). Perhaps it is due to the misperception that visual findings are relatively unimportant aspects of the disorder. Much evidence, including that cited above, and included in this e-book, shows that to be untrue.

The 30 papers included in this volume make important contributions toward clarifying the mechanisms involved in visual impairments, and their relevance for schizophrenia. These are divided into sections on: (a) visual processing impairments in schizophrenia; (b) visual processing impairments in at-risk states, and the implications of data on an inverse relationship between congenital blindness and incidence of schizophrenia; and (c) broader theoretical papers. The first section begins with three papers on low-level visual impairments in schizophrenia, including findings on: (1) the interaction of color and contrast sensitivity effects (Cadenhead et al., 2013) (2) a bias toward low spatial frequency processing in face perception (Laprevote et al., 2013); and (3) the influence of comorbid PTSD on contrast sensitivity in schizophrenia. The next three papers consider inhibitory effects, including: (4,5) surround suppression reductions with a variety of stimuli (Tibber et al., 2013; Yang et al., 2013); (6) the effects of change in clinical status on size contrast (Silverstein et al., 2013); (7) object-substitution masking (Wynn et al., 2013); and (8) a general approach to backwards masking impairment in schizophrenia (Herzog et al., 2013). The next four papers cover issues related to mid-level vision and perceptual organization. These include those on: (9) sex differences and clinical variables related to perceptual organization impairment in schizophrenia (Joseph et al., 2013); (10) an event-related potential marker of contour integration impairment (Butler et al., 2013); (11) neural oscillations and perceptual organization (Spencer and Ghorashi, 2014); and (12) a review of oscillatory activity and its relevance for understanding visual processing in schizophrenia. The next two papers (13,14; Christensen et al., 2013; Darke et al., 2013) address the issue of face processing abnormalities in schizophrenia. The final papers in this section cover topics related to motion processing, eye movements, temporal context processing effects and time perception. These include studies of: 15–17) biological motion perception in schizophrenia (Hastings et al., 2013; Kim et al., 2013; Spencer et al., 2013); (18–19) eye movement and scan pattern abnormalities (Delerue and Boucart, 2013; Sprenger et al., 2013); (20) visual and motor disorganization

*<sup>2</sup> Department of Psychology, Vanderbilt University, Nashville, TN, USA*

(Giersch et al., 2013b); (21) oscillatory markers of abnormal temporal context processing (Dias et al., 2013); and (22) the role of impaired temporal processing in visual processing abnormalities in schizophrenia (Giersch et al., 2013a).

The second section addresses issues related to risk and prevention. These include two papers on the nature of visual processing impairments in schizotypy (23,24; Bressan and Kramer, 2013; Ribolsi et al., 2013); and three papers (25–27) on the hypothesis that congenital blindness serves as a protective factor against schizophrenia, as well as the implications of these data for early cognitive-perceptual training in at-risk populations.

The third section addresses general issues. Skottun and Skoyles (2013) (28) call into question the view that the magnocellular pathway is disproportionally impaired in schizophrenia; they argue instead for a more generalized dysfunction in visual perception. The paper by Yoon et al. (2013) (29) demonstrates how studies of visual processes in schizophrenia can reveal abnormalities in general computational processes in schizophrenia. Finally, Phillips and Silverstein (2013) (30) argue that a general mechanism of context-sensitive gain control is basic to cognition and perception and is impaired in schizophrenia, which can account for many observed findings in the disorder.

Taken together, these papers provide a representative example of current work in the vision science of schizophrenia. Our hope is that this set of papers will be of use to those currently working in this field, and will stimulate others to investigate these issues. Findings addressing these questions would be of major benefit to the field of schizophrenia research, and would also inform the study of normal visual perception.

#### **REFERENCES**


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 21 November 2014; accepted: 09 January 2015; published online: 05 February 2015.*

*Citation: Silverstein S, Keane BP, Blake R, Giersch A, Green M and Kéri S (2015) Vision in schizophrenia: why it matters. Front. Psychol. 6:41. doi: 10.3389/fpsyg. 2015.00041*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2015 Silverstein, Keane, Blake, Giersch, Green and Kéri. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Schizophrenia spectrum participants have reduced visual contrast sensitivity to chromatic (red/green) and luminance (light/dark) stimuli: new insights into information processing, visual channel function, and antipsychotic effects

# *Kristin S. Cadenhead1\*, Karen Dobkins2, Jessica McGovern2 and Kathleen Shafer <sup>1</sup>*

*<sup>1</sup> Department of Psychiatry, University of California San Diego, La Jolla, CA, USA*

*<sup>2</sup> Department of Psychology, University of California San Diego, La Jolla, CA, USA*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Szabolcs Kéri, University of Szeged, Hungary Eunice Yang, University of California, Berkeley, USA*

#### *\*Correspondence:*

*Kristin S. Cadenhead, Department of Psychiatry, University of California San Diego, 9500 Gilman Drive, 0810 La Jolla, CA 92093-0810, USA e-mail: kcadenhead@ucsd.edu*

**Background:** Individuals with schizophrenia spectrum diagnoses have deficient visual information processing as assessed by a variety of paradigms including visual backward masking, motion perception and visual contrast sensitivity (VCS). In the present study, the VCS paradigm was used to investigate potential differences in magnocellular (M) vs. parvocellular (P) channel function that might account for the observed information processing deficits of schizophrenia spectrum patients. Specifically, VCS for near threshold luminance (black/white) stimuli is known to be governed primarily by the M channel, while VCS for near threshold chromatic (red/green) stimuli is governed by the P channel.

**Methods:** VCS for luminance and chromatic stimuli (counterphase-reversing sinusoidal gratings, 1.22 c/degree, 8.3 Hz) was assessed in 53 patients with schizophrenia (including 5 off antipsychotic medication), 22 individuals diagnosed with schizotypal personality disorder and 53 healthy comparison subjects.

**Results:** Schizophrenia spectrum groups demonstrated reduced VCS in both conditions relative to normals, and there was no significant group by condition interaction effect. *Post-hoc* analyses suggest that it was the patients with schizophrenia on antipsychotic medication as well as SPD participants who accounted for the deficits in the luminance condition.

**Conclusions:** These results demonstrate visual information processing deficits in schizophrenia spectrum populations but do not support the notion of selective abnormalities in the function of subcortical channels as suggested by previous studies. Further work is needed in a longitudinal design to further assess VCS as a vulnerability marker for psychosis as well as the effect of antipsychotic agents on performance in schizophrenia spectrum populations.

**Keywords: schizophrenia, schizotypal, visual contrast sensitivity, magnocellular, parvocellular**

# **INTRODUCTION**

Patients with schizophrenia (SZ), their first degree relatives and individuals diagnosed with schizotypal personality disorder (SPD) have deficient visual information processing as assessed by a variety of paradigms including visual backward masking (VBM), motion perception, spatial localization, eye tracking and visual contrast sensitivity (VCS) (Braff and Saccuzzo, 1981; Schwartz and Winstead, 1985; Schwartz et al., 1987, 1988; Green et al., 1994; O'Donnell et al., 1996; Saccuzzo et al., 1996; Cadenhead et al., 1998; Slaghuis, 1998; Chen et al., 1999; Green and Nuechterlein, 1999; Slaghuis and Curran, 1999; Keri et al., 2000, 2002; Butler et al., 2001). With increasing knowledge of information processing in the visual system, it has been possible to apply newer paradigms in SZ spectrum populations in order to specify the underlying mechanisms and possible neural substrates responsible for the observed deficits.

Physiological and anatomical data have documented the existence of two major subcortical pathways that originate in the eye and project to primary visual cortex, the magnocellular (M) and parvocellular (P) pathways (Breitmeyer and Ganz, 1976; Lennie, 1980; Kaplan and Shapley, 1986; Livingstone and Hubel, 1987; Bassi and Lehmkuhle, 1990). The M pathway is sensitive to low spatial frequency patterns, high temporal frequencies, and exhibits transient (on/off) responses to visual stimuli (Breitmeyer and Ganz, 1976; Lennie, 1980; Kaplan and Shapley, 1986; Livingstone and Hubel, 1987; Bassi and Lehmkuhle, 1990). By contrast, the P pathway is sensitive to high spatial frequency patterns, low temporal frequencies, and shows sustained

responses to stimuli (Breitmeyer and Ganz, 1976; Lennie, 1980; Kaplan and Shapley, 1986; Livingstone and Hubel, 1987; Bassi and Lehmkuhle, 1990). Most relevant to the current study, the M pathway exhibits high contrast sensitivity to luminance (light/dark) patterns, and low contrast sensitivity to chromatic (red/green) patterns; conversely, the P pathway exhibits high contrast sensitivity to chromatic (red/green) patterns and low contrast sensitivity to luminance (light/dark) patterns (Lee et al., 1990; Shapley, 1990; Smith et al., 1995; Dobkins et al., 2009). In terms of cortical projections, the M pathway provides input to both the dorsal processing stream (the "where" system, which processes motion and space information) and the ventral processing stream (the "what" system, which processes object information) (Maunsell et al., 1990; Thiele et al., 2001), while the P pathway provides input primarily to the ventral processing stream.

To investigate the integrity of the M and P pathways in SZ, various psychophysical and neurophysiological approaches have been used that tap into the known visual properties of the two, or tap into the dorsal/ventral pathways that receive asymmetrical input from the M and P subcortical pathways (see above). In an early visual information processing study in SZ patients, Schwartz and Winstead (1982) assessed visible persistence by sequentially presenting sine-wave gratings of varying spatial frequencies, in order to differentially stimulate the parallel visual pathways. Both acute and chronic patients with SZ required longer inter-stimulus intervals (ISI's) than controls, with chronic patients performing most poorly when exposed to low spatial frequency stimuli that were utilized to bias activity in the magnocellular pathway. These findings were interpreted as either abnormal visual pathway (mainly M pathway) activity or a dysfunction in inhibitory mechanisms between the pathways that prevented the termination of visible persistence (Schwartz and Winstead, 1982).

Visual Backward Masking (VBM) paradigms have been developed that use different spatial frequency masks or require participants to locate (dorsal stream function) or identify (ventral stream function) target stimuli (Balogh and Merritt, 1987; Green et al., 1994; Cadenhead et al., 1998; Slaghuis and Curran, 1999). In the VBM paradigm, patients with SZ have been shown to exhibit deficits in the location, but not the identification, task when compared to normal comparison participants (Green et al., 1994; Cadenhead et al., 1998). Because locating targets in space is thought to be a dorsal stream function, and because the dorsal stream receives primarily M pathway input, this result has been interpreted as evidence of abnormal M pathway (subcortical) or dorsal stream (cortical) processing in SZ. As noted, there is considerable overlap between the M and P visual channels as early as the primary visual cortex (V1) and this interaction extends into the dorsal and ventral streams occurring thereafter (Kovacs et al., 1995; Sawatari and Callaway, 1996; Keri et al., 2000; Butler et al., 2007; Skottun and Skoyles, 2007) making it difficult to conclude from the VBM paradigm alone whether there are specific subcortical pathway deficits.

Perhaps one of the most straightforward ways in which M vs. P pathway processing in SZ has been investigated has been to measure VCS to stimuli designed to activate one pathway more than the other (Schwartz and Winstead, 1985, 1997; Schwartz et al., 1987, 1988; Slaghuis, 1998; Chen et al., 1999; Keri et al., 2000, 2002; Butler et al., 2001, 2009; Kiss et al., 2010; Kent et al., 2011; Halasz et al., 2013). VCS is defined as the inverse of the contrast needed in a stimulus (typically, a sinusoidal grating) in order for that stimulus to be just barely detectable, referred to as "contrast threshold." The advantage of using the VCS paradigm is that it can *isolate* activity in a given pathway. The logic behind this approach is based on the fact that most stimuli will stimulate neurons in both the M and P pathways, for example, a high temporal frequency stimulus will activate both M and P neurons. However, given that one pathway is more sensitive to that stimulus (e.g., M neurons are more sensitive to high temporal frequency than are P neurons), one can assume that—*at contrast threshold*, only the more sensitive pathway (in this example, the M pathway) is responding to the stimulus (Smith and Edgar, 1991; Dobkins and Albright, 1995).

Using this approach, previous studies have measured VCS to different spatial/temporal frequencies, designed to isolate the M vs. P pathways. Slaghuis (1998; Slaghuis and Curran, 1999) examined VCS for stationary and moving stimuli across a range of spatial frequencies and reported that SZ patients with predominantly positive symptoms exhibited deficits at medium and high spatial frequencies, suggesting P pathway dysfunction, while the patients with predominantly negative symptoms performed poorly at all spatial frequencies, suggesting both M and P pathway dysfunction. Kéri and colleagues (2002) showed that patients with SZ have reduced VCS in the medium to high spatial frequency range in a stationary condition and over the full spatial frequency range in a moving condition. While the VCS dysfunction did not appear to be specific to either the M or P channel in the Kéri et al. study, more severe VCS deficits were associated with antipsychotic dose and Simpson Angus ratings of extrapyramidal side effects. Chen and colleagues (1999, 2003) did not find VCS deficits in SZ patients tested at low spatial frequency and varying temporal frequencies but reported that unmedicated patients (*N* = 6) exhibited VCS that was *greater* than that of normal participants. Furthermore, those patients on atypical antipsychotics were unimpaired while patients on typical antipsychotics had lower VCS. More recently, Kiss et al. (2010) have also reported that never medicated first episode patients with SZ (*N* = 20) have *greater* VCS in a pedestal condition that emphasizes M channel activity. The finding of reduced VCS in patients with SZ on antipsychotics in the Kéri and Chen et al. studies is similar to observations of reduced VCS in patients with Parkinson's Disease, suggesting that some of the effect may be secondary to modulation of spatiotemporal VCS functions by a hypodopaminergic state (Bulens et al., 1986, 1987, 1989; Bodis-Wollner et al., 1987). O'Donnell et al. (2006), however, found that both medicated and unmedicated patients with SZ were impaired relative to normals across both high and low spatial frequencies. In the same study, individuals meeting the DSM-IV criteria for SPD did not differ from normals. Similarly, Kent et al. (2011) reported that SPD patients had VCS deficits at low temporal frequencies. Using a visual evoked potential paradigm, Butler and colleagues (Butler et al., 2001, 2009) examined near threshold luminance and chromatic contrast (to emphasize M or P visual pathway activity, respectively) in patients with SZ in two studies who showed lower response levels to stimuli that were M biased (low luminance contrast with large squares) while P biased stimuli (low chromatic contrast) did not differentiate them from normals. All but 2 of the participants in the Butler et al. study were receiving antipsychotic medications at the time of testing; therefore, a medication effect could not be excluded.

In summary, it is not clear from the current literature whether the observed visual information processing deficits in SZ spectrum patients are specific to M vs. P visual channel dysfunction. An important finding that has emerged is that dopamine modulation via D2 receptor blockade likely affects VCS in SZ patients and may account for the observed deficits across a range of paradigms. In the current study, we likewise capitalized on M and P pathway differences in luminance and chromatic contrast sensitivities, respectively. To this end, VCS was measured in patients with SZ and individuals meeting criteria for SPD. The SPD participants are important because they do not have many of the confounding variables such as acute psychosis or medication effects seen in a chronic SZ population yet represent a SZ spectrum phenotype. It was predicted that patients diagnosed with SZ and individuals meeting the criteria for SPD would have lower contrast sensitivities consistent with other studies but we wanted to determine whether these differences were specific to either visual channel. Finally, to better understand the relationship of antipsychotic medication to VCS, the effect of antipsychotic medication were examined separately in *post-hoc* analyses.

# **MATERIALS AND METHODS PARTICIPANTS**

Participants included 53 (40M/13F) patients with SZ, 22 (11M/11F) individuals who met criteria for SPD and 53 (30M/23F) healthy comparison (HC) subjects. Participants provided written informed consent after receiving an explanation of the study. Individuals with a history of major medical or neurological disorders or significant drug abuse in the past were excluded. Additionally, all participants were screened for current drug use using urine toxicology tests. HC subjects were recruited through newspaper advertisements and had no history of Axis I or II disorders as assessed by the Structured Clinical Interview for DSM-IV Disorders (SCID I and SCID II) nor any family history of SZ in a first or second degree relative by self report. Established inter-rater reliability for the SCID in our laboratory is 0.98 (Perry and Braff, 1998).

SZ patients were recruited through inpatient and outpatient facilities at UCSD, a long term care facility, and the San Diego Alliance for the Mentally Ill. All participants received the SCID to confirm the diagnosis of SZ. Five SZ patients were not receiving antipsychotic agents.

SPD participants were recruited from outpatient facilities at UCSD and by newspaper advertisements per our established methods (Cadenhead et al., 2000). Additional SPD participants were identified through screening of potential normal comparison participants. All SPD participants were assessed with the SCID I and the Structured Interview for DSM-IV Personality Disorders (SIDP) by one of the investigators (KSC) to identify the diagnosis of SPD. Two SPD participants were receiving a low dose typical antipsychotic (perphenazine 6 Mg qd or trifluoperazine 2 mg).

All participants were screened for color blindness using Ishihara's tests for color deficiency (Ishihara, 1986) and were reported to have corrected visual acuity of 20/50 or better as measured by the Snellen Eye Chart.

# **VISUAL CONTRAST SENSITIVITY**

#### *Apparatus and stimuli*

Stimuli were generated using a Power Macintosh 8100/80 PC computer and Nanao T2-17TS FlexScan Color Monitor (1152 × 870 pixels, 75 Hz). The 8-bit video board allowed for 256 discrete levels of luminance. The CIE coordinates for the monitor primaries were: Red (0.61, 0.342), Green (0.298, 0.588), and Blue (0.151, 0.064). The maximum output for the monitor was calibrated to equal energy white (CIE chromaticity coordinates = 0. 333, 0.333), and the voltage/luminance relationship was linearized independently for each of the three guns in the display, using a PR-650 Colorimeter (Photoresearch). The PR-650 was used for photometric measurements to standardize to Vλ isoluminance, as well as for spectroradiometric measurements to compute L and M cone modulations produced by our visual stimuli.

Stimuli were 1.22 cycles/degree horizontally-oriented sinusoidal gratings moving at a temporal frequency 8.33 Hz and are either luminance (light/dark) or chromatic (red/green). Each trial consisted of a luminance or chromatic stimulus, subtending a visual angle of 1.64 degrees (a total of 2 cycles), centered 2.28 degree to the left or right of a fixation cross in screen center, for a duration of 120 Ms. Motion was produced by phase-shifting sinusoidal gratings at regular intervals in sync with the vertical refresh of the video monitor (75 Hz).

*Luminance (light/dark) gratings.* Luminance-defined gratings were produced by sinusoidally modulating the red and green phosphors in phase (with a small amount of blue gun also added in phase to match the mean chromaticity of the chromatic gratings). For luminance stimuli, r.m.s. (root mean square) cone contrast values directly correspond to the conventional Michelson contrast: [(Lmax − Lmin)/(Lmax + Lmin)], and cone contrasts up to 100% are readily produced. Although the calibration techniques allow for a specified amount of contrast, the actual luminance contrast of all stimuli was verified using the PR-650 Colorimeter.

*Chromatic (red/green) gratings.* Chromatic red/green gratings were produced by sinusoidally modulating the red and green phosphors 180 degrees out of phase, with a small amount of blue primary added in phase with the red portion of the grating so as to prevent modulation of the short-wavelength-sensitive (S) cones (Dobkins and Teller, 1996). The cone contrast of these stimuli was determined by spectroradiometry (with the PR-650) as described in detail previously (Gunther and Dobkins, 2002).

The red/green isoluminance point for each participant was determined using a 20 trial task during which participants were asked to adjust (increase or decrease) the luminance contrast (interval step = 0.5%) in a moving red/green grating (r.m.s. cone contrast = 7.2%) to the point where the pattern was no longer salient (not visible, faded, jagged or blurry). The mean isoluminance point was used in the chromatic portion of the contrast sensitivity paradigm (below).

#### *Psychophysical paradigm*

The VCS paradigm consisted of a standard 2-alternative-forcedchoice task with feedback. Participants sat in a comfortable chair and rest their head in a chin rest. They were asked to fixate on a small "+" symbol in the center of a monitor 57 cm away. The participant started each trial with a key press on the standard computer keyboard, after which a grating stimulus (luminance or chromatic) of varying contrast appeared on the left or right side of the display. After the stimulus disappeared, the participant reported its location (left or right) using color-coded keys. Using a method of constant stimuli paradigm, six different levels of contrast were presented for each condition, presented randomly across trials. Participants were given 100 practice trials to familiarize them with the task. The experimental session contained 360 trials (180 trials for each the luminance and chromatic conditions) and lasted approximately 45 Min. The experiment was self-paced and participants were encouraged to take breaks as often as needed.

#### *Data analysis*

*Contrast sensitivities.* Psychometric curves were fit to data using Weibull functions and maximum likelihood analysis (Weibull, 1958; Watson, 1979). Contrast threshold was defined as the contrast yielding 75% correct performance. In extreme cases where the performance of a subject was poor in both the luminance and chromatic conditions so that no threshold could be determined from the Weibull, the subject was excluded (1 SPD and 7 SZ). All thresholds were analyzed in terms of r.m.s. cone contrast. Contrast sensitivity was determined from the inverse of threshold (i.e., sensitivity = 1/threshold). Raw contrast sensitivity data was log10 transformed, because log but not linear sensitivities conform to normal distributions. Data were analyzed in Two-Way ANOVAs that included the stimulus type (luminance and chromatic) and participant group.

# **RESULTS**

#### **DEMOGRAPHICS**

**Table 1** presents the demographic data of subjects included in the analysis. The groups did not differ in age [*F(*2*,* <sup>127</sup>*)* = 1*.*8, ns] or ratio of males to females (Pearson Chi Square Value = 5.3, ns) but education differed between the groups [*F(*2*,* <sup>119</sup>*)* = 22*.*4, *p <* 0*.*001] with SZ patients having significantly fewer years of completed education compared to SPD and HC participants (both *p <* 0*.*001). As expected, SZ patients had poorer functioning (GAF: *t* = 3*.*9, *p <* 0*.*001) and more symptoms (SANS: *t* = 4*.*9, *p <* 0*.*001; SAPS: *t* = 2*.*8, *p <* 0*.*001) relative to SPD subjects.

#### **GROUP MEAN CONTRAST SENSITIVITY**

The results of a Two-Way ANOVA revealed a significant main effect of participant group [*F(*2*,* <sup>118</sup>*)* = 8*.*81, *p <* 0*.*001]. *Post-hoc*

#### **Table 1 | Demographics and visual contrast sensitivities.**


*GAF, global assessment of functioning; SAPS, schedule for the assessment of positive symptoms; SANS, schedule for the assessment of negative symptoms.*

*t*-tests indicated that this effect was driven by significantly lower contrast sensitivity (collapsed across luminance and chromatic) in both SZ spectrum groups as compared to HC subjects (*p <* 0*.*05 vs. SPD and *p <* 0*.*001 vs. SZ). The SZ and SPD participants did not significantly differ from each other. There was no interaction between participant group and stimulus condition [*F(*2*,* <sup>117</sup>*)* = 0*.*56, ns], indicating that the group differences did not differ between luminance and chromatic stimuli. There were no significant gender or interaction effects with gender, and therefore groups were collapsed across gender for the analyses presented below. Visual contrast sensitivity in the luminance but not the chromatic condition was significantly correlated with age while neither condition was associated with symptoms or functioning measures above.

#### **MEDICATION EFFECTS**

Because of the potential effects of antipsychotic agents on visual information processing, we conducted further analyses in which the 2 SPD participants on antipsychotic medication were removed from the analysis and the 41 patients with SZ on antipsychotics (SZ+AP) were compared to 5 patients with SZ who were not on antipsychotics (SZ-AP), the 53 HC and 18 SPD subjects (see **Figures 1**, **2**). A two-factor ANOVA revealed a significant effect of group [*F(*3*,* <sup>114</sup>*)* = 6*.*77, *p <* 0*.*001], driven by HC vs. SZ patients on antipsychotics (*p <* 0*.*001) and HC vs. SPD (*p <* 0*.*05), and a significant group by condition interaction [*F(*3*,* <sup>114</sup>*)* = 3*.*86, *p <* 0*.*01]. To investigate this interaction further, we performed separate one-factor ANOVAs for the luminance and chromatic conditions. Both analyses showed a significant main effect of group [luminance: *F(*3*,* <sup>117</sup>*)* = 7*.*56, *p <* 0*.*001; chromatic: *F(*3*,* <sup>117</sup>*)* = 5*.*03, *p <* 0*.*005], but for different reasons. For the luminance condition, *post-hoc t*-tests revealed impaired luminance contrast sensitivity in the SZ patients on antipsychotics as compared to HC (*p <* 0*.*001), yet *higher* luminance contrast sensitivity in unmedicated SZ patients vs. antipsychotic medicated patients with SZ (*p <* 0*.*005), and no significant difference between unmedicated SZ patients and HC subjects. Inspection of the data reveals that the unmedicated SZ patient have greater luminance contrast sensitivity than HCs (effect size = 0.62) but this analysis was not significant (*p* = 0*.*16). Power analyses indicate that at least 12 subjects per group would be needed to achieve a statistically significant result (power 0.80, *p <* 0*.*05). For the chromatic condition, *post-hoc* tests revealed impaired chromatic contrast

**FIGURE 1 | Visual contrast sensitivity (error bars represent SEM) in the luminance condition comparing healthy comparison (HC), schizotypal personality disorder (SPD), and schizophrenia (SZ) subjects.** Data from unmedicated SZ subjects as well as SZ subjects on antipsychotic (AP) medication are also shown relative to the original group analysis.

sensitivity in antipsychotic treated SZ patients as well as unmedicated patients with SZ and SPD subjects, as compared to HC subjects (*p <* 0*.*001, *p <* 0*.*05, and *p <* 0*.*05, respectively) while the antipsychotic treated vs. unmedicated patients did not differ from each other. Follow-up correlations between chlorpromazine equivalents and contrast sensitivities within the antipsychotic treated sample of patients were not significant. In summary, patients with SZ on antipsychotic medication have impaired VCS in both luminance and chromatic conditions and the small group of unmedicated patients had normal luminance contrast and impaired chromatic contrast. Or stated differently, if we could remove the effects of medication in SZ patients, they might as a group show atypically high luminance contrast sensitivity, yet impaired chromatic contrast sensitivity.

#### **DISCUSSION**

Schizophrenia spectrum groups demonstrated reduced VCS in both luminance and chromatic conditions relative to healthy subjects, but the relative sensitivity to luminance vs. chromatic stimuli did not differ between groups. These results replicate previous findings of visual information processing deficits in SZ spectrum populations but do not support the notion of selective abnormalities in the function of the subcortical M pathway as suggested by previous studies (Butler et al., 2007). Instead, the results suggest either abnormalities in both M and P pathways or a more general visual processing deficit at some point further downstream in individuals diagnosed with SZ spectrum illness. The M and P pathways are responsible for relaying specific visual information from the retina, through the lateral geniculate nucleus of the thalamus, to primary visual cortex (V1), and elsewhere in the cortex. A contrast detection deficit may arise from dysfunction at any of these levels.

As noted by previous investigators (Bodis-Wollner et al., 1982; Bulens et al., 1986, 1987, 1989; Keri et al., 2002; Chen et al., 2003), the possibility of medication effects accounting for the visual information processing deficits cannot be entirely ruled out. The differences between the SZ patients and the HC subjects in the luminance VCS condition were no longer present when the small sample of patients (*N* = 5) not receiving antipsychotics were compared to HCs. In fact, in the luminance condition, the five unmedicated SZ patients had contrast sensitivities *greater than* those of the antipsychotic treated patients (*p <* 0*.*005) and HC (*p* = 0*.*16, *d* = 0*.*62) subjects, consistent with the findings by Chen and colleagues (2003) in six unmedicated SZ patients and the twenty unmedicated SZ patients in the Kiss et al.study (2010). It is possible that the hypodopaminergic effect of the antipsychotics accounted for the original difference between groups in the luminance condition while a hyperdopaminergic state may increase contrast sensitivity as observed in the unmedicated patients with SZ. Interpretation of the current findings in unmedicated patients, are limited by the small sample size but add to a growing literature that suggests evidence of overactive M channel activity.

It is known that dopamine neurotransmission (specifically, of the horizontal, amacrine, and interplexiform cells) in the retina is involved in regulating the strength of lateral inhibition and center-surround antagonism (Tagliati et al., 1994; Sannita, 1995; Djamgoz et al., 1997). A hyperdopaminergic state may enhance the center-surround processing which could in turn, enhance contrast sensitivity. Supporting the role of dopaminemodulation effects, Bodis-Wollner and colleagues (1982) found that Parkinson's disease patients treated with L-dopa and unmedicated SZ patients had faster visual evoked potentials than Parkinson's patients without L-dopa and SZ patients treated with typical antipsychotics. Similarly, a study by Harris and colleagues (1990) showed that after a therapeutic injection of depot antipsychotic, unmedicated patients with SZ (*N* = 8) had enhanced sensitivity at low (0.5 c/degree), and reduced sensitivity at medium (2 c/degree) and high (8 c/degree) spatial frequencies.

In contrast to the findings in SZ patients, the SPD participants, who were not taking antipsychotic medications, had deficits in both luminance and chromatic conditions supporting the notion of general visual information processing deficits in SZ spectrum populations that are not accounted for by medication effects. These results differ from the findings of O'Donnell and colleagues (2006) who reported that individuals with SPD did not differ at any spatial or temporal frequency from healthy subjects and performed significantly better than individuals with SZ. The results may be due to methodological differences. O'Donnell and colleagues did not have a chromatic condition and had a larger age range. Consistent with our findings, Kent and colleagues (2011) did find deficits in SPD at all spatial frequencies but the deficits were more prominent in a pedestal condition that emphasizes magnocellular activity. Whether or not SPD subjects have visual contrast deficits, it remains possible that a hyperdopaminegic state of unmedicated psychotic patients may be enhancing contrast sensitivity while SPD patients, who are characterized by only mild subsyndromal psychotic-like symptoms, tend to have more prominent deficit symptoms that may reflect hypodopaminergia that is accounting for the VCS deficits (Siever and Davis, 2004).

Given the present findings in SPD, it remains possible that the visual information processing deficits as indexed by VCS could represent trait markers for SZ spectrum illness. The identification of trait markers for SZ has implications for genetic studies as well as early identification of individuals at risk for SZ (Cadenhead, 2002). If reliable trait markers can be utilized in conjunction with clinical risk factors in identification of individuals in the prodromal stage of SZ, it may be possible to intervene earlier and prevent many of the devastating effects of a first psychotic break. Further understanding of the underlying neuropathology in the developing brain of individuals in the early stages of SZ could shed insight into targets for early intervention.

Methodological differences may play a role in discrepancies across VCS studies. Different visual contrast studies vary in how sensitivity is measured (e.g, staircase method, method of constant stimulus), stimuli (random dots vs. sinusoidal gratings), how the stimuli are presented (side-by-side or sequentially) and which

#### **REFERENCES**


spatial and temporal frequencies are in the range for biasing processing toward the M vs. P pathways. Further, whereas the present study used luminance (light/dark) and chromatic (red/green) stimuli to isolate M and P pathway activity, other studies varied spatial and temporal frequencies.

Moreover, there is a large variance between studies among the SZ spectrum participants included. Studies differ in participants' ages (O'Donnell et al., 2006), whether participants are chronic vs. acute case (Keri and Benedek, 2007), how long they have been on or off antipsychotics (Chen et al., 2003), and if positive vs. negative symptoms are analyzed separately (Slaghuis and Thompson, 2003). Furthermore, growing research has shown that SZ spectrum patients may vary in their dopamine and other neurotransmitter levels at different points during their illness (Uchida and Mamo, 2009). With increasing age, there is also a decline in D2 receptor binding (Antonini and Leenders, 1993). All of these factors suggest that chronic SZ coupled with years of dopamine-blocking antipsychotics may contribute to a greater hypodopaminergic state.

Clearly, further work is needed using a longitudinal design to chart the course of visual information processing deficits both on and off antipsychotic medication over the course of SZ from prodrome to acute to chronic forms of the illness. The use of VCS paradigms in combination with functional measures of brain activity may help to determine the mechanism by which such deficits occur and lead to better treatments that target specific information processing deficits without causing further impairment.

#### **ACKNOWLEDGMENTS**

National Institute of Mental Health grants MH60720, MH01123 and the Department of Veterans Affairs VISN 22 Mental Illness Research Education and Clinical Center (MIRECC) provided support.


*Biol. Psychiatry* 43, 132–138. doi: 10.1016/S0006-3223(97)00316-8


posttraumatic stress disorder (PTSD). *Front. Psychol.* 4:136. doi: 10.3389/fpsyg.2013.00136


across the life-spectrum. *Prog. Neuropsychopharmacol. Biol. Psychiatry*. 104, 127–144. doi: 10.1016/j.pnpbp.2009.04.023

Watson, A. B. (1979). Probability summation over time. *Vis. Res.* 19, 515–522. doi: 10.1016/0042- 6989(79)90136-6

Weibull, C. (1958). *The Distribution of Reciprocal Choices in Sociometric Tests*. Göteborg: Almqvist and Wiksell Stockholm.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 29 April 2013; accepted: 30 July 2013; published online: 20 August 2013. Citation: Cadenhead KS, Dobkins K, McGovern J and Shafer K (2013) Schizophrenia spectrum participants* *have reduced visual contrast sensitivity to chromatic (red/green) and luminance (light/dark) stimuli: new insights into information processing, visual channel function, and antipsychotic effects. Front. Psychol. 4:535. doi: 10.3389/fpsyg. 2013.00535*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Cadenhead, Dobkins, McGovern and Shafer. This is*

*an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Low spatial frequency bias in schizophrenia is not face specific: when the integration of coarse and fine information fails

#### **Vincent Laprevote1,2\*, Aude Oliva<sup>3</sup> , Anne-Sophie Ternois <sup>4</sup> , Raymund Schwan1,2,5, Pierre Thomas 4,6 and Muriel Boucart <sup>6</sup>**

<sup>1</sup> Centre d'Investigation Clinique-INSERM 9501, CHU Nancy, Nancy, France

<sup>2</sup> Centre de Soins, d'Accompagnement et de Prévention en Addictologie, CHU Nancy, Nancy, France

<sup>3</sup> Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA, USA

<sup>4</sup> Pôle de Psychiatrie, Centre Hospitalier Régional Universitaire de Lille, Lille, France

<sup>5</sup> Faculté de Médecine, Université Lorraine, Nancy, France

<sup>6</sup> Laboratoire de Neuroscience Fonctionnelle et Pathologies EA 4559, Université Lille-Nord de France, Lille, France

#### **Edited by:**

Anne Giersch, Institut National de la Santé et de la Recherche Médicale, France

#### **Reviewed by:**

Michael Herzog, École Polytechnique Fédérale de Lausanne, Switzerland Jonathan K. Wynn, University of California Los Angeles, USA

#### **\*Correspondence:**

Vincent Laprevote, Centre d'Investigation Clinique CIC-INSERM 9501, CHU Nancy, Bâtiment Louis Matthieu, Hôpitaux de Brabois, F-54000 Nancy, France. e-mail: v.laprevote@chu-nancy.fr

Studies have shown that patients with schizophrenia exhibit visual processing impairments, particularly regarding the processing of spatial frequencies. In a previous work, we found that, compared to healthy volunteers, patients were biased toward low spatial frequencies (LSF) to identify facial expression at a glance. Given the ubiquity of faces in visual perception, it remains an open question whether the LSF bias is face specific or also occurs with other visual objects. Here, 15 patients with schizophrenia and 11 healthy control adults performed a categorization task with hybrid stimuli. These stimuli were single images consisting of two different objects, a fruit and an animal, each in a specific spatial frequency range, either low (LSF) or high (HSF). Observers were asked to report if they saw an animal or a fruit. The reported category demonstrated which spatial scale was preferentially perceived in each trial. In a control experiment, participants performed the same task but with images of only a single object, either a LSF or HSF filtered animal or fruit, to verify that participants could perceive both HSF or LSF when presented in isolation. The results on the categorization task showed that patients chose more frequently LSF with hybrid stimuli compared to healthy controls. However, both populations performed equally well with HSF and LSF filtered pictures in the control experiment, demonstrating that the LSF preference found with hybrid stimuli in patients was not due to an inability to perceive HSF. The LSF preference found in schizophrenia confirms our previous study conducted with faces, and shows that this LSF bias generalizes to other categories of objects. When a broad range of spatial frequencies are present in the image, as in normal conditions of viewing, patients preferentially rely on coarse visual information contained in LSF. This result may be interpreted as a dysfunction of the guidance of HSF processing by LSF processing.

**Keywords: schizophrenia, spatial frequency, vision, hybrid image, magnocellular, object**

#### **INTRODUCTION**

How do patients with schizophrenia use visual information to form a coherent representation of the world? Schizophrenia does not only impair high-level cognitive functions, such as executive functions (Dickinson et al., 2007), emotion recognition (Morris et al., 2009), or theory of mind (Sprong et al., 2007). It also impairs the processing of low level perceptual information such as spatial frequencies (Slaghuis, 1998; Butler et al., 2005). Visual spatial frequencies are often considered as the atomic element of perception (Valois and Valois, 1990). The lowest spatial frequencies (below 1.5 cycle/degree) contain a coarse representation of the visual stimuli and are preferentially conveyed by the magnocellular pathway. The detailed information contained in high spatial frequencies (HSF) is primarily processed by the parvocellular pathway (Kaplan and Shapley, 1986).

Several studies suggest that this early processing may be dysfunctional in schizophrenia: for instance, patients have poorer performance in contrast sensitivity tasks at different spatial frequency ranges, compared to healthy controls (Slaghuis, 1998, 2004; Kéri et al., 2002; Butler et al., 2005, 2007; Revheim et al., 2006; Kéri and Benedek, 2007; Martínez et al., 2008; Kantrowitz et al., 2009). Other studies have found a specific deficit of the processing of very low spatial frequencies (LSF), below 1.5 cycle/degree (Butler and Javitt, 2005; Revheim et al., 2006; Butler et al., 2007), which has been linked to subcortical magnocellular dysfunction (Butler and Javitt, 2005), although this interpretation remains controversial (Skottun and Skoyles, 2007).

Beside their implication in early visual processing, spatial frequencies also play a role in the formation of coherent visual representations. In realistic conditions of viewing, our visual system is exposed to a very large range of spatial frequencies and preferentially selects some spatial scales over others depending on temporal processing (e.g., exposure duration, Schyns and Oliva, 1994; Oliva and Schyns, 1997; Peyrin et al., 2006), task requirements (Schyns and Oliva, 1999; Peyrin et al., 2005), or distance of viewing (Brady and Oliva, 2012). In a previous study, our team has explored how different spatial frequencies are integrated to form a coherent representation of a face in patients with schizophrenia (Laprévote et al., 2010). We measured performance in a facial expression recognition task with hybrid faces. Hybrid faces are images made of two superimposed faces, one face filtered to only keep the LSF and the other filtered to only keep HSF. The results showed that patients suffering from schizophrenia more often used the LSF to recognize facial expression at a glance. A control experiment with single faces shown either in LSF or in HSF demonstrated that patients were able to process both spatial scale components when shown separately, meaning they could see very well the details of the HSF percept, when not in competition with a LSF percept. Therefore, the bias for LSF in hybrid images was interpreted as a deficit of the mechanism of spatial frequency integration in the brain.

One caveat of our previous study however, is that we used a facial expression recognition task, which might have introduced additional biases leading to the appearance of a HSF impairment. Faces are considered as special visual objects which activate specific brain networks (Kanwisher and Yovel, 2006; Atkinson and Adolphs, 2011). This specificity implies a differential processing of spatial frequencies for face stimuli: the optimal spatial band for face recognition is comprised between 8 and 16 cycles/face, so on the lower end of the spatial spectrum (Costen et al., 1996), and LSF may play a more significant role than HSF in rapid face processing because LSF support configural processing (Goffaux et al., 2005) and precede the integration of HSF (Awasthi et al., 2011). Moreover, it is known that people with schizophrenia have an emotion recognition deficit (Morris et al., 2009), and that the emotional content of a face modulates spatial frequency perception (Smith and Schyns, 2009; Kumar and Srinivasan, 2011). In a study using bubbles technique during a facial emotion discrimination task, Lee et al. (2011) have shown that patients with schizophrenia had an atypical strategy of using visual information to recognize different emotions: patients relied less frequently less frequently on high frequency information contained in the eyes of the fearful faces, whereas they used higher spatial frequencies to recognize happiness.

To further investigate the processing of rapid spatial frequency integration in patients with schizophrenia,we ran two psychophysical experiments with objects instead of faces. Is the bias found for low spatial frequency in hybrid stimuli (Laprévote et al., 2010) face specific or a generic mechanism of object recognition? In the first experiment we showed hybrid stimuli (**Figure 1A**) for a short glance, each made of a picture of an animal and a picture of a fruit, one in LSF and the other in HSF. With this simple design, the image category reported by the participant directly tells the spatial scale preferentially perceived. Accuracy results showed that patients with schizophrenia reported significantly more frequently the LSF image than the control group of participants. A control experiment tested the origin of the LSF bias in Experiment 1: all

participants performed the same object categorization task but were shown filtered single pictures of animals or fruits, showing only the HSF or the LSF components of each image (**Figure 1B**). This allowed us to determine whether the LSF bias observed in the first experiment was due to a specific deficit at processing HSF or a deficit in concurrently processing HSF and LSF in images containing a broad spectrum of spatial frequencies (as in normal perception).

# **MAIN EXPERIMENT**

# **METHOD**

# **Participants**

Fifteen adult individuals suffering from schizophrenia were recruited from the Public Mental Health Institute of Lille Metropole (Armentières, France), the Public Mental Health Institute of Flandres (Bailleul, France), and the Department of General Psychiatry in Lille University Hospital (Lille, France). The inclusion criteria required an age of 18–55 years and a diagnosis of schizophrenia based on standard DSM-IV criteria (American Psychiatric Association, 2000). All participants' visual acuity was measured by the Snellen chart. Only patients and controls with a normal or corrected-to-normal visual acuity (10/10 on Snellen chart) were included. The exclusion criteria were history of neurological illness, trauma occurring in the past 6 months, ophthalmic illness, and alcohol or drug abuse. All patients received antipsychotic medication and were clinically stable at testing time. Symptoms of schizophrenia were assessed with the Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987). Twelve age and gendermatched healthy controls were initially recruited. They were free from a DSM-IV axis-I diagnosis and reported taking no medication. The study was approved by the Ethics Committee of Lille University Hospital. A written consent was obtained from all participants. No participant was paid for taking part in the study.

# **Stimuli**

Hybrid stimuli were made of an animal and a fruit. The original picture set was taken from the Hemera Photo Object database. In that set, 35 animals and 35 fruits were grouped by pairs and were aligned so that inner and outer object characteristics overlapped (see **Figure 1**). The pictures grayscale images, centered in a matrix of 256 × 256 pixels. We created a low-pass version (below 8 cycles/image) and a high-pass version (above 24 cycles/image) of each object (see examples in **Figure 1**), for a total of 70 HSF-only objects and 70 LSF-only objects. Then,we created hybrid stimuli by overlapping a low-pass filtered object of one of the categories (animal/fruit) with the corresponding high-pass filtered object of the other category. Each pair of objects provided two hybrids images (LSF fruit/HSF animal and HSF fruit/LSF animal). As each hybrid was composed of two different objects, an animal and a fruit, the participant's response on a given hybrid image indicated which spatial scale was preferentially reported.

# **Procedure**

Participants were seated in a darkened room with their head stabilized by a chin-rest at a viewing distance of 140 cm. Stimuli subtended 2.5˚ of visual angle. A central fixation cross was shown for 1 s, followed by a hybrid stimulus displayed for 100 ms. This

**FIGURE 1 | Example of a hybrid stimulus (A)**. The hybrid combines the high spatial frequency information from the baby dog in image **(B)** with the low spatial frequency information from the litchi in image **(C)**. The high spatial frequency component of the hybrid can be seen more easily if you hold the

image close to your eyes, and the low spatial frequency information can be better seen if you step away from the image or slightly blur your gaze. Image **(B,C)** are examples of filtered images that were used in the control experiment.

presentation time was chosen to allow only one fixation on the stimulus, as the average human gaze fixation is around 300 ms (Harris et al., 1988). The stimuli were presented in a randomized order. Participants were asked to decide verbally whether the object was an animal or a fruit. The answer was coded by the experimenter on the keyboard of the computer. Before this experiment, participant performed a practice trial with five hybrids stimuli, in order to familiarize with the task. This hybrid set was different from the experimental set. On the basis of a pilot study on healthy participants, eight hybrid images were excluded in the final analysis because one of the two images proved to be more salient in terms of luminance.

# **Statistical analysis**

Statistical analyses were conducted with STATISTICA 6.1 software (StatSoft Inc.).

The main dependent variable was the percentage of object categorization responses to images for each spatial frequency component. Upon examination of the data distribution, we found that one healthy control's data were two standard deviations above the mean, so we did not include that data in the statistical analysis. The percentage of responses in each spatial frequency was compared between patients and healthy controls with a bilateral *t*-test.

Two-tailed Pearson correlations were used to check any relationship between the percentage of responses in each spatial frequency and antipsychotic daily dose, benzodiazepine dose, age, or PANSS dimensions.

#### **RESULTS**

The characteristics of the population are summarized in **Table 1**. Ages of patients and healthy controls were respectively 36.8 (SEM = 2.8) and 35.3 (SEM = 2.2) years.

Our main measure was the percentage of responses in each spatial frequency.

For hybrid stimuli, the percentage of responses based on the LSF object was respectively 50.67% (standard error of the mean, SEM,was 5.89%)for patients and 32.87% (SEM 5.55%)for healthy controls. This difference was significant [*t*(24) = 2.12, *p* < 0.05]. The complementary percentage of responses based on HSF was 49.32% (SEM 5.89%) for patients and 67.12% (SEM 5.55%) for healthy controls. As there was a small number of participants, a *post hoc* power analysis was conducted with the software G\_Power3 (Faul et al., 2007). It showed that the power of our design was 0.68 (α = 05).

We did not find any significant correlation between the percentage of responses, in each spatial frequency and antipsychotic daily dose, benzodiazepine dose, age, or any PANSS dimension.

# **CONTROL EXPERIMENT: HIGH AND LOW SPATIAL FREQUENCY FILTERED OBJECTS**

# **AIM**

This control experiment verifies if participants could perform the main task correctly and assesses if the bias found with hybrid images in main experiment resulted from an inability to process HSF components. To do so, filtered images containing only LSF or only HSF were used.

# **METHOD**

#### **Participants**

The same two populations performed the control experiment. This experiment was conducted after Experiment 1, after a 10 min pause, in the same conditions.

#### **Stimuli**

We used the same original animals and fruits sets as in those used with hybrid stimuli. We created a low-pass version (below 8 cycles/image) and a high-pass version (above 24 cycles/image) of each object (see examples in **Figure 1**),for a total of 70 HSF-only object images (35 animals and 35 fruits) and 70 LSF-only object images (35 animals and 35 fruits).

#### **Procedure**

This set of 140 images was presented to participants in the same conditions as in experiment 1 (distance was 140 cm, size of 2.5˚ of visual angle, and stimulus presentation time was 100 ms). Participants responded orally if the picture was an animal or a fruit, and the answer was coded by the experimenter on the keyboard of the computer. The 140 stimuli were presented in a randomized order in two blocks of 70 trials, separated by a short pause.

#### **Statistical analysis**

We collected the percentage of correct responses in each spatial frequency. This variable was analyzed with a repeated measures


ANOVA with "group" as between factor and "spatial frequency" and "object category" as within factors.

#### **RESULTS**

Healthy controls correctly reported the category in 96.6% (SEM 0.6%) for HSF-only images and 94.7% (SEM 0.8%) for LSFonly images. Patients' responses were 91.9% (SEM 1.9%) for HSF-only images and 91.0% (SEM 1.5%) for LSF-only images. The ANOVA showed a main effect of group [*F*(1, 24) = 4.65, *p* < 0.05), as on average healthy controls were more accurate than patients. The ANOVA revealed no effect of spatial frequency or stimulus category on correct answers. There was no spatial frequency × group interaction. Spatial frequency interacted with category [*F*(1,24) = 7.26,*p* = 0.01]. This interaction resultedfrom fruits being better recognized in high spatial frequency than in low spatial frequency [*F*(1, 24) = 5.86, *p* < 0.05] whereas there was no significant difference for animals. There was no spatial frequency × category × group interaction.

# **DISCUSSION**

The goal of this study was to verify if the low spatial frequency bias previously observed in schizophrenia (Laprévote et al.,2010) isface specific or generalizes to other object categories. Patients suffering from schizophrenia and healthy controls performed an object categorization task on hybrid stimuli, which combined an animal and a fruit, shown each at a different spatial frequency range. In the hybrids, patients categorized more frequently the images in low spatial frequency than did healthy controls. We verified that the LSF bias was not due to a task misunderstanding or to a deficit in the processing of HSF components as both patients and controls categorized objects with high accuracy when HSF-only and LSF-only stimuli were used.

This result replicates previous findings with hybrid faces, demonstrating a strong bias toward low spatial frequency in schizophrenia when categorizing images at a glance (Laprévote et al., 2010). This LSF bias contrasts with the predictions made on the basis of the subcortical magnocellular deficit hypothesis in schizophrenia (Butler and Javitt, 2005). Because LSF are preferentially conveyed by the magnocellular pathway, such a deficit would imply

a deficit of the perception of LSF. However, our results challenge this view, and suggest a framework where spatial frequencies are integrated in a dynamic fashion to form an object percept. Associating high and LSF is crucial for the recognition of complex visual stimuli: LSF are processed early on by the visual system and they precede the slower integration of HSF to form a full spatial scale percept (Schyns and Oliva, 1994; Bar, 2004). In our two studies, patients with schizophrenia based their decision on default LSF information, neglecting the potentially slower processing of HSF. Indeed, Bar et al. (2006) (Kveraga et al., 2007) have proposed that coarse visual information contained in LSF are quickly carried over by the dorsal cortical pathway and can rapidly provide a skeleton layout of visual information to orbito-frontal cortex, which influences by feedback connections the slower processing of details conveyed by HSF preferentially processed in the ventral cortical pathway. Our current results fit well with a dysfunction of spatial scale integration, in line with other studies suggesting a dysfunction of the interaction between dorsal and ventral pathways in patients with schizophrenia (Doniger et al., 2002; Schechter et al., 2003; Foxe et al., 2005; Ducato et al., 2008; Plomp et al., 2012). In addition, recent findings have also suggested that a dysfunction of early processing by the dorsal stream may impair high-level visual functions. For instance, Sehatpour et al. (2010) examined cortical activations with fMRI during the formation of a coherent object representation via a perceptual disclosure paradigm in patients with schizophrenia. Patients had less activity overall of the dorsal visual network which contributed to subsequent impaired activity of the ventral visual stream. Our results are also aligned with this proposal: coarse information contained in LSF may be processed at a minimum, but the signal strength may not be sufficient to allow amplification of subsequent detailed processing, implying a preference for LSF visual information.

A possible limitation to our study is that patients were under medication at the time of testing. Benzodiazepines have been shown to impair contrast sensitivity, mainly for LSF (Haris and

## **REFERENCES**


*Proc. Natl. Acad. Sci. U.S.A.* 103, 449–454.


Phillipson, 1995). Also, antipsychotics have been shown to impair contrast perception for HSF and to increase contrast perception for LSF (Harris et al., 1990). While our analysis failed to find any significant correlation between spatial frequency preference and benzodiazepine or antipsychotics daily dose, further studies will be necessary to measure the impact of those treatments on spatial frequency preference. The absence of any experiment testing directly performances of participant with hybrid faces may also constitute a bias. This methodological choice has been made in order to simplify experimental design and preserve attentional capacities of participants. However,we conducted the experiments of this paper in the same experimental condition as our previous work about hybrid faces (Laprévote et al., 2010). At last the small sample sizes may also constitute a limitation of this study.

To summarize, here we confirm that patients suffering from schizophrenia have a preference for coarse LSF information in fast visual categorization tasks, and show that this preference is not specific to faces but generalizes to other categories of objects. As patients, like controls, performed well with separated LSF and HSF stimuli, the LSF bias when processing full range hybrid images, as in normal conditions of viewing, may be interpreted as a dysfunction of the integration of spatial scales, a fundamental mechanism to form a rich and coherent representation of visual objects. Schizophrenia patients may preferentially rely on default LSF information to appreciate their visual environment.

#### **ACKNOWLEDGMENTS**

The authors thank Wilma Bainbridge for comments on the paper. We are grateful to the patients and the healthy controls who participated to this study. Many thanks to Public Mental Health Institute of Lille Metropole (Armentières – France) and Public Mental Health Institute of Flandres (Bailleul – France), which have contributed to patient's recruitment. **Funding Source:** Funding for this study was provided by the French National Research Agency (Grant ANR-12-SAMA-0016-01) and the CNRS.


as the trigger of top-down facilitation in recognition. *J. Neurosci.* 27, 13232–13240.


Brand, A., et al. (2012). Electrophysiological evidence for ventral stream deficits in schizophrenia patients*. Schizophr. Bull*. Available at: http://schizophreniabulletin.oxford journals.org/content/early/2012/01/ 18/schbul.sbr175 (Accessed March 29, 2013).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of Interest.

*Received: 31 January 2013; accepted: 15 April 2013; published online: 06 May 2013.*

*Citation: Laprevote V, Oliva A, Ternois A-S, Schwan R, Thomas P and Boucart M (2013) Low spatial frequency bias in schizophrenia is not face specific: when the integration of coarse and fine information fails. Front. Psychol. 4:248. doi: 10.3389/fpsyg.2013.00248*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Laprevote, Oliva, Ternois, Schwan, Thomas and Boucart. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Neuropsychological functions and visual contrast sensitivity in schizophrenia: the potential impact of comorbid posttraumatic stress disorder (PTSD)

# *Ibolya Halász 1, Einat Levy-Gigi 2, Oguz Kelemen3, György Benedek4 and Szabolcs Kéri 1,4\**

*<sup>1</sup> National Psychiatry Center, Budapest, Hungary*

*<sup>2</sup> Institute for the Study of Affective Neuroscience, University of Haifa, Haifa, Israel*

*<sup>3</sup> Psychiatry Center, Bács-Kiskun County Hospital, Kecskemét, Hungary*

*<sup>4</sup> Department of Physiology, Faculty of Medicine, University of Szeged, Szeged, Hungary*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Martin Brüne, Ruhr University Bochum, Germany Pamela Butler, Nathan Kline Institute, USA Gregory P. Strauss, University of Maryland School of Medicine, USA*

#### *\*Correspondence:*

*Szabolcs Kéri, Department of Physiology, University of Szeged, Dóm sq. 10, H6720 Szeged, Hungary. e-mail: keri.szabolcs.gyula@ med.u-szeged.hu; szkeri2000@ yahoo.com*

Previous studies have revealed a high prevalence of posttraumatic stress disorder (PTSD) in patients with other severe mental disorders, including schizophrenia. However, the neuropsychological and psychophysical correlates of comorbid PTSD are less exactly defined. The purpose of the present study was to assess immediate and delayed memory, attention, visuospatial skills, language, and basic visual information processing in patients with schizophrenia with or without PTSD. We recruited 125 patients with schizophrenia and 70 healthy controls matched for visual acuity, age, gender, education, and socioeconomic status. Twenty-one of patients with schizophrenia exhibited comorbid PTSD. We administered the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) and visual contrast sensitivity tasks for low spatial/high temporal frequency (0.3 cycle/degree and 18 Hz) and high spatial/low temporal frequency (10 cycles/degree and 1Hz) sinusoidal gratings. All patients were clinically stable and received antipsychotic medications. Results revealed that relative to healthy controls, patients with schizophrenia exhibited significant and generalized neuropsychological dysfunctions and reduced visual contrast sensitivity, which was more pronounced at low spatial/high temporal frequency. When we compared schizophrenia patients with and without PTSD, we found that patients with comorbid PTSD displayed lower scores for RBANS attention, immediate and delayed memory, and visuospatial scores. Schizophrenia patients with or without PTSD displayed similar visual contrast sensitivity. In conclusion, comorbid PTSD in schizophrenia may be associated with worse neuropsychological functions, whereas it does not affect basic visual information processing.

**Keywords: schizophrenia, posttraumatic stress disorder (PTSD), neuropsychology neuroimaging, vision tests, psychopathological features**

# **INTRODUCTION**

The presence, origin, and clinical characteristics of posttraumatic stress disorder (PTSD) in severe mental disorders, such as schizophrenia, are one of the most controversial issues in the literature. According to the meta-analysis of Achim et al. (2011), the prevalence of PTSD in schizophrenia and related psychotic disorders is 12.4%, with a wide confidence interval of 4–20.8%. Comorbid PTSD, which is characterized by upsetting re-experiencing the traumatic event (e.g., flashbacks and nightmares), avoidance of thoughts and cues related to the trauma, emotional numbing, and increased arousal, may be associated with adverse functioning, lower quality of life, and worse outcome (Grubaugh et al., 2011). Early trauma may be a pathogenic factor in the development of both schizophrenia and PTSD, but later traumatic events after the emergence of schizophrenia may also lead to PTSD (Schäfer and Fisher, 2011; Matheson et al., 2013).

In this study, we aimed to clarify two aspects of schizophrenia in relation to comorbid PTSD: early stage visual information processing and neuropsychological functions. There is a great deal of evidence that basic low-level vision, such as the detection of simple luminance-contrast gratings and elementary perceptual organization, is impaired in schizophrenia (reviewed by Javitt, 2009; Silverstein and Keane, 2011; Butler et al., 2012). Neuropsychological and social cognitive deficits are extensively documented in schizophrenia, exhibiting a definitive relationship with psychosocial functioning (Green, 2007; Fett et al., 2011; Keefe and Harvey, 2012). To our knowledge, there have been no studies investigating visual information processing in schizophrenia with comorbid PTSD, and reports on neuropsychological functions provided inconsistent results because of differences in methods and patient populations (Goodman et al., 2007; Fan et al., 2008; Duke et al., 2010; Peleikis et al., 2012). Therefore, we recruited a large sample of patients with schizophrenia and screened them for PTSD. We measured visual contrast sensitivity to characterize low-level perceptual functions and administered a standard battery of neuropsychological tests assessing immediate and delayed memory, attention, language, and visuospatial functions. We hypothesized that patients with schizophrenia with comorbid PTSD (SCZ + PTSD) would display worse visual and neuropsychological functions compared with schizophrenia patients without PTSD (SCZ). The hypothesis of early stage visual dysfunction was based on meta-analytic evidence revealing hypoactivation in the occipital cortex in PTSD, but not in other anxiety disorders (Etkin and Wager, 2007).

# **MATERIALS AND METHODS**

#### **PARTICIPANTS**

We enrolled 125 patients with schizophrenia and 70 healthy controls with a negative family history for mental disorders at the National Psychiatry Center, Budapest, Hungary. The diagnosis was based on structured clinical interviews (First et al., 1996) and medical records. The patients with schizophrenia received antipsychotic medications at the time of testing (risperidone, olanzapine, quetipaine, haloperidol, aripriprazole, zuclopenthixol, amisulpride), which were converted to chlorpromazine-equivalent doses (Woods, 2003). The severity of the symptoms was characterized by the Brief Psychiatric Rating Scale (Overall and Gorham, 1962), and the Clinician-Administered PTSD Scale (CAPS) (Blake et al., 1990). We used the Four Factor Index of Social Status (Hollingshead, 1975). All participants had normal or corrected-to-normal visual acuity. From the 125 patients with schizophrenia, 21 patients (17%) exhibited comorbid PTSD (physical assault: *n* = 11; sexual assault: *n* = 7; sudden violent death of a spouse or friend: *n* = 1; traffic accident: *n* = 1; natural disaster: *n* = 1). In each patient PTSD was a current, and not a lifetime, comorbid condition. From the 21 SCZ + PTSD individuals, 7 patients (33%) also had major depressive disorder (MDD), and 4 patients (19%) had current substance misuse. From the 104 SCZ patients, 15 individuals (14%) had MDD, and 16 individuals (15%) had current substance misuse. The clinical and demographic data are summarized in **Table 1**. The study was done in accordance with the Declaration of Helsinki, and each participant gave written informed consent. The study was approved by the institutional ethics board.

# **VISUAL CONTRAST SENSITIVITY**

The procedure, which is suitable for the reliable measurement of visual contrast sensitivity in individuals with less efficient general cognitive functions, has been described elsewhere (Kogan et al., 2004; Kéri and Benedek, 2009). We presented vertical sinusoidal luminance-contrast gratings on a gamma-corrected ViewSonic PF815 monitor. During visual contrast sensitivity measurements, the minimal luminance-contrast is measured, which is indispensable for the detection of the gratings (**Figure 1**). The first type of stimuli had low spatial frequency (SF) and high temporal frequency (TF) (0.3 cycle/degree and 18 Hz, respectively), and the second type of stimuli had high SF and low TF (10 cycles/degree and 1 Hz, respectively). The stimulus area was a circular window (mean luminance: 31 cd/m2, size: 8◦). The initial Michelsoncontrast was 12%. We used a Yes/No one-up/two-down staircase procedure (step size: 0.1 log contrast unit). The participants gave oral responses (grating is seen or not), and the experimenter entered all responses on the computer keyboard. We used catch trials to control for spurious responding. The staircase was finished when the slope and SD of the last 12 trials was less than the step size. The detection threshold was the mean of the last 12 reversals (for methodological details, see Kogan et al., 2004; Kéri and Benedek, 2009).

# **NEUROPSYCHOLOGICAL ASSESSMENT**

We used the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) battery, which is suitable

#### **Table 1 | Clinical and demographic characteristics.**


*Data are mean (standard deviation) with the exception of gender ratio. CONT, controls; SCZ, schizophrenia; PTSD, posttraumatic stress disorder; BPRS, brief psychiatric rating scale; CAPS, clinician-administered PTSD scale. The three groups did not differ in age, education, and socioeconomic status (p > 0.1, t-test). In the SCZ* + *PTSD group, females were overrepresented, although it was not significant compared with the other two groups (chi-square test, p > 0.05). SCZ and SCZ* + *PTSD patients did not differ in total BPRS scores (p > 0.1).*

Halász et al. Neuropsychology and vision in schizophrenia with PTSD

for rapid and comprehensive evaluation of neurocognitive functions (Randolph, 1998; Gold et al., 1999; Juhász et al., 2003). The RBANS battery consists of 12 tests, combined to five index scores. Each index score is standardized (normal mean: 100, *SD* = 15 based on a normative study group of 200 healthy Hungarian volunteers, 20–80 years of age). The index domains are as follows: (1) immediate memory [word list learning (10 words repeated in four trials), story recall in two trials]; (2) language (confrontation naming of 10 pictures, category fluency); (3) visuospatial functions (figure copy, line orientation); (4) attention (digit span, digit-symbol coding); (5) delayed memory (delayed recall of the story, complex figure, and word list, recognition of the word list). The RBANS battery has two psychometrically matched forms; here we used version "A."

#### **DATA ANALYSIS**

We used STATISTICA 11 (StatSoft, Inc., Tulsa) for data analysis. Data quality assumptions were evaluated with Kolmogorov– Smirnov test (normality of distribution) and Levene's tests (homogeneity of variance). Analyses of variance (ANOVAs) were used in the case of contrast sensitivity and RBANS data, followed by Tukey Honestly Significant Difference (HSD) tests, corrected for unequal sample sizes. We used two-tailed *t*-tests and chi-square tests for the comparison of demographic data. The relationship among the relevant variables was investigated using Pearson's product moment correlation coefficients (*r*). Where appropriate and informative, we also calculated Cohen's effect size values (*d*). The level of statistical significance was set at α *<* 0*.*05.

#### **RESULTS**

#### **VISUAL CONTRAST SENSITIVITY**

The ANOVA conducted on the contrast sensitivity data indicated significant main effects of group [*F(*2*,* <sup>192</sup>*)* = 95*.*47, *p <* 0*.*0001], stimulus type [*F(*1*,* <sup>192</sup>*)* = 208*.*83, *p <* 0*.*0001], and a Two-Way interaction between group and stimulus type [*F(*2*,* <sup>192</sup>*)* = 10*.*83, *p <* 0*.*001]. The *post-hoc* comparisons revealed that both SCZ and SCZ + PTSD groups exhibited reduced contrast sensitivity at low SF/high TF relative to healthy controls (*p <* 0*.*001), but there was no significant difference between the SCZ and SCZ + PTSD groups (*p* = 0*.*74). At high SF/low TF, the results were similar (CONT *>* SCZ = SCZ + PTSD; significant difference between CONT and SCZ groups: *p <* 0*.*001) (**Figure 2**). The Two-Way interaction between group and stimulus type appeared because the difference between patients (SCZ plus SCZ + PTSD) and controls was larger on the low SF/high TF condition (*d* = 1*.*47) than on the high SF/low TF condition (*d* = 0*.*83).

#### **NEUROPSYCHOLOGICAL PERFORMANCE**

**Figure 3** demonstrates the RBANS performance. The statistical analyses (One-Way ANOVAs followed by Tukey HSD for unequal Ns and Cohen's effect size values) are summarized in **Table 2**. Overall, SCZ patients exhibited significant deficits in attention, memory, visuospatial functions, and language. Patients with SCZ + PTSD showed more severe dysfunctions in attention, memory, and visuospatial functions relative to the SCZ group.

**FIGURE 2 | Visual contrast sensitivity.** CONT, control; SCZ, schizophrenia; PTSD, posttraumatic stress disorder; SF, spatial frequency; TF, temporal frequency. Error bars indicate 95% confidence intervals. SCZ and SCZ + PTSD patients showed similar reductions in sensitivity relative to controls (*p <* 0*.*001, Tukey HSD for unequal Ns).

schizophrenia; PTSD, posttraumatic stress disorder; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; Att, attention; MemI, immediate memory; MemD, delayed memory; VisSpat, visuospatial; Lang, language. Error bars indicate 95% confidence intervals. For statistical details, see **Table 2**.

#### **CORRELATIONS AND GENDER DIFFERENCES**

Correlations among RBANS, contrast sensitivity, symptoms (BPRS and CAPS scores), and chlorpromazine-equivalent antipsychotic dose did not reach the level of statistical significance when SCZ and SCZ + PTSD patients were analyzed separately or together (−0*.*1 *< r <* 0*.*1, *p >* 0*.*1). There were no significant differences in any visual perceptual and neuropsychological measures between male and female patients (*p >* 0*.*1, *t*-tests).

#### **EFFECT OF COMORBID MDD AND SUBSTANCE MISUSE**

We compared SCZ and SCZ + PTSD patients with and without MDD/substance misuse on visual contrast sensitivity and RBANS



*The table present results from One-Way ANOVAs. RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; CONT, healthy controls; SCZ, schizophrenia; PTSD, posttraumatic stress disorder. See also Figure 2.*

performance. We did not find significant differences between patients with and without MDD/substance misuse in the SCZ and SCZ + PTSD groups (ANOVAs, *p >* 0*.*4).

# **DISCUSSION**

The results of the present study indicate that SCZ + PTSD is associated with more severely disrupted memory, attention, and visuospatial functions compared with SCZ. In contrast, low-level visual processing is not affected by the presence of comorbid PTSD, which suggests that occipital hypoactivation observed in PTSD may not affect basic contrast processing (Etkin and Wager, 2007). In order to examine visual processing in more detail, patients with PTSD should be tested with the parallel application of psychophysical and neuroimaging techniques.

By the assessment of elderly individuals, Goodman et al. (2007) demonstrated that Holocaust survivor SCZ + PTSD patients scored lower on tests of speed of information processing, recognition memory, and general mental status than elderly SCZ patients without trauma exposure. Lysaker et al. (2001) found that individuals with schizophrenia-spectrum disorders with selfreported childhood sexual abuse performed worse in the case of working memory and executive functions than did individuals without self-reported abuse. Fan et al. (2008) also found a more severe generalized cognitive dysfunction in SCZ + PTSD than in SCZ, with a special reference to attention, working memory, and executive functions. However, two studies failed to find any evidence that SCZ + PTSD is associated with substantially increased neuropsychological impairment (Duke et al., 2010; Peleikis et al., 2012), although in the Duke et al. (2010) study SCZ + PTSD patients showed a qualitatively different neuropsychological profile compared with SCZ patients, which was influenced by differences in age, education, and symptoms. In the present study, we identified a relatively high number of SCZ + PTSD patients, which is consistent with the literature (Achim et al., 2011). SCZ + PTSD and SCZ patients displayed similar clinical and demographic characteristics, and, therefore, differences in neuropsychological performance cannot be attributed to these potential confounding variables. Comorbid MDD and substance misuse did not affect neuropsychological and visual performance. Correlation analysis also revealed negative results as regards symptoms and antipsychotic medications. A categorical comparison of symptom severity (mild, moderate, severe) is not possible because studies rarely use CAPS to characterize PTSD severity in schizophrenia.

Contrary to the differences in neuropsychological functions, SCZ + PTSD and SCZ patients demonstrated statistically indistinguishable visual contrast sensitivity deficits; in both groups, reduction in sensitivity was similar relative to controls, which was more pronounced at low SF/high TF. Several previous studies demonstrated altered visual contrast sensitivity in SCZ, but the data varied regarding the specific values measured at different SFs and TFs (e.g., Slaghuis, 1998; Kéri et al., 2002; Chen et al., 2003; Butler et al., 2009; Kiss et al., 2010; for a critical review, see Skottun and Skoyles, 2007, 2011). A more pronounced reduction in contrast sensitivity at low SF/high TF can be explained by the differential impairment of the magnocellular visual pathway, which project from the retina to the primary visual cortex via the lateral geniculate nucleus; this pathway participates in the processing of stimuli with low SF/high TF (Merigan and Maunsell, 1990). However, the magnocellular origin of contrast sensitivity alteration in SCZ has been debated (Skottun and Skoyles, 2007, 2011) because the magnocellular system shows a considerable overlap with other cells in the visual system regarding contrast and SF response. Nevertheless, evidence suggests that the cortical processing of low SFs is compromised in SCZ (Martínez et al., 2008; Kéri et al., 2012; Calderone et al., 2013), and that low SF achromatic gratings activate the putative magnocellular layers of the human lateral geniculate nucleus (Denison et al., 2012). The rapid development of high-resolution imaging techniques capable of identifying and visualizing the magnocellular layers may provide answer to this debate question.

The most important limitation of the present study is that no PTSD patients without SCZ were included. Therefore, it is not clear how PTSD itself affects neuropsychological and particularly visual functions in our study context. By conducting a direct comparison between SCZ + PTSD and PTSD alone, Duke et al. (2010) demonstrated much milder neuropsychological deficits in PTSD relative to PTSD + SCZ. Nevertheless, several studies have shown that PTSD itself is associated with compromised attention, memory, and executive functions (Golier and Yehuda, 2002), although the causal link between trauma and impaired cognition is controversial (Danckwerts and Leathem, 2003). Regarding low-level vision, even less information is available in PTSD. Hendler et al. (2003) showed that the visual cortex displayed more activation in PTSD when trauma-related images were exposed at below recognition threshold, which suggests that the earliest stage of visual information processing is corrupted in PTSD. However, it remains to be explored how PTSD patients react to emotionally neutral basic and more complex visual stimuli (e.g., Levy-Gigi and Kéri, 2012).

The second limitation was that we had to restrict the assessment of the symptoms, and we did not depict positive, negative,

# **REFERENCES**


in schizophrenia with comorbid posttraumatic stress disorder. *J. Clin. Exp. Neuropsychol.* 32, 737–751.


disorganized, and affective symptom separately. In general, however, BPRS and CAPS scores did not correlate with neuropsychological and visual contrast sensitivity performance. Finally, the sample size was still small in the SCZ + PTSD group, which may limit the generalizability of the findings, given that the symptoms, clinical course, and outcome of the illness show a high degree of individual variability and heterogeneity. Therefore, the results must be replicated and extended in a larger sample.


atypical antipsychotics. *J. Clin. Psychiatry* 64, 663–667.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 14 January 2013; accepted: 04 March 2013; published online: 20 March 2013.*

*Citation: Halász I, Levy-Gigi E, Kelemen O, Benedek G and Kéri S (2013) Neuropsychological functions and visual contrast sensitivity in schizophrenia: the* *potential impact of comorbid posttraumatic stress disorder (PTSD). Front. Psychol. 4:136. doi: 10.3389/fpsyg. 2013.00136*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Halász, Levy-Gigi, Kelemen, Benedek and Kéri. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Visual surround suppression in schizophrenia

#### **Marc S. Tibber 1,2\*, Elaine J. Anderson1,2,3,Tracy Bobin<sup>4</sup> , Elena Antonova<sup>4</sup> , Alice Seabright <sup>5</sup> , BerniceWright <sup>5</sup> , Patricia Carlin<sup>4</sup> , Sukhwinder S. Shergill <sup>4</sup> and Steven C. Dakin1,2**

1 Institute of Ophthalmology, University College London, London, UK

<sup>2</sup> NIHR Biomedical Research Centre at Moorfields Eye Hospital, London, UK

3 Institute of Cognitive Neuroscience, University College London, London, UK

4 Institute of Psychiatry, King's College London, London, UK

<sup>5</sup> Department of Cognitive, Perceptual and Brain Sciences, University College London, London, UK

#### **Edited by:**

Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA

#### **Reviewed by:**

Brian P. Keane, UMDNJ – Robert Wood Johnson Medical School; Rutgers University Center for Cognitive Science, USA Ariel Rokem, Stanford University, USA Jong Yoon, University of California Davis, USA

#### **\*Correspondence:**

Marc S. Tibber, Institute of Ophthalmology, University College London, Bath Street, London EC1 9EL, UK. e-mail: mtibber@yahoo.com

Compared to unaffected observers patients with schizophrenia (SZ) show characteristic differences in visual perception, including a reduced susceptibility to the influence of context on judgments of contrast – a manifestation of weaker surround suppression (SS). To examine the generality of this phenomenon we measured the ability of 24 individuals with SZ to judge the luminance, contrast, orientation, and size of targets embedded in contextual surrounds that would typically influence the target's appearance. Individuals with SZ demonstrated weaker SS compared to matched controls for stimuli defined by contrast or size, but not for those defined by luminance or orientation. As perceived luminance is thought to be regulated at the earliest stages of visual processing our findings are consistent with a suppression deficit that is predominantly cortical in origin. In addition, we propose that preserved orientation SS in SZ may reflect the sparing of broadly tuned mechanisms of suppression. We attempt to reconcile these data with findings from previous studies.

**Keywords: schizophrenia, surround suppression, cortex, contrast, luminance, size, orientation, perception**

# **INTRODUCTION**

Schizophrenia (SZ) is a mental disorder characterized by a range of cognitive, affective, and perceptual symptoms, the diversity of which presents a considerable challenge to any single pathophysiological model of the condition (Cohen and Servan-Schreiber, 1992; Barch and Ceaser, 2012). However, recent studies indicate that visual deficits in SZ may result from abnormalities in *gain control* (Butler et al., 2008), the inhibitory processes by which neurons regulate their overall levels of activity to optimize information transmission (Heeger, 1992). That gain control is an example of a *canonical* neural computation (Carandini and Heeger, 2012), i.e. one that is likely to be repeated across different brain regions and modalities, makes it a potential candidate for involvement in the wide range of symptoms that characterize SZ.

In terms of visual processing, gain control is thought to play a critical role in *contextual effects,* whereby the presence of a surround influences or biases the perception of a target (Albright and Stoner, 2002). There is evidence that such phenomena are reduced or absent in people with SZ (Silverstein et al., 2000). For example, the perceived contrast of a target is normally reduced when embedded in a high contrast surround (**Figure 1B**) – an instance of a more general phenomenon known as *surround suppression* (SS; Chubb et al., 1989). However, patients with SZ are much less susceptible to this effect. As a result, under conditions that would normally lead to SS, patients select a perceptual match that is closer to veridical than do controls (Dakin et al., 2005; Yoon et al., 2009, 2010; Barch et al., 2012). (See **Figure 1** legend).

Converging evidence from psychophysical (Solomon et al., 1993), electrophysiological (Ohtani et al., 2002; Haynes et al., 2003) and functional imaging (Williams et al., 2003; Zenger-Landolt and Heeger, 2003) studies suggest that SS is mediated by the inhibition of a neuron's response to a stimulus by the pooled activity of cells in surrounding cortex (Heeger, 1992; Solomon et al., 1993; Xing and Heeger, 2001). Abnormal SS in SZ is therefore consistent with reduced levels of cortical inhibition (Butler et al., 2008). This hypothesis is supported by the finding that impoverished contrast SS in SZ correlates with a visuo-cortical deficit of gamma-aminobutyric acid (GABA; Yoon et al., 2010), the brain's primary inhibitory neurotransmitter (see Wassef et al., 2003 for a review of GABAergic models of SZ). Further, we have previously suggested that abnormal performance on a number of visual tasks, e.g. contour integration and visual crowding, can be explained by reduced levels of cortical suppression from a stimulus' surround (Robol et al., in press).

Analogous SS effects, which may involve similar mechanisms of gain control, have also been demonstrated for visual dimensions other than contrast (**Figures 1A,C,D**). For example, the brightness of a target is reduced when it is embedded in a high luminance surround (**Figure 1A**; Adelson, 1993), the perceived orientation of a target is shifted when it is presented within a surround with a different orientation (**Figure 1C**; Wenderoth and Johnstone, 1988), and the perceived size of a circle is reduced by the presence of large flanking circles (**Figure 1D**; the Ebbinghaus illusion; Weintraub, 1979). Although the extent to which these effects rely on common mechanisms is not well understood (Webb et al., 2005;

Smith, 2006), there is evidence for multiple gain control processes operating at different levels within the visual stream. Whilst luminance gain control is largely mediated by retinal processes (Shapley and Enroth-Cugell, 1984), the locus of SS for judgments of size, orientation and motion is thought to reside further downstream in striate and extra-striate areas once signals from the two eyes have converged. Thus, SS effects for these latter dimensions survive dichoptic presentation, i.e. when the target and surround are presented separately to different eyes (Marshak and Sekuler, 1979; Mather and Moulden, 1980; Wade, 1980; Song et al., 2011). This is not the case for contrast SS however, which only persists if the target and surround are presented to the same eye, implicating an intermediary locus of contrast gain control in pre-cortical or early cortical areas (Chubb et al., 1989).

**(A)** higher luminance, **(B)** higher-contrast, **(C)** more anti-clockwise orientation,

There is evidence to suggest that several of these SS effects may be diminished in SZ, potentially implicating a widespread deficit in gain control mechanisms. Thus, in addition to well-documented abnormalities in contrast SS as described above (Dakin et al., 2005; Yoon et al., 2009, 2010; Barch et al., 2012), reduced SS is seen for judgments of motion direction in patients with SZ (Tadin et al., whereas the true/physical match is at "6 O'clock."

2006) as well as judgments of size in patients with disorganized SZ (Uhlhaas et al., 2006a,b) and non-clinical adults who score highly on a disordered thought sub-score of the Schizotypal Personality Questionnaire (Uhlhaas et al., 2004). However, not all findings reported are consistent with the notion of a widespread deficit. One study attributed *elevated* motion SS effects to individuals with SZ (relative to controls; Chen et al., 2008), and in a recent report of SS effects in the luminance, size, contrast, orientation, and motion domains, weakened contextual modulations in SZ were only reported for judgments of contrast (Yang et al., 2013). Hence, there is conflicting evidence for a general versus a dimension-specific deficit in contextual modulation in SZ.

One problem in establishing the generality of SS deficits in SZ, is that with the exception of a single report (Yang et al., 2013), data for the various versions of the task have typically been obtained using different patient groups, sample sizes, and/or experimental paradigms, thereby hindering direct comparison. To address this limitation, we assessed SS in a single patient group (*n* = 24) using a standardized battery of tasks (**Figures 1A–D**) and stimuli designed to probe visual function at multiple stages of the

processing hierarchy. Observers made judgments of relative luminance, contrast, orientation and size in the context of a reference stimulus embedded in a suppressive surround. We hypothesized that reduced SS stems from a generalized deficit in SZ, and consequently, predicted that patients would be less influenced than controls by the presence of a surround for all four judgment types.

# **MATERIALS AND METHODS**

#### **OBSERVERS**

Twenty-four observers with SZ (eightfemale;mean age = 39.96 ± 9 years; mean I.Q. = 102 ± 10) and 24 age-/sex- and IQ-matched controls (mean age = 38.21 ± 12 years; mean I.Q. = 107 ± 9) gave informed written consent to take part in this study (see **Table 1**). Patient and control groups did not differ significantly with respect to age [*t*(46) = −0.58; *P* = 0.57] or I.Q. [*t*(46) = 1.74; *P* = 0.09]. Patients were recruited from inpatients at the Churchill London Clinic (*n* = 5) and from outpatients at the Institute of Psychiatry (IoP); all had been diagnosed with SZ according to DSM-IV criteria. At the IoP and Churchill London Clinic clinical assessments

were undertaken by a Masters level research nurse and clinical psychologist, respectively, both of whom have extensive knowledge and training in the field. Of the 24 patients tested, 12 were diagnosed with paranoid SZ; however, none of the other patients fell firmly into any other specific sub-category. Details of patients' medication are given in **Table 1**. Ethics approval was granted by the UK National Research Ethics Committee.

#### **APPARATUS**

Stimuli were presented on a CRT monitor (LaCie Electron Blue 22), which was viewed at 120 cm at a spatial and temporal frequency of 1024 × 768 pixels and 75 Hz, respectively. The monitor was fitted with a Bits++ box (Cambridge Research Systems) operating in Mono++ mode to give true 14-bit contrast accuracy. The display was calibrated with a Minolta LS110 photometer and linearized using custom software using a look-up table. Experiments were run in the Matlab programming environment (MathWorks, Cambridge, MA, USA) – in conjunction with Psychtoolbox (Brainard, 1997; Pelli, 1997) – on an Apple MacBook Pro computer.

**Table 1 | Patient details, including medication type (Med), medication dose (Dose; chlorpromazine equivalent in mg/day), diagnosis (SZ, schizophrenia; PS, paranoid schizophrenia), intelligence quotient (I.Q.), total scores on tPANSS, scores for the positive symptoms of the PANSS test (tPSS), scores for the negative symptoms of the PANSS test (tNSS), scores for the general symptoms of the PANSS test (tGSS), scores on a cognitive factor which overlaps heavily with the concept of disorganization syndrome (tDIS) and scores for item P2 on the PANSS test, "conceptual disorganization" (DIS).**


Only one patient had a history of substance abuse. Typ, typical; Atyp, atypical; diag, diagnosis; std, standard deviation.

## **PROCEDURE**

All experiments (presented in pseudo-random order) involved a two-(temporal)-interval-forced choice (2-IFC) task in which the observer had to report which of two stimuli (the reference or target) was"the brighter"(luminance task),"stronger"(contrast task), "tilted closer to the horizontal" (orientation task), or "larger" (size task). Observers gave a verbal response on each trial, which the experimenter then relayed to the computer by button-press. The entire testing session (including obtaining of consent and debriefing) lasted approximately 1.5 h, with around 20 min allocated for each experiment. For examples of the stimuli used see **Figure 1**, but note that in the actual experiment the reference and target were presented centrally in two separate temporal intervals. Stimulus presentation time was fixed at 500 ms with an inter-stimulus interval of 1250 ms. Fixation was assisted by the presence of a central black cross, which turned white when stimuli were presented onscreen.

The target stimulus consisted of an isolated disk with a radius subtending 0.34˚ of visual angle. The reference was a disk of identical dimensions embedded in an annular surround with an outer radius of 1.91˚. For all four experiments the reference (surrounded) stimulus did not vary across trials, whereas the test stimulus varied either in luminance, contrast, orientation, or size, depending on experiment. The test value presented on any single trial was controlled by an adaptive method (Watt and Andrews, 1981) that sought to probe responses that would be maximally informative about the slope and offset (bias) of the underlying psychometric function, which was approximated by a cumulative Gaussian (see below). All runs consisted of 64 trials, during which

the signal level (luminance, contrast, orientation, or size of the test) was manipulated.

For each task and each individual the probability that the observer reported that the test had a higher signal than the reference (**Figure 2** upper left panel) was plotted against the signal level of the test (luminance in the luminance task for example). Data follow a sigmoidal distribution that is well fit by a cumulative Gaussian function defined by two parameters: a slope and a bias. The form of the cumulative Gaussian captures the fact that when the test is considerably brighter than the reference, observers (almost) always report the test as having the higher signal, whereas when the reference is much brighter than the test, participants (almost) never report the test as having the higher signal. The slope of the line connecting these two asymptotes defines how sensitive the observer is to changes in the relative signal of the test and reference. A steep slope represents high sensitivity, such that small changes in the signal difference elicit large changes in the responses of the observer. Sensory *threshold* is the inverse of the slope of a cumulative Gaussian function, and defines the difference in test signal and reference-signal that is needed for the observer to respond correctly on a certain percentage of trials (defined here as 84%). For example, a threshold of 5 cd/m<sup>2</sup> implies that the test and reference must differ in luminance by 5 cd/m<sup>2</sup> for the observer to correctly discriminate their brightness on 84% of trials. Thus, a high threshold represents poor performance.

The second parameter of the cumulative Gaussian, the offset or bias, defines the function's mid-point. Also known as the point of subjective equality (PSE), it is the test signal level at which the observer reports the test as having the higher signal

**FIGURE 2 | Data taken from a single control observer (upper plots) and a single observer with schizophrenia (lower plots).** These were selected on the basis that they were typical of group trends. Red triangles denote the point of subjective equality (PSE) on the abscissa, which represents the test signal level at which the test

and reference were perceptually matched. The white triangles denote the test signal when test and reference were physically (i.e. veridically) matched. Error bars represent 95% confidence intervals for parameter estimates obtained through boot-strapping of the observer's responses.

on 50% of trials, i.e. the point at which the test and reference are indistinguishable for the dimension of interest, and hence are perceived as matched. For the experiments reported here, the bias is expressed relative to the *veridical* match point and represents the shift in the perceived signal level of the reference as a result of being embedded in the surround. A negative bias is therefore indicative of a strong suppressive effect. For example, a bias of −10 cd/m<sup>2</sup> would imply that the perceived luminance of the reference is reduced by 10 cd/m<sup>2</sup> when embedded in the surround.

Ninety-five percent confidence intervals (CIs) and the standard deviation of fit parameter estimates (threshold and bias) were obtained through boot-strapping (re-sampling) and re-fitting of the raw data. These were subsequently *Z*-transformed (expressed as units of standard deviation relative to the mean) and used to derive weightings for each parameter estimate; these followed an inverse cumulative Gaussian such that parameter estimates associated with smaller CIs (i.e. higher confidence) contributed most heavily to the analyses [weighted *t*-tests and correlations were undertaken using Matlab and SPSS statistical analysis software (version 18.0; SPSS Inc., Chicago, IL, USA)]. In addition, data with an associated confidence interval for the bias parameter >2.58 standard deviations from the mean were excluded on the basis that they were extreme outliers and reflected data that were poorly fit by the cumulative Gaussian model. (The mean ± 2.58 standard deviations captures the 1st to 99th percentile of a normal probability distribution function). All statistical analyses undertaken employed two-tailed tests, unless stated otherwise.

#### **STIMULI**

For the **luminance task**, the reference patch, the test patch, and the surround-annulus were random-noise filtered with a spatialfrequency (SF) band-pass LogGabor filter passing a mean SF of 11.25 c/deg. with a bandwidth of 0.4 octave (the σ of the log-Gaussian defining the filter in the fourier domain). Michelson contrast was fixed at 50%. The mean-luminance of the reference and its uniform background were fixed at 50 cd/m<sup>2</sup> and the reference-surround at 75 cd/m<sup>2</sup> . The luminance of the test fell between 37.5 and 62.5 cd/m<sup>2</sup> depending on performance.

For the **contrast task**, the reference, test, and surround again consisted of LogGabor filtered noise (same characteristics as in the luminance task, except that luminance was now fixed at 50 cd/m<sup>2</sup> and the contrast of the reference and surround were set at 40 and 95%, respectively). The contrast of the test patch varied between 4 and 80%.

For the **orientation task**, the reference, test, and surround were comprised of similar-(LogGabor) filtered noise which had also been orientation-limited using a wrapped Gaussian weighted passband with a 5˚ (σ) bandwidth. Test luminance was 50 cd/m<sup>2</sup> and contrast 95%. The reference had a mean orientation of 55˚ (anticlockwise from horizontal), and was embedded in a 75˚ surround. The test orientation varied between 35˚ and 75˚.

For the **size task** the reference was a ring with a radius of 0.34˚ surrounded by four large circles, each with a radius of 0.7˚ with a 1.1˚ separation. The test ring radius varied within a range of 0.22˚– 0.44˚. Ring edges had a sinusoidal profile with a SF of 11.25 c/deg. presented at 95% contrast and an average luminance of 50 cd/m<sup>2</sup> .

# **RESULTS**

# **INDIVIDUAL DATA**

**Figure 2** shows data from one control observer (top row) and one observer with SZ (bottom row). On the abscissa the test signal level (i.e. luminance, contrast, orientation, or size of the test) is plotted against the probability that the observer reported that the test was brighter, higher-contrast, clockwise, or larger compared to the reference (respectively). Data have been fit with a cumulative Gaussian function (solid black line) defined by two parameters: the threshold and bias (the positioning of the curve along the abscissa). In addition, 95% CIs were generated for each of these parameters using a method of boot-strapping. The threshold indicates the smallest difference in signal between the reference and the test that would allow the observer to correctly discriminate the two on 84% of trials. The bias (red triangle) (with associated 95% CIs; red lines) represents the test signal level for which the reference and test are *perceived* as matched, so that the observer reports the test as brighter on 50% of the trials. The *actual* signal level of the reference is denoted by a white triangle for comparison; if judgments were unbiased, the red and white triangles would coincide. For the luminance task (**Figure 2**; upper left plot), note that the perceived match point (red triangle) lies to the left of the actual reference stimulus level (white triangle). Thus, for the test to be perceptually matched to a reference with a luminance of 50 cd/m<sup>2</sup> , it must have a luminance of approximately 36 cd/m<sup>2</sup> , consistent with the midgray reference patch appearing darker when presented in a bright surround.

#### **CONFIDENCE OF PARAMETER ESTIMATION**

Before comparing thresholds and biases between the control and patient groups, data were filtered to remove any values that were associated with high uncertainty. Thus, parameter estimates with an associated CI that was >2.58 standard deviations from the group mean were excluded from the analysis (see Materials and Methods). This resulted in the exclusion of 7% of the patients' data, compared to only 1% of the control group's data. Following exclusion of these extreme outliers, control, and patient CIs were compared across the four tasks using a multivariate analysis of variance (MANOVA) with four dependent variables (performance on the luminance, contrast, orientation, and size tasks) and one independent variable (group at two levels: patients and controls). Bias and threshold CIs were analyzed independently. These revealed a main effect of group for biases [*F*(4,33) = 2.92; Wilk's λ = 0.74, *P* = 0.04, partial η <sup>2</sup> = 0.26], but not for thresholds [*F*(4,33) = 2.5;Wilk's λ = 0.77, *P* = 0.06, partial η <sup>2</sup> = 0.23]. To explore these findings further, a series of independent-samples *t*tests were undertaken. Significance was defined at an alpha level of 0.0125 (correction for four multiple comparisons). CIs for both the bias and threshold parameter were found to be significantly elevated in the patients (relative to controls) for judgments of orientation only (*P*s < 0.01; **Table 2**). Consequently, after filtering for extreme outliers, we find some evidence for a poorer fit to the patient group data, although the effect is not consistent across tasks.

#### **GROUP BIASES AND THRESHOLDS**

**Figures 3A–D** plot bias against threshold data derived from the psychometric functions fit individually to each observer's data.

**Table 2 | Independent-samples t-tests comparing patient and control group 95% confidence intervals for biases and thresholds.**


Alpha level = 0.0125, reflecting correction for four multiple comparisons. P, significance level; d, Cohen's d; \*significant effect for given alpha level.

Blue/white and red/white square data-points represent group averages for the control and patient groups respectively. Notice that for all tasks the patient group data fall above – and with the exception of the luminance task – to the right of the control group data. To explore this separation quantitatively we used MANOVA with four dependent variables (performance on the luminance, contrast, orientation, and size tasks) and one independent variable (group at two levels: patients and controls). Biases and thresholds were analyzed separately. Analyses highlighted a significant main effect of group on biases [*F*(4,33) = 2.74; Wilk's λ = 0.75, *P* = 0.05, partial η <sup>2</sup> = 0.25] as well as thresholds [*F*(4,33) = 2.96; Wilk's λ = 0.74, *P* = 0.03, partial η <sup>2</sup> = 0.26].

To examine which tasks underlie these effects a series of independent-samples *t*-tests (weighted by parameter confidence) were carried out to compare group biases and group thresholds on individual tasks. Statistical significance was defined at an alpha level of 0.025, reflecting adjustment of the standard value (0.05) for a single-tailed test and Bonferroni correction for four multiple comparisons (reflecting a total of four separate tasks, with biases and thresholds once again tested independently). Singletailed tests were employed since on the basis of previous literature our hypothesis was unidirectional: with the exception of a single study of motion (which was not tested here; Chen et al., 2008), measures of SS have only ever highlighted significant associations between *high* schizotypal traits (or SZ itself) and *reduced* contextual modulation (Uhlhaas et al., 2004, 2006a,b; Dakin et al., 2005; Tadin et al., 2006; Yoon et al., 2009, 2010; Barch et al., 2012; Yang et al., 2013). Thresholds were found to be significantly higher in patients with SZ (relative to controls) for the contrast judgments only (*P* < 0.001; see **Table 3** and **Figure 3B**). For biases there was a general trend for all data to fall to the left of the zero bias line (veridical match), reflecting the effect of a suppressive surround. However, relative to control values, these biases were found to be significantly reduced (less negative) in the patients with SZ for both contrast and size judgments (*P* = 0.025 and *P* = 0.02, respectively). This suggests that the patients with SZ were less susceptible to the suppressive effects of context in both the contrast and size domains.

To compare contextual modulation effects across tasks and to facilitate comparison of effect sizes with previous studies, SZ group biases were re-plotted following *z*-score transformation relative to control group data, as in Yang et al. (2013), but with all parameters weighted by confidence (**Figure 4**). Note that this is a signed measure of the effect size: a variation of Cohen's *d*, in which group differences are normalized by the control group variance as opposed to the pooled variance (also known as Glass's delta). Negative values reflect reduced suppression in the patients relative to controls. Reinforcing the findings of statistical comparisons, the data show reduced suppression in the contrast, orientation, and size domains, with the largest effects for judgments of relative contrast and size (0.68 and 0.52, respectively; **Figure 4** and **Table 3**). FollowingYang et al. (2013),we also generated a *general* contextual modulation index (CMI) for each patient; this was calculated by averaging individuals' *Z*-transformed biases across all four tasks. Once again, negative values imply a suppressive effect of the surround. The group mean CMI for the patient group was −0.4 with a standard deviation of 0.62 (**Figure 4**). A comparison of control and patient CMIs reveal a significant difference at the single-tailed, but not two-tailed, level [*t*(46) = 1.71, *P* = 0.09].

#### **CORRELATIONS BETWEEN TASKS**

If SS along distinct visual dimensions relies on shared mechanisms of gain control one might expect to find correlations between biases from distinct SS tasks. In **Table 4**, weighted Pearson's correlation coefficients and associated *P* values are presented for correlations between biases for all four tasks. We also tested for correlations between thresholds and biases in order to check that the reduced biases we report in the patient group are not simply an artifact of elevated thresholds. If this were the case we would expect biases and thresholds to correlate within each SS task. Statistical significance was defined at an alpha level of 0.0063, reflecting Bonferroni correction for eight multiple comparisons/parameter. Analyses show that none of the biases from one task correlated with biases from another task (all *P*s > 0.08), a finding that is consistent with independent mechanisms of SS across the different dimensions. In addition, we report no significant correlations between thresholds and biases on any single task (all *P*s > 0.16), suggesting any group differences in biases are unlikely to be driven by an artifact of elevated thresholds. Interestingly, there was a strong and significant correlation between luminance thresholds and contrast thresholds (*P* = 0.005).

#### **CORRELATIONS BETWEEN TASK PERFORMANCE, CLINICAL SYMPTOMS AND MEDICATION DOSE**

To determine whether group differences in bias and threshold could be attributed to the patients' anti-psychotic medication we also tested for correlations between patients' daily medication dose (converted into chlorpromazine equivalents; Woods, 2003; Taylor et al., 2012) and scores on behavioral measures including CMIs (**Table 5**). Statistical significance was defined at an alpha of 0.025, reflecting adjustment of the standard value (0.05) for a single-tailed test and Bonferroni adjustment for clinical symptoms multiple comparisons. Single-tailed tests were employed since our experimental hypothesis – that *elevated* SS in the patients can be attributed to medication – was unidirectional. Weighted Pearson's correlation coefficients and associated *P* values indicate that none of the comparisons approached statistical significance (all *P*s > 0.25). Consequently, we find no evidence that differences

between patient and control group data can be attributed to patients' medication.

Finally, to determine whether any of the behavioral measures tested correlated with symptom severity we calculated weighted Pearson's correlation coefficients between behavioral measures (including CMIs) and individual total PANSS scores the positive and negative symptoms scale (tPANSS), as well as negative, positive, general psychopathology sub-scale scores, and a cognitive factor that overlaps heavily with the concept of disorganization syndrome (Lindenmayer et al., 1994). This cognitive factor is based on the scoring of patients to a subset of questions in the PANSS test [Poor attention (G11); mannerisms and posturing (G5); conceptual disorganization (P2); difficulty in abstract thinking (N5); disorientation (G10)] and has previously been shown to correlate with poor performance in SZ on a contour integration task (Silverstein et al., 2000). In addition, following Uhlhaas et al. (2006a), we also looked for correlations between task performance and scores on question P2 of the PANSS test (conceptual

disorganization; DIS). The alpha level was set to 0.0083, reflecting Bonferroni correction for six multiple comparisons (a total of six PANSS measures recorded). No significant correlations were found between behavioral measures and any of the PANSS scores (**Table 6**).

#### **DISCUSSION**

We report that patients were more accurate (less biased) than healthy controls for judgments of relative contrast and size, implicating a reduced influence of surrounding context within these visual domains. However, with respect to our stated hypothesis (that attenuated contextual modulation is a general property of the visual system in SZ), we do not report evidence for a deficit across *all* tasks employed; patients showed SS effects that were statistically indistinguishable from controls' for stimuli defined both by luminance and orientation.

These findings are consistent with a number of studies that have reported comparable deficits in contextual modulation using

**Table 3 |Weighted t-tests comparing patient and control group biases and thresholds.**


Alpha level = 0.025, reflecting correction for single-tailed test and four multiple comparisons. P, significance level; d, Cohen's d with individual values weighted by parameter confidence; \*significant effect for given alpha level.

analogous SS tasks in studies of SZ and schizotypal traits. Thus, reduced SS effects have previously been reported for judgments of relative contrast (Dakin et al., 2005; Yoon et al., 2009, 2010; Barch et al., 2012) and motion (Tadin et al., 2006) in patients with SZ, as well as size in patients with disorganized SZ (Uhlhaas et al., 2006a,b) and non-clinical adults who scored highly on a disordered thought sub-score of the Schizotypal Personality Questionnaire (Uhlhaas et al., 2004). Although this implicates attenuated SS in SZ for a number of visual dimensions, not all studies support this conclusion. Using random-dot motion stimuli Chen et al. (2008) reported a reversed pattern of effects: *elevated* SS in patients with SZ relative to matched controls. In addition, as the majority of these studies have typically employed distinct experimental paradigms and heterogeneous patient groups it is difficult to compare findings across tasks and draw general conclusions.

One recent publication has attempted to address this limitation in the literature directly. A study by Yang et al. (2013) used a similar design and experimental approach to test a single group of patients with SZ on a batch of SS tasks that measured contextual modulations for judgments of relative luminance, contrast, orientation, size *and* motion. They reported attenuated contrast SS in the patient group relative to controls – with a similar effect size to our own: Cohen's *d* = 0.64 compared to 0.68 – but found no evidence for group differences on any of the other dimensions tested (see points raised below however). This effect size is considerably weaker than that found in the original study by Dakin et al. (2005), but larger than reported by Barch et al.(2012; Cohen's *d* = 0.31). This is unsurprising however: whilst Barch et al. (2012) only tested stable outpatients and Dakin et al. (2005) forensic inpatients (who were chronically ill), our own data were based on a mixture of inpatient and outpatient populations. Taken together, these findings support the notion that despite preserved mechanisms of luminance gain control in SZ, contrast SS is attenuated relative to controls. Further, whilst several studies have demonstrated an analogous deficit for size SS in a subgroup of patients and nonclinical adults with disordered thought, there is inconsistent and somewhat contradictory evidence as to whether or not orientation and motion judgments are similarly affected (Tadin et al., 2006; Chen et al., 2008; Yoon et al., 2009, 2010; Yang et al., 2013).

**FIGURE 4 | Bias data from the schizophrenia group have been converted into z-scores relative to control group means and standard deviations for each of the four tasks, with individual parameter estimates weighted by their associated confidence interval.** Negative and positive values, respectively, denote weaker and stronger contextual effects in the patient group. For each patient, a mean of these four standardized z-scores was calculated, generating a contextual modulation index (CMI) that represents a measure of general susceptibility to surround suppression (white bar). Error bars indicate the standard error of the mean (SEM) of the patient group; the blue shaded region indicates the SEM of the control group.\*Significant effect at the 5% level following correction for multiple comparisons and single-tailed tests.

One potentially important distinction between Yang et al. (2013) and other studies (including our own), is that with the exception of the motion task, unlimited exposure times were used: stimuli remained onscreen until the observer gave a response, a design that may be suboptimal for uncovering group differences. With respect to orientation at least, the magnitude of SS is dependent on stimulus presentation time (Calvert and Harris, 1988), such that the effect diminishes at durations >100 milliseconds. Consequently, prolonged exposure to the stimulus may reduce the likelihood of uncovering group differences by minimizing baseline biases, thereby risking floor effects. In support of this possibility, whilst Yang et al. (2013) report that a high-signal (oriented) surround shifted the perceived orientation of a target by an average of 2.86˚ in the control group, we report a mean shift of 10.84˚ using our briefly presented stimuli. Despite these discrepancies, both the work of Yang et al. (2013) and the findings of our own study point to the existence of preserved mechanisms of luminance gain control in SZ: patients showed normal SS effects for luminance judgments. This raises the possibility that the notional visual dysfunction in SZ may be restricted to cortical as opposed to pre-cortical loci and places a theoretical lower bound on the deficit. Relative to other visual dimensions, e.g. motion, orientation, and size (Marshak and Sekuler, 1979; Mather and Moulden, 1980; Wade, 1980; Song


#### **Table 4 | Inter-correlations between biases and thresholds for each of the four tasks tested.**

Alpha level = 0.0063, reflecting correction for 8 multiple comparisons/task. \*Significant effect for given alpha level.



Alpha level = 0.025, reflecting correction for single-tailed test and four multiple comparisons. Lum, luminance; Cont, contrast; Orient, orientation; B, bias;T, threshold; CMI, contextual modulation index.

et al., 2011), luminance signals are processed at the very earliest stages of the visual hierarchy, within the retina and lateral geniculate nucleus (LGN; Shapley and Enroth-Cugell, 1984). Indeed, there is some support from post-mortem anatomical studies of SZ for deficits within the visual system being restricted to cortical loci: whilst there is a 25% reduction in neuron number (and 22% reduction in total volume) in the primary visual cortex in patients with SZ (relative to controls; Dorph-Petersen et al., 2007), no such deficit was found in the LGN (Lesch and Bogerts, 1984; Selemon and Begovic, 2007; Dorph-Petersen et al., 2009) – an important pre-cortical site of luminance gain control.

Meta-analyses of magnetic resonance imaging (MRI) voxelbased morphometric studies, which quantify regional differences in gray matter volume, do not highlight such a distinction between cortical and pre-cortical deficits in SZ however (Ellison-Wright et al., 2008; Honea et al., 2008; Fornito et al., 2009). Whilst frontal and temporal abnormalities are strongly associated with SZ (see Hulshoff Pol and Kahn, 2008 for a review), several subcortical – including thalamic-loci have also been implicated (Andreasen et al., 1994; Blennow et al., 1996; Staal et al., 1998), although at least some of these may reflect

secondary effects of treatment with anti-psychotic medication (Dazzan et al., 2005). Morphometric (Wright et al., 1995) and diffusion tensor imaging studies (Shergill et al., 2007) have also highlighted a number of white matter (neural fiber) defects in SZ that extend to fronto-thalamic connections (Suzuki et al., 2002). Thus, although reported anatomical abnormalities in SZ are commonly *cortical* in nature, the existing literature clearly does not rule out the possibility of related deficits in subcortical structures.

What is to be made of our finding that levels of *orientation* SS also did not differ significantly between patient and control groups, a finding that corroborates the work of Yang et al. (2013)? This was contrary to our prediction: orientation SS (**Figure 1C**) is putatively driven by inhibition between cell populations tuned to similar orientations (Wenderoth and Johnstone, 1988), and as such, should be reduced in magnitude if cortical suppression is deficient in SZ. It is worth noting that although levels of orientation SS in SZ were statistically indistinguishable from controls', there was a trend in the same direction as for contrast and size: biases were lower in the patient group, raising the issue of statistical power. With a greater sample size it is possible that a group difference may have been uncovered. The findings of Yang et al.


**Table 6 | Correlations between the behavioral measures (biases and thresholds) and PANSS scores (patient data only).**

Alpha level = 0.0083, reflecting correction for six multiple comparisons. R, Pearson's correlation coefficient; P, significance level; Lum, luminance; Cont, contrast; Orient, orientation; B, bias; T, threshold; CMI, contextual modulation index.

(2013) make this unlikely however, as they report *elevated* biases in their patient group relative to controls, although this trend did not reach significance. In addition, as noted earlier, Yang et al. (2013) used unlimited exposure times, which may be critical to their findings.

There are several possible explanations as to why orientation SS is relatively normal in SZ. First, orientation SS may rely on distinct mechanisms and cortical networks from other forms of SS that have been implicated in SZ (e.g. contrast SS), and these may be relatively spared in SZ. This also seems unlikely however, as the suppressive inputs that drive contrast SS itself (which is affected in SZ) are tuned for orientation (Chubb et al., 1989; Solomon et al., 1993). An alternative possibility is that the stimuli used here (and previously) were suboptimal for capturing a deficit in the patient group. We used relatively broad-band reference and surround textures with mean orientations separated by 20˚. This design may have driven broadly tuned mechanisms of suppression, which are seemingly unaffected in SZ. Thus, there is evidence that orientation tuning curves are abnormally broad in SZ (Rokem et al., 2011), and further, that suppression deficits are specific to closely oriented (i.e. near-parallel) stimuli. In a study of contrast SS using oriented narrow-band stimuli suppression from a parallel surround was found to be reduced in SZ (relative to controls), whilst suppression from an orthogonal surround was, if anything, elevated (see Yoon et al., 2009, Figure 2B). Similarly, in a recent study of contour detection in SZ, whilst the presence of near-parallel flankers

selectively impaired performance in the patient group relative to controls (putatively via elevated suppression of activity driven by the contour), near-orthogonal flankers had no such differential effect on the two groups (Robol et al., in press). Therefore, in future studies it may be more informative to test for orientation effects using closely oriented narrow-band reference and surround textures.

With respect to the underlying pathophysiology of visual dysfunction in SZ there are a number of candidate neurotransmitter systems that have been heavily implicated in the disorder, most notably GABA (Wassef et al., 2003), dopamine (Howes and Kapur, 2009) and glutamate (Javitt, 2010). However, the data we report are largely consistent with studies that highlight the importance of reduced levels of GABA in SZ. GABA has been linked to reduced SS (Yoon et al., 2010) and broader orientation tuning (Rokem et al., 2011) in SZ, and GABA-mediated inhibition is thought to be critical to a number of other tasks that are affected in SZ, e.g. orientation discrimination (Edden et al., 2009; Robol et al., in press) and contour integration (Silverstein et al., 2000, 2006; Uhlhaas et al., 2006a,b). Further, a number of studies have highlighted the potential benefits of targeting GABAergic networks with pharmacological interventions (Wassef et al., 1999). For example, improvements in cognitive function have been demonstrated in patients with chronic SZ following administration of a GABA type A receptor sub-unit selective agonist (Lewis et al., 2008), whilst pharmacological induction of a GABA deficit in patients with SZ has been shown to exacerbate psychotic symptoms and perceptual abnormalities (Ahn et al., 2011). However, a meta-analysis of randomized controlled drug studies comparing the use of benzodiazepines, which directly enhance the effects of GABA at the receptor level, to anti-psychotics or placebo concluded that current evidence does not warrant their use in the treatment of SZ (Volz et al., 2007), although the authors simultaneously emphasized the poor quality and relative paucity of existing studies. Consequently, further research is needed to determine whether abnormalities of GABAergic function could underlie the visual dysfunction we report here.

To investigate a possible relationship between anti-psychotic medication and surround suppression in the patient group, we tested for correlations between patients' behavioral measures and prescribed drug dosages following conversion into chlorpromazine equivalents; no significant correlations were found. However, this approach does not provide a particularly rigorous test of medication-related confounds: equivalent doses are calculated on the basis of DA receptor type-2 binding affinity only, whilst antipsychotics typically affect multiple neurotransmitter systems and receptor types. Nonetheless, several other lines of evidence lead us to believe that the effects we report are unlikely to be driven by medication. First, the main effect reported (reduced SS) has been observed in disparate samples of patients with a wide range of medication regimes (Dakin et al., 2005; Tadin et al., 2006; Uhlhaas et al., 2006a,b; Yoon et al., 2009, 2010; Barch et al., 2012; Yang et al., 2013). Further, we have previously shown that the effect (for contrast) was specific to patients with SZ, and did not extend to a clinical control group with bipolar disorder, several of whom were also treated with low-dose anti-psychotics (Dakin et al., 2005). Lastly, the effects we report were not seen in all dimensions tested.

Other potential confounds of a non-visual nature, e.g. differences in general cognitive function, attention, or motivation (see O'Carroll, 2000; Rund, 1998 for reviews), should also be considered. Barch et al. (2012), using a similar contrast SS paradigm, have recently shown that following exclusion of participants on the basis of high lapse rates – a putative measure of attentional engagement – group differences in levels of SS between patients and controls essentially disappeared. The authors suggest that their findings imply reduced contrast SS in SZ may be largely due to impaired mechanisms of attention. The underlying logic is that the patients failed to attend to the stimulus on a significant proportion of trials, thereby leading to random responses that masked inherent (perceptual) biases. However, the effect size they report prior to filtering is relatively small (Cohen's *d* = 0.27–0.31). An alternative interpretation of the data therefore, is that by filtering outliers the authors simply reduced their power, and hence, chance of finding a significant effect. In support of this possibility it is worth noting that even after filtering out 25% of their patient data (66 out of 262), their reported effect remained near-significant (*P* = 0.08 – ANOVA interaction), and in fact, *would be* significant if a single-tailed test were applied, an approach that would be justified on the basis of previous published data. Hence, it may be too premature to conclude that reduced SS in SZ is due to impaired mechanisms of attention.

With respect to our own data, there are a number of reasons why we believe that the main findings reported, i.e. reduced contrast and size SS in the patient group, are unlikely to be due to an inability to attend to the task at hand. First, if this were the case, we would expect to find consistent inter-group differences across all tasks. Thus, why would lapses of attention be specific to a subset of visual judgments? Even if one *were* to posit a possible mechanism by which effects might be specific to a subset of tasks, for example if they were differentially demanding of available attentional resources, this leads to a clear prediction: that inter-group differences in the size of associated parameter CIs should be maximal – or at least evident – on those very tasks which are seemingly affected in SZ (i.e. judgments of contrast and size). In fact, the findings we report show clearly that this predicted relationship is *not* upheld: CIs were larger in the patient group for the orientation task only (for which biases did *not* differ between groups), and did *not* differ for the contrast task (on which biases *did* differ between groups). In addition, there was no correlation between biases and thresholds on individual tasks as one might also expect if levels of bias were an artifact of attentional lapses.

Taken together, the findings reported are inconsistent with inter-group differences in levels of SS being driven by higher attentional lapses in our patient group. This conclusion is further reinforced by the fact that the effect we report is evident following removal of extreme outliers defined on the basis of CIs, and the inverse weighting of individual data-points according to CIs. Both of these data processing stages would have excluded – and minimized the contribution of – participants who were unable to attend to the task, thereby fulfilling a similar function to excluding participants on the basis of lapse rates. Nonetheless, the inclusion of catch-trials to the basic experimental design represents an invaluable improvement; however, we would recommmend integrating these trials into the data-fitting stage by including a lapse rate as an additional parameter to the psychometric function. In this way data need not be discarded and power is retained.

Although we have argued against an explanation of our findings based on the notion of increased attentional lapses in the patient group, is it possible that some other inter-group difference in patterns of attentional deployment could be invoked? One possibility that should be considered is that patients with SZ simply have a smaller spotlight of attention than control participants. If this were the case, then patients might attend less to the high-signal surround, thereby attenuating its suppressive effects (Shulman, 1992; Sundberg et al., 2009). In support of this possibility, there is evidence that on any given fixation, patients with SZ process information from a smaller area of the visual field than do controls (Elahipanah et al., 2011). In addition, patients with SZ are less sensitive at detecting peripheral stimuli during a concurrent foveal discrimination task (Cegalis and Deptula, 1981) or a visual search task (Elahipanah et al., 2010). However, as with attentional lapses, any explanation of this kind needs to address the fact that reported group differences are specific to a subset of SS effects (i.e. luminance judgments are not affected). Further, it is worth noting that for contrast judgments at least (the most robust finding to date), reduced SS in the patients persists when stimuli are maintained onscreen until a response is given (Yang et al., 2013), conditions under which observers are likely to have made multiple saccades, thereby minimizing any influence of differences in the spotlight of attention.

In conclusion, our data suggest that reduced SS characterizes the visual system in SZ, affecting visual judgments of relative contrast and size. However, as not all visual dimensions were affected, we must reject the hypothesis that attenuated suppression is a general (ubiquitous) property of the brain in SZ. Specifically, as the data did not implicate abnormal *luminance* gain control – a finding that has been reported previously – we propose that the putative dysfunction may be predominantly cortical in origin. Considered within the context of previous research, our data suggest that an attenuated contrast SS effect represents a robust and pronounced feature of SZ, with clinical/diagnostic value. On the other hand, whilst the weight of evidence points toward the existence of analogous abnormalities in the contextual processing of other visual dimensions (particularly size), several effects (e.g. for orientation and motion) seem to be more fragile. Future studies involving parametric manipulation of stimulus parameters and testing conditions are therefore needed if critical variables are to be identified and remaining discrepancies in the literature are to be resolved.

#### **ACKNOWLEDGMENTS**

This work was funded by the Wellcome Trust. The authors gratefully acknowledge the generous assistance of Cambian Healthcare in the data collection. The authors have no financial interests to disclose.

# **REFERENCES**


dopamine: a connectionist approach to behavior and biology in schizophrenia. *Psychol. Rev.* 99, 45–77.


thought disorder. *Schizophr. Res.* 68, 249–260.


direct and indirect tilt illusions. *Vision Res.* 28, 301–312.


schizophrenia. *Schizophr. Bull.* 35, 1078–1084.

Zenger-Landolt, B., and Heeger, D. J. (2003). Response suppression in v1 agrees with psychophysics of surround masking. *J. Neurosci.* 23, 6884–6893.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 19 November 2012; accepted: 07 February 2013; published online: 28 February 2013.*

*Citation: Tibber MS, Anderson EJ, Bobin T, Antonova E, Seabright A, Wright B, Carlin P, Shergill SS and Dakin SC (2013) Visual surround suppression in schizophrenia. Front. Psychol. 4:88. doi: 10.3389/fpsyg.2013.00088*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Tibber, Anderson, Bobin, Antonova, Seabright, Wright, Carlin, Shergill and Dakin. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Visual context processing in bipolar disorder: a comparison with schizophrenia

# *Eunice Yang1,2\*, Duje Tadin3,4, Davis M. Glasser 3, Sang Wook Hong1,5, Randolph Blake1,6 and Sohee Park1*

*<sup>1</sup> Department of Psychology, Vanderbilt University, Nashville, TN, USA*

*<sup>6</sup> Department of Brain and Cognitive Sciences, Seoul National University, Seoul, Republic of Korea*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Brian P. Keane, UMDNJ–Robert Wood Johnson Medical School; Rutgers University Center for Cognitive Science, USA Matthew W. Roché, University of Medicine and Dentistry of New Jersey, USA*

#### *\*Correspondence:*

*Eunice Yang, School of Optometry, University of California, Berkeley, 486 Minor Hall, Berkeley, CA 94720, USA*

*e-mail: euniceyang@berkeley.edu*

Anomalous perception has been investigated extensively in schizophrenia, but it is unclear whether these impairments are specific to schizophrenia or extend to other psychotic disorders. Recent studies of visual context processing in schizophrenia (Tibber et al., 2013; Yang et al., 2013) point to circumscribed, task-specific abnormalities. Here we examined visual contextual processing across a comprehensive set of visual tasks in individuals with bipolar disorder and compared their performance with that of our previously published results from schizophrenia and healthy participants tested on those same tasks. We quantified the degree to which the surrounding visual context alters a center stimulus' appearance for brightness, size, contrast, orientation and motion. Across these tasks, healthy participants showed robust contextual effects, as indicated by pronounced misperceptions of the center stimuli. Participants with bipolar disorder showed contextual effects similar in magnitude to those found in healthy participants on all tasks. This result differs from what we found in schizophrenia participants (Yang et al., 2013) who showed weakened contextual modulations of contrast but intact contextual modulations of perceived luminance and size. Yet in schizophrenia participants, the magnitude of the contrast illusion did not correlate with symptom measures. Performance on the contrast task by the bipolar disorder group also could not be distinguished from that of the schizophrenia group, and this may be attributed to the result that bipolar patients who presented with greater manic symptoms showed weaker contrast modulation. Thus, contrast gain control may be modulated by clinical state in bipolar disorder. Stronger motion and orientation context effects correlated with worse clinical symptoms across both patient groups and especially in schizophrenia participants. These results highlight the complexity of visual context processing in schizophrenia and bipolar disorder.

**Keywords: bipolar disorder, contextual effects, perception deficit, visual processing, schizophrenia**

# **INTRODUCTION**

Visual dysfunction represents a core dimension of schizophrenia, but its role in the etiology of the disease has yet to be defined. To address this shortcoming, recent studies have investigated a number of visual functions as potential biomarkers for the disease, with contextual processing being one of those candidates (Carter and Barch, 2007; Gold et al., 2012). Contextual processing serves to enhance differences among visual features and consequently facilitate their segmentation from their background (Albright and Stoner, 2002). As a result, the perceptual appearance of a visual feature is altered in such a way as to emphasize its relative difference from features in the surrounding spatial context. Recent studies suggest that individuals with schizophrenia (SZ) are less influenced by visual context on some tasks, thereby enabling them to perceive the absolute characteristics of visual features more accurately (e.g., Tadin et al., 2006; Uhlhaas

Tibber et al., 2013). Considered together, these results seem to suggest existence of a generalized contextual processing deficit in schizophrenia. However, we recently showed that this deficit in contextual processing does not generalize across all contextual cues when explored within the same group of SZ: the magnitude of contextual modulations of luminance, size, orientation, and motion, were comparable to those of healthy controls, despite a weakened contextual effect of contrast (Yang et al., 2013). In addition, the strength of certain contextual illusions (i.e., orientation and motion repulsion) was predictive of symptom severity and

et al., 2006). For example, in the center-surround contrast illusion presence of a high-contrast background decreases the apparent contrast of smaller foreground features. Several studies have reported more accurate performance at judging stimulus contrast in SZ relative to controls, which implicates a weakened contextual effect of contrast (Dakin et al., 2005; Barch et al., 2012;

*<sup>2</sup> School of Optometry, University of California Berkeley, Berkeley, CA, USA*

*<sup>3</sup> Center for Visual Science and Department of Brain and Cognitive Sciences, University of Rochester, Rochester, NY, USA*

*<sup>4</sup> Department of Ophthalmology, University of Rochester, Rochester, NY, USA*

*<sup>5</sup> Department of Psychology, Florida Atlantic University, Boca Raton, FL, USA*

social functioning in SZ. Thus, impairments in contextual processing in schizophrenia may not be as wide-ranging as previously thought, and in those visual sub-modalities where impairment is evident the degree of impairment may be modulated by illness severity.

Knowing the diagnostic specificity of putative contextual processing abnormalities in schizophrenia is as important as understanding the conditions under which contextual deficits arise. It may be that contextual disturbances are related more broadly to psychosis rather than just the phenotype of schizophrenia. One approach for addressing this issue is to investigate contextual processes in individuals with other forms of psychosis, for example bipolar disorder. In one study, the surround contrast illusion was examined in individuals with bipolar disorder (BD), but they were a part of heterogeneous clinical "control" group, which consisted of individuals with affective, personality, and posttraumatic stress disorders (Dakin et al., 2005). Thus, while the clinical control group showed no contextual deficits, there was no information specific to bipolar disorder. Investigating contextual effects in bipolar disorder may also speak to the issue of whether schizophrenia and bipolar disorder occupy different regions of a continuum or are distinct disorders. They share similar symptoms such as hallucinations and delusions, are often treated with identical antipsychotic medications, and may share some genetic liability (Purcell et al., 2009; Van Snellenberg and De Candia, 2009). Some commonly reported visual deficits in schizophrenia are also found in bipolar disorder, including impairments in visual backward masking (Green et al., 1994; review by McClure, 1999), in vernier acuity (Kéri et al., 2004, 2007), and in early sensory processing measured with visually evoked potentials (Yeap et al., 2009). However, it must be noted that SZ and BD are distinguished by their performance on a very broad range of tasks from perceptual/cognitive to motor domains. In the visual perceptual domain, SZ and BD patients perform differently on tasks measuring photoreceptor sensitivity (Balogh et al., 2008), motion discrimination (Chen et al., 2006), and notably, contrast sensitivity modulation by collinear flanking stimuli (Kéri et al., 2005).

Given the current state of the literature, one cannot say whether the two disorders are distinct or fall on a continuum. Examining visual function in bipolar disorder may provide evidence for the specificity of contextual disturbances in schizophrenia if the two groups show distinct patterns of deficits and intact functions. This information, in turn, might be important to determine biomarkers specific to schizophrenia. The current study aims to systematically explore contextual processing in bipolar disorder in order to determine the extent to which contextual abnormalities are shared or specific to schizophrenia. We employed the same contextual tasks in BD as those used in our previous study of SZ and healthy controls (CO) and compared the pattern of contextual modulation in bipolar disorder to our previously published results (Yang et al., 2013).

# **MATERIALS AND METHODS**

#### **PARTICIPANTS**

Sixteen individuals who met the DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, fourth edition) criteria for bipolar disorder were recruited from Nashville, Tennessee. Diagnosis was confirmed by trained master's- and doctoral-level psychologists using the Structured Clinical Interview for the DSM-IV (First and Gibbon, 1997). Excluded from the study were individuals who reported any substance use within the last 6 months, and individuals with a history of neurological disorders or head trauma, or an IQ lower than 80 on the National Adult Reading Test (NART; Nelson, 1982). All participants had visual acuity of 20/30 or better (Optec Vision Tester 5000, Stereo Optical, Chicago, IL), with refractive correction if needed using a kit of trial lenses. Behavioral and clinical data of BD were compared with those of SZ (*N* = 30) and CO (*N* = 23) reported in our previous study (Yang et al., 2013).

**Table 1** summarizes the demographic information for the BD group tested in this study together with the SZ and CO individuals tested in our earlier work. The mean illness duration of BD was significantly briefer than the duration of illness in SZ [*t(*44*)* = 2*.*4, *p* = 0*.*02]. All but 2 BD were medicated (79% on atypical antipsychotic drugs, 86% on mood stabilizers, and 64% taking both). The mean chlorpromazine equivalent dose (CPZ) was significantly higher in SZ than in BD at the time of testing [*t(*34*)* = 3*.*1, *p* = 0*.*004]. Clinical symptoms in both patient groups were assessed with the Brief Psychiatric Rating Scale (BPRS; Overall and Gorham, 1962) and SZ and BD showed comparable BPRS scores (*p >* 0*.*05). BD were also rated on the Young Mania Rating Scale (YMRS; Young et al., 2000) and the Hamilton Rating Scale of Depression (HRSD; Hamilton, 1960). SZ were rated on the Scale of Assessment for Positive and Negative Symptoms or SAPS and SANS, respectively (Andreasen, 1983, 1984). Both patient groups were clinically stable at the time of testing, as assessed by the ratings scales mentioned above and by self-reports of episodes or hospital admittance in the last 6 months.

**Table 1 | Demographic and clinical information on subject groups.**


*Asterisks indicate significant group differences (p < 0.05). Parentheses denote standard deviation (SD). NART, national adult reading test; CPZ, chlorpromazine daily equivalent; BPRS, brief psychiatric rating scale; SAPS and SANS, scale of assessment for positive and negative symptoms, respectively; YMRS, young mania rating scale; HRSD, Hamilton rating scale of depression.*

There were no significant differences in mean age, mean NART IQ, and in the proportion of women among all three groups (all *p >* 0*.*05). Social functioning, as assessed with the Social Functioning Scale (Birchwood et al., 1990) was worse in both patient groups relative to CO [*F(*2*,* <sup>61</sup>*)* = 13*.*2, *p <* 0*.*001; BD vs. CO: *t(*34*)* = 4*.*0, *p <* 0*.*001; SZ vs. CO: *t(*43*)* = 5*.*4, *p <* 0*.*001]. The Institutional Review Board of Vanderbilt University approved this study protocol. All participants provided written informed consent and were paid.

#### **APPARATUS**

The study design was identical to that of Yang et al. (2013). Stimuli were created in MATLAB and the Psychophysics Toolbox (Brainard, 1997; Pelli, 1997) and were presented on a linearized CRT monitor (1280 × 960 resolution; 120 Hz). Viewing distance was 73 cm. Head position was stabilized by a chin rest. The display background was gray (luminance <sup>=</sup> 35.2 cd/m2, except in the brightness induction task, where luminance was 0.11 cd/m2). The ambient illumination was 0.16 cd/m2.

#### **CONTEXT BATTERY**

To assess contextual effects in a broad range of stimulus dimensions (luminance, contrast, size, orientation, and motion direction), we developed a battery of five psychophysical tasks (Yang et al., 2013). All tasks involved a center stimulus (**Figure 1**), whose perceptual appearance was altered by the presence of surrounding stimuli. In these tasks, participants were instructed to judge the appearance of the center stimulus by comparing it with a fixed reference stimulus (luminance, size, and contrast tasks) or by judging its deviation from vertical (motion and orientation tasks). To quantify the magnitude of the contextual modulation, the point of subjective equality (PSE) was measured for each task (as described below). PSEs were estimated by adaptive staircases for all tasks except the brightness induction task, where the method of adjustment was used. Stimuli were always presented until a response was made, except for the motion task, where stimulus duration was fixed at 200 ms. To establish baseline performance and to ensure that participants accurately judged stimulus dimensions tested in different tasks, all tasks included a no-context control condition. This condition was identical to the main context condition except that no surrounding context was present.

#### **BRIGHTNESS INDUCTION TASK**

The stimulus consisted of two circles (0.5◦ radius) surrounded by annuli (2.4◦ radius). They were simultaneously presented 15◦ apart (**Figure 1A**). The luminance of the reference circle (always shown on the left) was fixed at 6 cd/m2, while its surrounding annulus was set to 8, 12, 16, 20, or 24 cd/m2. A range of surround luminance values was included to allow comparison of the pattern of surround modulation between groups. The initial luminance of the target stimulus (always shown on the right) was randomly chosen from a range of 2–14 cd/m2, while the luminance of its annulus was fixed at 24 cd/m2. Fixed stimulus positions were used to control spatial inhomogeneities in screen luminance. Participants' task was to adjust the luminance of the target circle on the right to match the luminance of the reference circle

on the left. By pressing one of two keys, participants adjusted the target luminance, decreasing or increasing luminance in steps of 0.2 cd/m2. Three such adjustments were performed for each surround luminance, with their average taken as the PSE. The strength of brightness induction was defined as the difference (in cd/m2) between fixed luminance of the reference circle (6 cd/m2) and the perceived (i.e., adjusted) luminance of the target (which was typically much higher).

stimuli in 1A is not on the same scale as the size of the stimuli.

#### **SURROUND CONTRAST ILLUSION TASK**

The stimulus display (**Figure 1B**) was similar to surround contrast illusion stimuli used in previous studies (Chubb et al., 1989; Dakin et al., 2005; Barch et al., 2012). The display consisted of two circular patches (1.67◦ radius; 13.5◦ horizontal center-center separation). Each patch was filled with spatial frequency filtered noise (1 cycle/◦ center frequency; 0.25-octave bandwidth). The Michelson contrast of the reference patch was fixed at 20%, while the surrounding high-contrast noise annulus (6.67◦ radius) was shown at 97% contrast. The starting contrast of the target stimulus was randomly chosen (10–30%). On each trial, the positions of reference and target stimuli were randomly assigned, and participants judged which patch appeared higher in contrast by a key press. These responses were used to adaptively adjust the contrast of the target stimulus to match the apparent contrast of the reference stimulus.

#### **SURROUND MOTION REPULSION TASK**

The display (**Figure 1C**) consisted of a stimulus moving within a small circular aperture (1◦ radius) surrounded by another stimulus moving within a large annulus (6◦ radius). Stimulus speed for both the center and the annulus was 3◦/s. The stimuli were composed of spatial frequency filtered noise (80% contrast; 1◦/degree center frequency; 0.25-octave bandwidth). The direction of surround motion was either 45◦ clockwise or 45◦ counterclockwise from vertical. The direction of the center motion was either 18◦ clockwise or 18◦ counterclockwise at the start of the task and thereafter was varied by the staircase procedure. The stimuli were presented for 200 ms and, then, were immediately replaced with a blank screen. This was done to avoid pursuit eye movements. Participants' task was to indicate by a key press whether the central motion direction was clockwise or counterclockwise relative to vertical.

#### **SURROUND ORIENTATION REPULSION TASK**

The display (**Figure 1D**) consisted of a small circular grating (0.5◦ radius, 50% contrast, 3 cycles/◦) surrounded by a large, high-contrast annulus (4◦ radius, 97% contrast, 3 cycles/◦). The phase of each grating was random. The orientation of the annulus was always 15◦ counterclockwise from vertical. At the start of the task the center orientation was either 11◦ clockwise or counterclockwise and thereafter determined by the staircase procedure. Participants' task was to judge whether the center patch appeared tilted clockwise or counterclockwise relative to vertical and to indicate their responses by a key press.

# **EBBINGHAUS SIZE ILLUSION TASK**

This task was a variant of the classic Ebbinghaus illusion (**Figure 1E**). The display consisted of the target and the reference stimuli presented 15◦ apart (center-center). Their positions (left or right) were randomly assigned on each trial. The fixed reference stimulus consisted of a small dark circle (1.08◦ radius) surrounded by five evenly spaced large circles (2.17◦ radius and a 4◦ center-to-center distance from the reference stimulus). The target stimulus was a small circle. Its initial radius was randomly chosen between 0.92◦ and 1.08◦, and thereafter varied by the staircase procedure (described below). All stimuli were presented at 97% contrast. Participants' task was to judge which of the two center circles was larger and to indicate their responses by a key press.

#### **PROCEDURE**

The order of tasks was randomized for each participant. The experiment for each task consisted of four blocks, starting with the no-context control block and followed by three context blocks. In each block, two interleaved one-up/one-down staircases were used to estimate PSEs. The step size of these staircases decreased after every two reversals. The staircases converged after seven reversals. For each staircase, the PSE estimate was based on the average of the last four reversals. The resultant PSE for each participant was an average of six such staircases (two staircases in each of three blocks). For control tasks, PSEs were based on the average of two staircases. One exception was the brightness induction task, where the above-described adjustment method was used. No feedback was provided and there was no time limit for making a response. The entire context battery took about 1–1.5 h to complete. Before starting each task, participants were given detailed instructions. Each task started with five practice trials.

The strength of contextual effects was measured by quantifying the change in PSE values measured in the presence of a surrounding context relative to PSE values measured in the control condition with no surrounding context (i.e., as the degree to which a participant's perception changed after adding the surrounding context). The measurement units for luminance, contrast, and size tasks were cd/m2, log10 contrast, and arcmin, respectively. Orientation and motion angular repulsions were measured in degrees.

#### **PSYCHOMETRIC PROPERTIES**

We considered the following psychometric issues: ceiling effects, floor effects, and measurement reliability (**Table 2**). All tasks had inherent stimulus-constrained ceilings (e.g., 90◦ repulsion in the orientation task). All results were considerably weaker than these ceilings. Floor effects would be manifested as a "no contextual effect" for each task. However, because CO participants exhibited strong contextual effects, we had ample dynamic ranges to reveal potential weakening of contextual processing in clinical groups. Finally, we found no deviation from normality and equality of variance, as assessed with Kolmogorov-Smirnov test and Levene's test, respectively.

To examine measurement reliability, we split each data set into halves or thirds and correlated these partial data sets. **Table 2** depicts split-half reliability scores for each task and for each group. For size and contrast tasks, where we obtained six independent PSE estimates, we split the data into halves. For motion and orientation tasks, we obtained three pairs of measurements, where each pair consisted of two center directions/orientations. To assess measurement reliability, we correlated the second and third estimates. The modest correlation for CO (*r* = 0*.*58) in the orientation task is largely due to a single CO participant who failed to show a contextual effect on one measurement; without that individual's data, the reliability is 0.71. Note that somewhat lower numbers in motion and orientation tasks are expected, given that only two thirds of the data are used to compute reliabilities. Finally, second and third adjustment estimates in the brightness task were correlated. In sum, we found reliabilities for BD, SZ, and CO to be comparable for each task and relatively high. For BD, all split-data reliabilities were between 0.70 and 0.89 (average = 0.78).

# **ANALYSIS**

For all tasks except the brightness induction task, univariate analysis of variance (ANOVA) was used to compare performance measures of BD, SZ, and CO. In the brightness induction task,



repeated measures analysis of variance (ANOVA) compared performance between the three groups with five surround luminance conditions as the within-subjects factor. *Post-hoc* comparisons were performed using Welch's *t*-test. Effects sizes were reported for ANOVAs and *t*-tests using partial η<sup>2</sup> and Cohen's *d*, respectively. Below, we compare performance measures of BD with those of SZ and CO for each task. This is followed by combined analyses across tasks that compare the three groups using mixed model ANOVA. Pearson's *r* was used to determine correlations among contextual tasks, and Spearman's *rho* (*rs*) was used to test for correlations among contextual effects and clinical measures. We have reported both raw *p*-values and alpha levels adjusted for Bonferroni correction of multiple comparisons.

#### **RESULTS**

Owing to experimenter error, data for contrast and brightness tasks were missing for 1 BD participant. However, data on the other tasks were retained. Moreover, if any participant's data fell three standard deviations or more from the group mean, his or her baseline and context data were excluded for that task. This resulted in the exclusion of five data sets (three for motion and two for contrast), accounting for approximately 3% of all data. Two of the three outliers for the motion task were control participants. The two outliers in the contrast task were bipolar participants, one being the same outlier as in the motion task. ANOVAs revealed that BD, SZ, and CO did not significantly differ in each of the baseline conditions in which surround stimuli were absent (all *p >* 0*.*1). This result shows that patient groups had no problems accurately performing the visual tasks used in this study.

Surround contextual effects were observed across all tasks for each group (one-sample *t* tests, all *p* ≤ 0*.*001, adjusted α = 0*.*01, reflecting 5 comparisons per group, **Figure 2**). In the brightness induction task, we found a main effect of surround luminance [*F(*4*,* <sup>242</sup>*)* <sup>=</sup> <sup>423</sup>*.*0, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, partial <sup>η</sup><sup>2</sup> <sup>=</sup> <sup>0</sup>*.*87], but no main effect of group [*F(*2*,* <sup>63</sup>*)* <sup>=</sup> <sup>0</sup>*.*77, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*47, partial <sup>η</sup><sup>2</sup> <sup>=</sup> <sup>0</sup>*.*02] and no interaction between luminance and group [*F(*8*,* <sup>252</sup>*)* = 1*.*15, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*33, partial <sup>η</sup><sup>2</sup> <sup>=</sup> <sup>0</sup>*.*04]. We also compared group performance on the surround luminance condition that would evoke strongest illusion (surround luminance of 8 cd/m2) and found no significant difference across groups (**Table 3**; **Figure 2E**). The contextual effects of size, motion, and orientation, were not significantly different across the three groups (**Table 3**; **Figures 2A–C**). In the contrast task, the group difference reached significance only at a single-tailed unadjusted alpha level (*p* = 0*.*096, α = 0*.*05). Our previous study showed weakened contextual modulation of contrast in SZ relative to CO [*t(*42*)* = 4*.*87, *p* = 0*.*03, *d* = 0*.*64]. For the purposes of our study, we thought it was worth employing paired *t*-tests to determine whether the contextual contrast effect in BD was similar in magnitude to that of SZ and to CO. Indeed, BD was not significantly different from either CO [*t(*25*)* = 0*.*76, *p* = 0*.*39, *d* = 0*.*29] or SZ [*t(*20*)* = 0*.*75, *p* = 0*.*4, *d* = 0*.*33; **Figure 2D**] in the contrast illusion.

To examine the pattern of results across all contextual tasks, we normalized effect sizes for each task relative to the performance of CO to derive z scores (**Figure 3**). For the brightness induction

**with the strongest surround modulation (8 cd/m2) is represented).** Data points represent individuals within each group and bars denote mean group performance. The only significant group difference (∗) was weaker contextual modulation of contrast in SZ relative to CO [*t(*42*)* = 4*.*87, *p* = 0*.*03, *d* = 0*.*64].

task, we used the PSE estimate in the surround luminance condition that would evoke the strongest illusion (surround luminance of 8 cd/m2). Using a mixed model ANOVA with task (5) and group (3) as fixed factors, we found no significant main effect of group, *F(*2*,* <sup>65</sup>*)* = 0*.*54, *p* = 0*.*59; or task, *F(*4*,* <sup>64</sup>*)* = 1*.*6, *p* = 0*.*18; nor a significant interaction between group and task, *F(*8*,* <sup>64</sup>*)* = 1*.*5, *p* = 0*.*16.

To estimate a general measure of contextual processing, we derived a contextual modulation index (CMI) for each patient by averaging *z* scores across tasks (relative to CO). If BD showed a general weakening of contextual processing, then CMI should be negative. A positive CMI would indicate a general strengthening of contextual processing. The result for BD, however, is a *z* value of −0.2, with an associated *p*-value of 0.84 (**Figure 3**). In other words, CMI is nearly zero for BD, as was the CMI for SZ (*z* = −0*.*048, *p* = 0*.*96). Furthermore, variance did not differ between groups [Levene's test: *F(*2*,* <sup>66</sup>*)* = 0*.*24, *p* = 0*.*79], ruling out the possibility that the absence of CMI differences is due to

**Table 3 | Results of ANOVAs comparing contextual effects of schizophrenia, bipolar, and control groups in each task.**


*Adjusted alpha level* = *0.01, reflecting correction for 5 multiple comparisons.*

equal numbers of BD with abnormally strong and abnormally weak CMIs.

We also examined intertask correlations to test whether a weak contextual effect on one task would predict a weak contextual effect on other tasks. However, no significant correlations were found within any of the three groups (**Table 4**). It should be noted that some intertask correlations were trending toward significance at an unadjusted alpha level (0.05) but the relationship differed in each group: orientation and motion were modestly correlated in SZ (*r* = 0*.*39, *p* = 0*.*07), size and motion in BD (*r* = 0*.*5, *p* = 0*.*06), luminance and motion in CO (*r* = 0*.*4, *p* = 0*.*07). However, there was no consistent trend of intertask correlation across groups. It is worth noting that these low correlations are not caused by low measurement reliability, as split-half reliabilities were high (**Table 2**).

Finally, we examined the relationships between the strength of contextual illusions and clinical measures in patient groups (**Table 5**). Unless otherwise noted, significance was defined at an adjusted alpha level of 0.01, reflecting Bonferroni correction for five multiple comparisons per clinical measure. In BD, the contextual modulation of contrast negatively correlated with YMRS score (**Figure 4A**): greater manic symptoms were associated with a weaker surround contrast illusion (*rs* = −0*.*76, *p* = 0*.*003). When excluding three potential outliers based on YMRS score (see **Figure 4A**), the correlation remained significant at the unadjusted alpha level (*rs* = −0*.*68, *p* = 0*.*03, α = 0*.*05). There was also a trend for the severity of depressive symptoms (HRSD) to positively correlate with the magnitude of orientation repulsion illusion in BD (*rs* = 0*.*45, *p* = 0*.*08). Similarly in SZ, the strength of the orientation illusion was associated with greater positive symptoms (SAPS; *rs* = 0*.*38, *p* = 0*.*05) and negative symptoms

**FIGURE 3 | The magnitude of contextual modulation in BD and SZ.** The magnitude of contextual effect in individuals with bipolar disorder (BD) and individuals with schizophrenia (SZ) was converted into *z* scores for each task relative to the respective mean and variance of the control group. The contextual modulation index represents the average *z* score across tasks for each participant. Negative values indicate weaker

contextual modulation in patients, whereas positive values indicate stronger contextual modulation in patients relative to the control group. As noted in **Figure 2**, SZ exhibited a significantly weaker contrast illusion compared to CO. Error bars denote the standard error of the mean of the *z* scores in clinical groups, and the shaded region denotes the standard error of the mean of the control group.

(SANS; *rs* = 0*.*46, *p* = 0*.*02). However, these correlations in SZ did not survive Bonferroni correction.

When examining patient groups together, the magnitudes of both motion (**Figure 4B**) and orientation (**Figure 4C**) illusions



*Adjusted alpha level* = *0.005, reflecting correction for 10 multiple comparisons per group. r, Pearson's correlation coefficient; p, significance level.*


#### **DISCUSSION**

In this study, we examined contextual interactions in bipolar disorder to determine the diagnostic specificity of contextual abnormalities reported in schizophrenia. We (Yang et al., 2013) and others (Dakin et al., 2005; Barch et al., 2012; Tibber et al., 2013) have found that the contextual effect of contrast is weakened in SZ. Yet within the same group of schizophrenia patients, we found the magnitude of contextual modulations associated with luminance, size, orientation, and motion to be similar between SZ and CO (Yang et al., 2013). Tibber et al. (2013) reported similar findings of intact contextual luminance and orientation effects, despite weakened contextual effects of contrast



*Pearson's correlation coefficient is shown except for correlations with symptoms scores in which Spearman's correlation coefficient is displayed. p, significance level; BD, bipolar disorder participants; SZ, schizophrenia participants; BPRS, brief psychiatric rating scale; YMRS, Young Mania rating scale; HRSD, Hamilton rating scale of depression; SAPS and SANS, scale of assessment for positive and negative symptoms, respectively; DOI, duration of illness; CPZ, chlorpromazine daily equivalent; SFS, social functioning scale.*

*\*significant at the unadjusted alpha level* <sup>=</sup> *0.05.*

*\*\*significant at the adjusted alpha level* <sup>=</sup> *0.01, reflecting correction for 5 multiple comparisons per clinical/demographic variable.*

effects. Correlations remained significant when excluding three potential YMRS outliers in (A), *rs* = −0*.*68, *p* = 0*.*03, and 2 potential BPRS outliers in

**(B)** and **(C)**, *rs* = 0*.*53, *p* = 0*.*001; *rs* = 0*.*47, *p* = 0*.*002, respectively.

and size (discussed below) in SZ. Utilizing the same contextual tasks from our previous study (**Figure 1**), here we report relatively normal magnitudes of contextual effects in BD across all features tested, including contrast (**Figure 2**). With regard to overall contextual modulation strength, the three groups could not be distinguished (**Figure 3**). Yet, the strength of some contextual illusions covaried with clinical state (**Figure 4**). In BD, a weaker contrast illusion was associated with greater manic symptoms at the time of testing. We previously reported in SZ that stronger positive and negative symptoms were associated with stronger orientation and motion repulsion illusions. When examining patient groups together, stronger orientation and motion context effects were also associated with greater symptom severity assessed with BPRS. In summary, our findings suggest that the weak contextual modulation of contrast may be a general characteristic of schizophrenia, whereas contextual contrast effects may covary with manic state in bipolar disorder. In addition, the strength of other contextual effects may be modulated by clinical state, especially in schizophrenia. In the following paragraphs, we discuss the implications of our findings with BD and refer readers to our previous study (Yang et al., 2013) for more detailed discussion of visual context processing in SZ.

#### **CONTEXTUAL MODULATION OF CONTRAST**

In the first study to examine the surround contrast illusion in schizophrenia, Dakin et al. (2005) compared SZ to a healthy control group and a clinical control group that included individuals with affective, personality, and post-traumatic stress disorders. Comparing SZ with such a heterogeneous clinical group cannot probe the diagnostic specificity of the contextual deficit since impairments shared within psychotic spectrum disorders could be washed out by normal performance associated with unrelated illnesses. Thus, we specifically tested an array of contextual interactions specifically in bipolar disorder, an illness that shares many clinical features with schizophrenia (see Introduction). Our study showed that the weakened contextual modulation of contrast is indeed specific to schizophrenia, as BD showed a similar contrast illusion to that of CO (**Figures 2D**, **3**). This is reminiscent of results from a previous study reporting abnormal contrast sensitivity modulation in presence of collinear flanking stimuli in SZ but not in BD (Kéri et al., 2005).

In our study, it is important to note that BD performance on the contrast task could not be distinguished from either that of CO or of SZ. It is possible that certain clinical characteristics were modulating performance within the bipolar group and as a result, some bipolar patients behaved more similarly to schizophrenia patients than control participants. Our findings seem to support this account: Bipolar individuals with greater manic symptoms exhibited a weaker contrast illusion (**Figure 4A**). Surround contrast modulation is believed to reflect gain control mechanisms in early visual cortical areas (Chubb et al., 1989; Lotto and Purves, 2001). Perhaps, then, hyperdopaminergia related to mania leads to anomalous gain control mechanisms in BD. Indeed, it is known that dopamine mediates processes involved in contrast gain control, particularly in the modulation of visual contrast detection (e.g., Chen et al., 2003).

Given that psychotic symptoms frequently accompany manic phases, another speculation is that the presence of psychotic symptoms allows gain control mechanisms to be modulated by manic phases in BD. Simply put, manic bipolar patients who are prone to psychosis may exhibit weakened gain control mechanisms similar to those of SZ. Such an account would be consistent with early studies reporting impaired backward masking functions in actively psychotic manic patients—similar to impairments in SZ (e.g., Green et al., 1994)—and relatively normal backward masking in non-psychotic hypomanic patients (Saccuzzo and Braff, 1981, 1986). Our study did not have the statistical power to directly compare performance of bipolar patients with (*n* = 9) and without (*n* = 7) a prior history of psychosis. The severity of psychotic symptoms does not appear to modulate the contrast illusion in our pool of bipolar patients, possibly because of the limited range of symptom scores. While similar contextual contrast deficits could be taken as evidence for a shared pathophysiological mechanism between schizophrenia patients and sub-groups of bipolar patients, similar deficits could also manifest from different pathophysiological processes (e.g., Green et al., 1994; further discussion below). Future studies will provide greater insight into this debate by investigating contextual effects along the course of the illness and among different patient sub-groups.

There are at least two caveats in the interpretation of these results. Although BD did not significantly differ in performance on the contrast task from either SZ (*d* = 0*.*33) or CO (*d* = 0*.*29), the effect sizes obtained were roughly equivalent to the effect size reported by Barch et al. (2012) who demonstrated a significant group difference between SZ and CO (*d* = 0*.*31) in their contextual contrast task. Since our data was acquired from a relatively modest sample of bipolar participants, it is possible that given enough subjects the group differences between BD and SZ and between BD and CO would reach significance. Power analysis revealed that approximately 188 participants in each group of BD and CO and 146 participants in each group of BD and SZ would be required to achieve statistically significant group effects for our contextual contrast task, given the obtained effect sizes (power = 0.8, alpha = 0.05).

The second caveat was raised by Barch et al. (2012). In their study, the contextual contrast deficit in SZ was substantially weakened when they excluded individuals who performed poorly on catch trials. Barch and colleagues argued that weakened surround contrast effects in SZ might be attributed to general impairments in attention. Our finding of abnormally weak contrast modulation in SZ is unlikely due to attentional impairments, as the deficit was specific to the contrast task and was not observed in the baseline contrast condition. However, Barch et al.'s results underscore the need for further research into contextual contrast processing in schizophrenia.

#### **CONTEXTUAL MODULATION OF ORIENTATION AND MOTION**

Deficits in motion perception are well established in SZ (review by Chen, 2011) and recent evidence suggests that orientation processing may be disturbed as well (Rokem et al., 2011). Studies have identified abnormal contextual modulation of moving stimuli (Tadin et al., 2006; Chen et al., 2008) and orientation-specific surround suppression (Yoon et al., 2009), although the exact nature of these deficits is still under debate (see Yang et al., 2013). In one study, Chen et al. (2008) reported abnormally strong surround motion repulsion in mildly symptomatic SZ. Consistent with Chen et al.'s results, we previously showed that SZ with stronger motion and orientation repulsion effects also showed greater symptom severity, as assessed with BPRS (Yang et al., 2013). Here, we found that this relationship remained significant when including data from another clinical population—BD—within the psychosis spectrum of disorders (**Figures 4B–C**). The magnitudes of these repulsion illusions further predicted the severity of a range of clinical symptoms: SZ with greater positive (SAPS) or negative (SANS) symptoms exhibited stronger orientation and motion repulsion effects (Yang et al., 2013) and there was a trend for BD with more severe depressive symptoms (HRSD) to show stronger orientation illusions. Patients with greater duration of illness were more likely to have stronger repulsion effects as well. However, other studies did not find a relationship between clinical symptoms and motion or orientation illusions in SZ and in BD (Chen et al., 2008; Tibber et al., 2013). The discrepancy in results could be attributed to task and stimulus differences or the fact that clinical symptom scores were much higher in these studies (Chen et al., 2008; Tibber et al., 2013). Further investigation will be necessary to ascertain the usefulness of these particular contextual illusions for clinical studies of schizophrenia.

# **CONTEXTUAL MODULATION OF SIZE AND BRIGHTNESS**

We previously reported relatively normal Ebbinghaus illusion in SZ (Yang et al., 2013). In contrast, Uhlhaas et al. (2004, 2006) reported that both SZ and schizotypal individuals showed a reduced size illusion effect. Notably, this result was observed in only a subset of individuals with disorganization symptoms or thought disorder. Our findings are consistent with Uhlhaas et al.'s in that SZ participants in our study exhibited few, if any, symptoms of disorganization. However, Tibber et al. (2013) reported weakened size illusion among SZ who mostly exhibited few disorganized symptoms. Thus, the relationship between the size illusion and clinical symptoms in SZ is an issue that requires further inquiry.

As far as we know, ours is the first study to examine the role of surrounding context in perceived brightness in bipolar disorder. Our findings show that BD exhibit relatively intact brightness induction. Our group and Tibber et al. (2013) similarly reported normal contextual modulation of luminance in SZ. Taken together, these findings suggest that the early cortical and subcortical mechanisms responsible for the contextual effects in brightness perception (Rossi and Paradiso, 1999; Kinoshita and Komatsu, 2001) may be spared in SZ and in BD.

# **CONTEXTUAL PROCESSING: A BIOMARKER FOR SCHIZOPHRENIA?**

There has been a rapid growth in the use of context tasks in clinical trials and large-scale, NIH-supported studies of schizophrenia, particularly tasks focusing on visual context. Given this trajectory, it is imperative to identify the conditions under which contextual processing is compromised in schizophrenia and importantly, to examine the diagnostic specificity of these abnormalities. Recently, we found no clear evidence for a general weakening of contextual visual processing in SZ (Yang et al., 2013), which was later confirmed by Tibber et al. (2013). Using different sets of contextual tasks, both studies reported a weakened contextual contrast effect in SZ. The current study further suggests that the contextual contrast deficit may be specific to SZ, as it was not found in BD (see caveats above). Taken together with previous studies, these findings support the notion that, among different visual context tasks, the contrast illusion may be a more viable candidate for a biomarker of schizophrenia.

The remaining question is whether the contextual contrast deficit is a state- or trait-related characteristic of schizophrenia. The contextual contrast deficit has been reported in both inpatient and outpatient populations and has failed to correlate with any clinical measure at the time of testing (Dakin et al., 2005; Barch et al., 2012; Tibber et al., 2013; Yang et al., 2013). This suggests that the contrast deficit is not influenced by clinical state. Comparison of effect sizes across studies, however, show that the largest effect size was reported in a study of inpatients (Dakin et al., 2005), whereas the smallest effect size was found in a study of outpatients (Barch et al., 2012). Intermediate effect sizes were reported in a smaller cohort of outpatients (Yang et al., 2013) and in a mixture of inpatients and outpatients (Tibber et al., 2013). On this basis, one could speculate that the contrast deficit is indeed modulated by clinical state, as inpatients tend to be actively and severely ill in comparison to clinically stable outpatients. Yet other factors may have contributed to the differences in effect sizes across studies including task differences, medication effects, and sample size (smaller samples tend to enhance effect sizes). Thus, it may be too early to draw conclusions about the role of clinical state in the contextual contrast deficit. This issue would be best addressed with studies examining the contrast illusion along the course of the illness and across a wide range of schizophrenia patients varying in symptom severity.

#### **DIMENSIONAL vs. CATEGORICAL CLASSIFICATION OF PSYCHOSIS SPECTRUM DISORDERS**

Converging lines of evidence implicate commonalities between schizophrenia and bipolar disorder, including overlaps in genetic susceptibility, in epidemiologic characteristics, and in neural dysfunction [reviews by Möller (2003) and Maier et al. (2006)]. These findings have revived a long-standing debate as to the relationship between schizophrenia and other psychotic disorders, including bipolar disorder. The traditional dichotomy in the diagnosis of schizophrenia and bipolar disorder has long been challenged by the notion that schizophrenia is not a singular, distinct entity but, instead, forms part of a psychosis continuum (McIntyre, 1949; Craddock and Owen, 2005). However, not all evidence supports a continuous account of psychosis (David, 2010). Abnormalities in neurodevelopment and cognitive function follow distinctly different time courses in the two disorders (Lewandowski et al., 2011). Non-shared genetic risk factors (e.g., Grozeva et al., 2010) and neurobiological distinctions (e.g., structural and functional differences in the brain) also exist between schizophrenia and bipolar disorder (Geuze et al., 2005). Moreover, several empirical findings across multiple domains differentiate SZ from BD. A particularly important example is that of oculomotor control; smooth pursuit eye tracking deficit is a candidate endophenotypic marker for schizophrenia but is intact in bipolar disorder (e.g., Levy et al., 1993; Holzman, 2000; Levy and Sweeney, 2008). Similarly, higher cognitive deficits are severe in schizophrenia but mild or absent in bipolar disorder (Krabbendam et al., 2005) especially with respect to working memory (Park and Holzman, 1992; Pirkola et al., 2005).

Therefore, a singular pathophysiological mechanism is unlikely to account for the two psychotic disorders (Whalley et al., 2012). Similar to the current state of the literature, our findings neither fit perfectly into the continuous or categorical account of psychosis but suggest an alternative approach. The two disorders may be differentiated by their distinct *profiles* of impaired, intact and even enhanced functions. Identifying such profiles across different tasks within a neurobiologically constrained domain may prove to be extremely useful in elucidating the nature of these disorders. Thus, our results emphasize the need for a hybrid model that better captures the complexity in symptoms, deficits, and prognosis within and across diagnostic categories.

Recent initiatives, such as the Research Domain Criteria (RDoC) project, were developed for this purpose (Morris and Cuthbert, 2012). RDoC supports a multi-dimensional approach framed within neuroscience and genomic research to identify core processes underlying clinical features and diagnostic groups. It is too soon to tell whether contextual processing abnormalities contribute to one of those core processes. Future studies should include the investigation of epidemiological characteristics (e.g., risk factors, heritability) to elucidate the role of contextual dysfunction in schizophrenia and psychosis spectrum disorders.

# **SUMMARY**

Our study systematically measured contextual processing in bipolar disorder and compared those results to equivalent measurements in schizophrenia, to determine the extent to which abnormal contextual interaactions are characteristic of psychosis spectrum disorders in general. We measured contextual effects across a range of visual tasks in individuals with bipolar disorder and compared their performance with that of our previously published findings with schizophrenia and healthy participants tested on those same tasks. Participants with bipolar disorder showed robust contextual effects that were comparable in magnitude to those reported in healthy participants. The contextual contrast illusion, in particular, distinguished performances of bipolar disorder and schizophrenia groups, as individuals with schizophrenia exhibited weakened contrast illusion relative to controls whereas individuals with bipolar disorder did not. Yet, bipolar patients with worse manic symptoms were more likely to have a weaker contrast illusion. Furthermore, the severity of psychiatric symptoms was associated with stronger orientation and motion repulsion illusions, especially in individuals with schizophrenia. These findings may suggest that the pathophysiological mechanisms underlying contextual effects may differ in bipolar disorder compared with schizophrenia.

# **ACKNOWLEDGMENTS**

The authors thank Amanda Cumming, Katherine Thakkar, Natasha Matthews, Heath Nichols, and Joel Peterman for their assistance in subject recruitment and in clinical assessments. This work was supported by grants from National Alliance for Research on Schizophrenia and Depression, the National Institute of Health (MH073028, EY007135, EY001319, EY007125, and EY019295), P30 HD15052 to the Vanderbilt Kennedy Center for Research on Human Development and the World Class University Program through the Korea Science and Engineering Foundation funded by the Ministry of Education, Science and Technology (R31-10089).

# **REFERENCES**


*Biol. Psychiatry* 64, 74–77. doi: 10.1016/j.biopsych.2007.11.017


study. *Prog. Neuropsychopharmacol. Biol. Psychiatry* 31, 86–91. doi: 10.1016/j.pnpbp.2006.07.001


dimensions of behavior. *Dialogues Clin. Neurosci.* 14, 29–37.


Evidence for impaired visual context processing in schizotypy with thought disorder. *Schizophr. Res.* 68, 249–260. doi: 10.1016/S0920-9964 (03)00184-1


magnetic resonance imaging studies comparing bipolar disorder and schizophrenia. *Bipolar Disord.* 14, 411–431. doi: 10.1111/j.1399-5618. 2012.01016.x


Diminished orientation-specific surround suppression of visual processing in schizophrenia. *Schizophr. Bull.* 35, 1078–1084. doi:10.1093/schbul/sbp064

Young, R. C., Biggs, J. T., Ziegler, V. E., and Meyer, D. A. (2000). "Young Mania Rating Scale," in *Handbook of Psychiatric Measures*, (Washington, DC: American Psychiatric Association), 540–542.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 05 April 2013; accepted: 09 August 2013; published online: 30 August 2013.*

*Citation: Yang E, Tadin D, Glasser DM, Wook Hong S, Blake R and Park S (2013) Visual context processing in bipolar disorder: a comparison with schizophrenia. Front. Psychol. 4:569. doi: 10.3389/fpsyg. 2013.00569*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Yang, Tadin, Glasser, Wook Hong, Blake and Park. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Effects of short-term inpatient treatment on sensitivity to a size contrast illusion in first-episode psychosis and multiple-episode schizophrenia

#### *Steven M. Silverstein1 \*, Brian P. Keane1,2, Yushi Wang1, Deepthi Mikkilineni 1, Danielle Paterno1, Thomas V. Papathomas <sup>2</sup> and Keith Feigenson1*

*<sup>1</sup> Rutgers University Behavioral Health Care, and Department of Psychiatry, Robert Wood Johnson Medical School, Rutgers Biomedical and Health Sciences, Piscataway, NJ, USA*

*<sup>2</sup> Center for Cognitive Science, Rutgers University, New Brunswick, NJ, USA*

#### *Edited by:*

*Michael Green, University of California, Los Angeles, USA*

#### *Reviewed by:*

*Carol Jahshan, VA Greater Los Angeles Healthcare System, USA Amanda McCleery, UCLA Dept. Psychiatry, USA*

#### *\*Correspondence:*

*Steven M. Silverstein, Rutgers University Behavioral Health Care, and Department of Psychiatry, Robert Wood Johnson Medical School, Rutgers Biomedical and Health Sciences, 151 Centennial Avenue, 08854 Piscataway, NJ, USA e-mail: silvers1@umdnj.edu*

**Introduction:** In the Ebbinghaus illusion, a shape appears larger than its actual size when surrounded by small shapes and smaller than its actual size when surrounded by large shapes. Resistance to this visual illusion has been previously reported in schizophrenia, and linked to disorganized symptoms and poorer prognosis in cross-sectional studies. It is unclear, however, when in the course of illness this resistance first emerges or how it varies longitudinally with illness phase.

**Method:** We addressed these issues by having first-episode psychosis patients, multipleepisode schizophrenia patients and healthy controls complete a psychophysical task at two different time points, corresponding to hospital admission and discharge for patients. The task required judging the relative size of two circular targets centered on either side of the screen. Targets were presented without context (baseline), or were surrounded by shapes that made the size judgment harder or easier (misleading and helpful contexts, respectively). Context sensitivity was operationalized as the amount of improvement relative to baseline in the helpful condition minus the amount of decrement relative to baseline in the misleading condition.

**Results:** At hospital admission, context sensitivity was lower in the multiple-episode group than in the other groups, and was marginally less in the first episode than in the control group. In addition, schizophrenia patients were significantly more and less accurate than the other groups in the misleading and helpful conditions, respectively. At discharge, all groups exhibited similar context sensitivity. In general, poorer context sensitivity was related to higher levels of disorganized symptoms, and lower level of depression, excitement, and positive symptoms.

**Discussion:** Resistance to the Ebbinghaus illusion, as a characteristic of the acute phase of illness in schizophrenia, increases in magnitude after the first episode of psychosis. This suggests that visual context processing is a state-marker in schizophrenia and a biomarker of relapse and recovery.

**Keywords: schizophrenia, vision, perception, cognition, state marker, biomarker, context, disorganization symptoms**

## **INTRODUCTION**

Recent years have seen a renewed interest in visual perception in schizophrenia. Reasons for this include: (1) vision is arguably the best understood domain of mental functioning (Palmer, 1999); (2) reliable and valid measures from the field of vision science are available to assist with answering specific questions about brain and cognitive functioning in schizophrenia (Butler et al., 2008, 2012); (3) studies have consistently demonstrated specific perceptual differences between people with schizophrenia and matched controls (Butler et al., 2008; Chen, 2011; Green et al., 2011; Silverstein and Keane, 2011a), and this can be done independently of a generalized deficit in many cases (Place and Gilmore, 1980; Knight, 1984; Knight and Silverstein, 2001; Dakin et al., 2005; Koethe et al., 2009; Yoon et al., 2009; Tibber et al., 2013); (4) patterns of abnormal regional activation, connectivity/circuitry, and/or neurotransmitter activity have been associated with visual impairments in schizophrenia, and these are consistent with what is known about normal vision from the neurobiology literature (Spencer et al., 2004; Silverstein et al., 2009; Sehatpour et al., 2010; Uhlhaas and Singer, 2010; Yoon et al., 2010; Butler et al., 2013; Plomp et al., 2013); (5) theoretical models and empirical data link visual impairments with aspects of behavioral and cognitive functioning, in some cases suggesting that perceptual impairments are low-level manifestations of widespread canonical computations that are impaired in the disorder (Phillips and Silverstein, 2003, 2013; Silverstein and Keane, 2011a,b); (6) visual impairments are related to problems in daily functioning in schizophrenia (Green et al., 2012); and (7) some visual abnormalities in schizophrenia are related to clinical state (Silverstein et al., 1996; Uhlhaas et al., 2005; Keane et al., 2013), suggesting they may be biomarkers of relapse, recovery, or treatment response, whereas other abnormalities are stable over time and can be found in unaffected relatives, suggesting they may be genetic or endophenotype markers (Yeap et al., 2006).

The primary goal of this study was to examine whether scores on an index of visual context processing covary with clinical state over the course of short-term inpatient treatment. To do this, patients were tested at admission and discharge/transfer from an acute care inpatient unit. The visual task involved a variant of the Ebbinghaus illusion in which a circle appears larger than its actual size when surrounded by smaller circles, and smaller than its actual size when surrounded by larger circles (see **Figures 1**, **2**). On each experimental trial, subjects were shown two target circles—one on the left of the screen and one on the right—and the task was to decide which was larger. On half of the trials, the targets were surrounded by larger or smaller circles that would make giving a correct response easier (helpful condition) or harder (misleading condition; see **Figure 1**). As discussed further below, we chose this illusion because it has been established over decades of research, because it is experienced to a lesser extent among schizophrenia patients relative to healthy and psychiatric controls (Uhlhaas et al., 2006a,b; Tibber et al., 2013), and because reduced illusions have been linked to a more acutely ill clinical state and to more disorganized symptoms in crosssectional studies (Uhlhaas et al., 2006a,b; Horton and Silverstein, 2011). Another advantage to this task is that it can side-step generalized deficit confounds, in which low accuracy can be attributed to reduced motivation or attention (Knight and Silverstein, 2001; Silverstein, 2008). In the Ebbinghaus task, patients are expected to perform *better* than healthy controls in the misleading context condition, worse than controls in the helpful context condition, and about the same in the no-context condition.

The Ebbinghaus illusion was discovered by German psychologist Hermann Ebbinghaus (b. 1850, d. 1909), was popularized by Edward Titchener's 1902 psychology textbook (Titchener, 1902), and has been the subject of numerous experiments since the 1970s (e.g., Massaro and Anderson, 1971; Girgus et al., 1972; Weintraub and Schneck, 1986; Coren and Enns, 1993; Rose and Bressan, 2002; Doherty et al., 2010; Schwarzkopf and Rees, 2013). The illusion depends on basic stimulus parameters such as the relative sizes of target and surrounds, distances between targets and surrounds, and differences in form between targets and surrounds (Massaro and Anderson, 1971; Choplin and Medin, 1999). It can also be affected by the conceptual similarity between target objects and their surrounds (Coren and Enns, 1993) or by affective cues (Van Ulzen et al., 2008). However, in the absence of such high-level manipulations, the illusion is thought to primarily reflect: (1) perceptual organization, since the illusion requires integration of targets and surrounds (Kovacs, 2000); and (2) size constancy, which involves "top-down" effects of prior knowledge of depth cues on the representation of sensory input (Phillips et al., 2004; Doherty et al., 2008, 2010; Caparos et al., 2012). Regarding the latter, when the surrounding circles are larger than

the center target circle, this creates the implicit assumption that the stimulus set is relatively close to the observer, and the center object is therefore perceived as smaller than its actual size. In contrast, smaller surrounds lead to the implicit assumption that the stimulus set is relatively far from the observer, and the center object is then perceived as larger than its actual size (see **Figure 2**) (Doherty et al., 2010). These effects are consistent with the tendency, in adults, to overestimate the size of distant objects and to underestimate the size of near objects (Kavsek and Granrud, 2012). The Ebbinghaus illusion can also be considered a form of surround suppression, in the sense that perception of a central target is modulated by surrounding context in a direction opposite to characteristics of the surround (Tibber et al., 2013).

A second goal of this study was to compare the performance of people with a first episode of psychosis to that of healthy controls and people with an established diagnosis of schizophrenia. To date, there have been no studies of size contrast illusions—or any other type of surround suppression—in first-episode psychosis. It is therefore unclear whether reduced

**FIGURE 2 | (A)** Most people see the further circle as being larger than the nearer one, though they are equal. They would also judge the 'real' size of the further circle within the pictured space to be much larger than the nearer circle. This shows that pictorial cues to depth and size influence perception of the markings on the picture surface. **(B)** Adding surrounds, as in the Ebbinghaus illusion, increases the perceived size difference between the two circles. This suggests that surround size adds to the other pictorial depth cues. **(C)** In texture gradients the mean size and separation of elements decreases with depth. The size of the elements on the picture surface is seen as decreasing with depth, but their 'real' size within the pictured space would be judged to be approximately constant. The large element in the center of the second row from the top may be seen as being larger than that arrowed below, but they are equal. Its 'real' size within the pictured space would be judged to be much larger. The bottom and top three rows are versions of the Ebbinghaus illusion. Therefore, this suggests that the illusion may in part be due to the visual system learning to use such pictorial cues. Figure reprinted, with permission by John Wiley and Sons, from Doherty et al. (2010).

illusion effects, and the abnormalities in neural mechanisms that subserve these reductions, are an aspect of psychosis in general, schizophrenia in general, or psychosis and/or illness progression in schizophrenia. However, evidence suggests that perceptual organization impairments are associated with illness chronicity and progression. For example, reduced perceptual organization has been observed among patients requiring long-term hospitalization compared to patients requiring short-stays who can usually function in the community, with the latter group performing normally (Silverstein et al., 1998, 2006b; Uhlhaas et al., 2006b; Silverstein and Keane, 2011a). At the same time, past studies in prodromal or first-episode schizophrenia indicate that perceptual organization is intact at those time points (Parnas et al., 2001; Silverstein et al., 2006a). Because the evidence so far is limited to two studies (both cross-sectional) and because no study has explicitly examined surround suppression (let alone size contrast illusions) in first episode patients, however, it remains an open question as to how persons with first episode psychosis perform relative to our other groups on the type of task we report on here.

#### **METHOD**

#### **SUBJECTS**

Three subject groups participated: (1) patients hospitalized for their first episode of psychosis (FEP) (*n* = 16, 9 males), and so, for whom, the eventual diagnosis (e.g., mood disorder with psychotic features vs. schizophrenia spectrum disorder) is unknown at this time point; (2) patients in their second or later episode of schizophrenia (SCZ) (*n* = 21, 16 males) recruited from the same short-term inpatient unit as the first episode subjects; and (3) healthy controls (CON) screened to rule out the presence of a psychotic or mood disorder (*n* = 27, 14 males). Demographic characteristics and symptom profiles of each group can be found in **Table 1**. For the FEP group, research diagnoses (and *n*) on admission were as follows: psychotic disorder NOS (8), schizophrenia (1), schizoaffective disorder (1), delusional disorder (1), and major depression with psychotic features (5). Average length of stay on the inpatient unit for patients was 16.46 days (*SD* = 8*.*10, median = 15.00). On average, patients were initially tested 5.28 days after hospital admission (*SD* = 4*.*54, median = 4.00), and then again 13.60 days later (*SD* = 6*.*64, median = 13.00). To be included in the study, patients had to be between the ages of 18–60, and had to be diagnosed with either schizophrenia, or a first episode of a psychiatric disorder with psychotic symptoms. Exclusion criteria included: (1) any history of TBI or head injury with loss of consciousness greater than 10 min; (2) history of a neurological or developmental disorder; (3) current mood disorder; (4) current substance abuse or dependence disorder (within past 6 months) or positive urine toxicology screen on any day of testing; (5) estimated premorbid (Wechsler) IQ *<* 70, as determined by the Shipley Institute of Living Scale (Shipley et al., 2009) or evidence of intellectual disability as indicated in the electronic medical record; or (6) ECT within the past 8 weeks. All patients were receiving antipsychotic medication. Exclusion criteria for the CON group included those listed for patients, as well as: (1) any lifetime Axis-I disorder (as assessed by SCID) with the exception of past substance use disorders; (2) psychotropic medication use in the last 6 months; and (3) a first-degree relative(s) with a diagnosis of schizophrenia, schizoaffective disorder, or bipolar disorder (based on subject self report). All subjects had normal or corrected-to-normal visual acuity as assessed via a Snellen chart.

#### *Apparatus*

Stimuli were presented on a Samsung 2243BWX LCD monitor with viewable dimensions of 47.5 by 29.8 cm. The viewing distance was 24 inches (60.9 cm). The screen resolution was 1680 × 1050, and therefore, the viewable screen subtended 43◦ × 27◦ of visual angle. Spyder 3 Elite software was used to calibrate the monitors across sites at the start of the study and then monthly afterwards. Monitor parameters were a gamma value of 2.2, color temperature (white point) of 6500K, and luminance of 120 cd/m2.

#### *Ebbinghaus Illusion Task*

Stimuli were presented and responses were recorded and analyzed with a C++ program developed by Phillips et al. (2004). This task has been used in four prior studies of the Ebbinghaus illusion, including one of schizophrenia (Phillips et al., 2004; Doherty et al., 2008, 2010; Horton and Silverstein, 2011). On each trial, the task was to press a key to indicate whether the target on the left or



*<sup>a</sup> Factor score based on a 4-item scale (each item rated 1–7).*

*<sup>b</sup> Factor score based on a 6-item scale (each item rated 1–7).*

*<sup>c</sup> Facor score based on a 5-item scale (each item rated 1–7).*

the right half of the screen was larger (see **Figure 1**). All circles were black and presented on a white background. The stimulus appeared on the screen until the subject responded or after 2 s (whichever happened first). If a response was not recorded within 2 s of stimulus onset, the trial was counted as a guess (0.5 correct). Trials were separated by 200 ms. The targets were centered on either side of the screen and appeared either with or without surrounding circles (see below). The two target circles always differed in actual size, and this difference varied in magnitude across trials. The center circle on one side was always 2.67◦ of visual angle in diameter, while the center circle on the other side was always 0.05◦, 0.16, 0.27, 0.37, or 0.48◦ larger or smaller. The side on which the larger circle appeared was randomized. This size comparison was presented in 3 conditions. (1) In the *misleading* condition, the target circles were always surrounded by 8 larger circles arranged in a square configuration (i.e., 3 above, one on each side, and 3 below, see **Figure 1**). Each of the five size differences was shown sixteen times, with the larger central circle always surrounded by larger circles (3.33◦ in diameter) and the smaller central circle always surrounded by smaller circles (1.33◦ in diameter). In this condition, size contrast impairs discrimination by biasing the observer to perceive the larger target as smaller and the smaller target as larger (Doherty et al., 2008). (2) In the *helpful* context condition, the 2.61 and 2.72◦ target circles were presented eight times each, again surrounded by 8 circles around the edges of an imaginary square, with the smaller center circle surrounded by circles 3.33◦ in diameter and the larger central circle surrounded by circles 1.33◦ in diameter. In this condition, size contrast increases accuracy. Note that in this condition, if subjects choose the array with larger *surrounds* then they will be wrong on every trial. As in prior studies, only 16 trials were presented in the helpful condition, and these were all at the hardest difficulty level (0.05◦ size difference between center circles) (Phillips et al., 2004; Doherty et al., 2008). The 96 trials in the context conditions (80 in the misleading and 16 in the helpful conditions) were presented in a different random order for each subject. (3) In addition to these 96 trials, 96 additional trials were presented in a control (*no-context*) condition, also in a different random order for each subject, using the same 80 size comparisons as in the misleading condition, plus 16 additional trials at the smallest size difference. In other words, the no-context trials were exactly the same as the block of trials with context, except that the surrounding circles were invisible. This block of trials was presented either before or after the trials containing context, with the order of context and no context blocks counterbalanced across subjects. In total, the task contained 192 trials, and typically took less than 10 min to complete.

During the course of the study, it was discovered that the location on the screen of the entire stimulus display would be shifted slightly to the left or right, corresponding to the side that contained the larger target circle. This occurred 83% of the time within the context block (only on misleading trials) and occurred the same percentage of time within the nocontext blocks. This heretofore unknown feature of the program is not deemed problematic for our analyses because it occurred with an equal incidence in the helpful and no-context trials (used for the facilitation calculation) and also in the misleading and no-context trials (used for the impairment calculation; see below). Therefore, the facilitation and impairment calculations were not biased by any aspect of the display presentation.

#### **CLINICAL ASSESSMENT MEASURES**

All patients were interviewed with the Structured Clinical Interview for DSM-IV Diagnosis (SCID), patient version (First et al., 2002b). Information was also obtained from medical records and discussions with staff to confirm the final research diagnosis. The CON group was screened for psychopathology using the non-patient version of the SCID (First et al., 2002a). For patients, symptoms were rated, based on the past week, using the Positive and Negative Syndrome Scale (Kay et al., 1987), which was scored using a 5-factor model (Lindenmayer et al., 1994a,b, 1995a,b) that includes positive, negative, cognitive/disorganized, excitement, and depression factors. All interviews were conducted by trained research staff that had established inter-rater reliability on these measures (i.e., intraclass correlations greater than 0.80) in previous studies.

#### **ANALYSIS**

All analyses were performed in SPSS version 20. Data were analyzed first by recoding timed out trials as 0.5 correct (so that subjects who preferred to guess rather than time out on a trial would not have an advantage). Next, contextual *facilitation* was calculated as the proportion correct in the helpful condition minus that in the no-context condition, using only the 0.05◦ size difference difficulty trials (since the helpful condition included only this difference magnitude). Then, the amount of contextual *impairment* was calculated as the proportion correct in the misleading condition minus the proportion correct in the no context condition (all difficulty levels). C*ontext sensitivity* was the critical metric and corresponded to the difference between facilitation and impairment (with higher difference scores indicating greater sensitivity). Planned *t-*tests determined differences between pairs of groups on the context sensitivity variables. Context sensitivity was examined at each time point for each group and this was compared between groups. The groups were also compared across time points with a 3 (group) × 2 (context) × 2 (time) analysis of variance. Because the levels of the context factor were facilitation and impairment, a main effect of context is equivalent to significant context sensitivity.

## **RESULTS**

#### **DEMOGRAPHIC DATA**

Means and standard deviations for demographic variables can be found in **Table 1**. The groups did not differ in gender composition: chi square (2) = 2.69, *p >* 0*.*26. As expected, there was a significant difference in age [*F(*2*,* <sup>68</sup>*)* = 12*.*84, *p <* 0*.*001], with the FEP group being younger than either the SCZ or CON groups (Scheffe *p*s *<* 0.001), who did not differ in age from each other (*p >* 0*.*97). Also as expected, the groups differed on education level [*F(*2*,* <sup>68</sup>*)* = 4*.*30, *p <* 0*.*05], with the CON group having more years of education than the SCZ (*p <* 0*.*05) but not FEP (*p >* 0*.*58) group, and the two patient groups not differing from each other (*p >* 0*.*29). There were no group differences for mother's education level or father's education level (*p*s *>* 0.70).

#### **TIME 1**

At initial testing (which for the patient groups, represented hospital admission), the CON group performed as expected (see **Figure 3**). The critical context sensitivity score (facilitation minus impairment) was enormous [*F(*1*,* <sup>26</sup>*)* = 397*.*46, *p <* 0.001, partial eta squared = 0.939], with strong facilitation [30.0%; *t(*26*)* = 11*.*17, *p <* 0*.*001] and impairment [-48.3%; *t(*26*)* = 18*.*16, *p <* 0*.*001]. Indeed, all 21 controls exhibited both facilitation and impairment. For the FEP group, there was also significant context sensitivity [*F(*1*,* <sup>15</sup>*)* = 65*.*01, *p <* 0*.*001, partial eta squared = 0.813], with significant contextual facilitation [23.4%; *t(*15*)* = 5*.*18, *p <* 0*.*001] and impairment [−39*.*5%; *t(*15*)* = 7*.*72, *p <* 0*.*001]. The SCZ group, however, exhibited no context sensitivity [*F(*1*,* <sup>20</sup>*)* = 1*.*14, *p* = 0*.*30), partial eta squared = 0.053], demonstrating a non-significant negative amount of facilitation [−4*.*2%; *t(*20*)* = −0*.*53, *p* = 0*.*60] and some degree of impairment, although less than half as much as observed in the other groups [−17*.*5%; *t(*20*)* = −2*.*94, *p* = 0*.*008]. Given the negative direction of the facilitation for the SCZ group, it is possible that some of the impairment arose simply because the surrounds had a general disruptive effect on performance regardless of the illusion, perhaps by making the target harder to isolate.

Context sensitivity was next directly compared between groups. The main effect of group on the context sensitivity metric was significant: *F(*2*,* <sup>61</sup>*)* = 17*.*03, *p <* 0*.*001, partial eta squared = 0.358. Planned comparisons indicated that the SCZ group was less context sensitive than the FEP [*t(*32*.*1*)* = 3*.*32, *p* = 0*.*002] and CON [*t(*23*.*6*)* = 4*.*90, *p <* 0*.*001] groups, whereas the FEP and CON groups differed only marginally from each other [*t(*41*)* = 1*.*95, *p* = 0*.*059]. Of note, and as expected, the SCZ group was significantly *more* accurate than the FEP and CON groups in the misleading condition (69% vs. 52% vs. 46%; *p*s *<* 0.01 and 0.001, respectively), while being less accurate than these

**FIGURE 3 | Context sensitivity at Time 1 (hospital admission for the SCZ and FEP groups), by group.** Performance in the facilitation and impairment context conditions is expressed relative to performance in the no-context condition, which is represented by the value 0.0 on the *Y* axis. FEP, First Episode Psychosis; SCZ, Schizophrenia (multiple episode); CON, Healthy Control.

groups in the helpful condition (57% vs. 87% vs. 95%; *p*s *<* 0.005 and 0.001, respectively; see **Table 2**).

Next, correlations between context sensitivity and PANSS symptom factors were examined for each patient group separately. Despite the modest samples sizes, an interesting pattern of findings was revealed. Positive, depression and excitement symptoms positively correlated with context sensitivity in the SCZ group (*rho* = 0*.*45, *p <* 0*.*05; *rho* = 0*.*62, *p <* 0*.*005, and *rho* = 0*.*49, *p <* 0*.*05, respectively). In addition there were trends toward significant correlations with negative and cognitive/disorganized symptoms (*rho* = 0*.*40, *p <* 0*.*08; *rho* = −0*.*37, *p <* 0*.*10, respectively). Keeping in mind that the results are not corrected for multiple comparisons, these data suggest that poorer context sensitivity was associated with lower levels of positive, depression, excitement, and negative symptoms, and higher levels of disorganized symptoms. Similar to a prior study (Uhlhaas et al., 2006a), we dichotomized the conceptual disorganization score (P2 on the PANSS) so that SCZ subjects who had moderate or severe disorganization (*>*3; *n* = 7) were compared with those who had lower scores (*n* = 14). Replicating the past effect, we found that disorganized SCZ patients had significantly less context sensitivity [*t(*19*)* = 2*.*24, *p <* 0*.*05]. This analysis could not be performed for the FEP group as only 2 FEP patients met criteria for the disorganized group. There were no significant symptom correlates for the FEP group (all *p*s *>* 0.55). These results are especially interesting because the symptom profiles of the FEP and SCZ groups did not differ on positive, negative, excitement, cognitive, or depression symptoms (*p*s *>* 0.12). Therefore, symptoms *per se* do not yield lower context sensitivity; *symptoms along with a more advanced illness do*.

It must be noted that the effect for positive symptoms was not expected. The more positive symptoms that a SCZ patient divulged during the PANSS interview, the more normal their perceptual performance (and hence more context sensitivity). This paradoxical inverse relationship in surround suppression tasks is not unprecedented (Yang et al., 2013) and will be discussed further below. Importantly, however, this symptom effect was driven primarily by depression. When both depression and positive symptoms were entered together as predictors of the context sensitivity index for SCZ patients in a multiple regression analysis (for which *<sup>R</sup>*<sup>2</sup> <sup>=</sup> <sup>0</sup>*.*44), depression (*Beta* <sup>=</sup> <sup>0</sup>*.*56, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*01) but not positive symptoms (*Beta* = 0*.*19, *p* = 0*.*36) was a significant predictor.

Finally, for the patient sample as a whole, poorer visual context sensitivity was significantly related to fewer depression (*rho* = 0*.*40, *p* = 0*.*01) and excitement (*rho* = 0*.*36, *p* = 0*.*03) symptoms, and—as expected—greater cognitive/disorganized symptoms (*rho* = −0*.*39, *p* = 0*.*019). When the patient sample as a whole was divided into those with and without conceptual disorganization, the group difference was significant: *t(*34*)* = 2*.*31, *p <* 0*.*05.

#### **TIME 2**

At the second testing point (which, for the patient groups, represented hospital discharge or transfer), the sample sizes for each group were FE*P* = 14, SCZ = 14, and CON = 25. The smaller sample sizes for each group at discharge compared to admission reflected sudden discharges or transfers of patients that occurred before discharge testing could take place, or—in the case of 2 control subjects—unwillingness to return for a second session. There was a trend toward a group difference in subject attrition rates from Time 1 to Time 2: chi squared (2) = 5.91, *p* = 0*.*052. Importantly, in the SCZ group, which had the largest number of subjects unavailable to the study at Time 2, there were significant Time 1 differences between those whom could be tested at Time 2 (*n* = 14) versus those whom were unavailable (*n* = 7). The SCZ subgroup that was unavailable at Time 2 demonstrated, at Time 1, less context sensitivity [*F(*1*,* <sup>19</sup>*)* = 5*.*59, *p* = 0*.*029, partial eta squared = 0.237] than the SCZ subgroup that attended both sessions. Therefore, the between group data from Time 2 reported below are biased in a *conservative* direction by the loss of those patients with the least context sensitivity at Time 1, in the sense of our data potentially underestimating the magnitude of change in the SCZ group from Time 1 to Time 2. This is because poorer context sensitivity at Time 1 was (non-significantly) associated with *greater* improvement in context sensitivity from Time 1 to Time 2 for the SCZ group (*rho* = −0*.*35, *p <* 0*.*23). Moreover, although both the FEP and CON groups both demonstrated significant context sensitivity at Time 1, for both of these groups, poorer context sensitivity at Time 1 was also associated with greater improvement in context sensitivity across time points, with correlation values very similar to those observed in the SCZ group: FEP *rho* = −0*.*35, *p* = 0*.*23; CON *rho* = −0*.*36, *p* = 0*.*08. For both patient groups combined *rho* = −0*.*36, *p <* 0*.*07 and for the sample as a whole *rho* = −0*.*37, *p* = 0*.*007. Although only the correlation for the sample as a whole reached statistical significance, the values are remarkably similar across groups, and are all in the negative direction and indicative of a medium effect size (Cohen, 1992). All of this suggests that it is *un*likely that patients with the weakest context sensitivity at Time 1 would have had the

**Table 2 | Percent correct in each condition in the Ebbinghaus illusion task, by group, at each time point.**


smallest amount of improvement across Times 1 and 2, and thus that it is unlikely that the available data on change in performance over time for the SCZ group overestimate the true effect. It is worth noting as well that if the seven unavailable (at Time 2) SCZ patients, and the two unavailable subjects in each of the other two groups, were removed from the Time 1 analyses reported above, the context sensitivity, facilitation, and impairment differences between the SCZ and other groups would all remain significant (all *p*s *<* 0.005).

At discharge, as before, the CON group was strongly context sensitive [*F(*1*,* <sup>24</sup>*)* = 66*.*08, *p <* 0*.*001], partial eta squared = 0.734 (see **Figure 4**), with significant facilitation [22.13%; *t(*24*)* = 3*.*95, *p* = 0*.*001] and impairment [−42*.*13%; *t(*24*)* = −13*.*72, *p <* 0*.*001]. A similar, but somewhat weaker, pattern of performance was observed among FEP patients, who displayed context sensitivity [*F(*1*,* <sup>13</sup>*)* = 23*.*23, *p <* 0*.*001, partial eta squared = 0.641], facilitation [18.5%; *t(*13*)* = 2*.*78, *p* = 0*.*016], and impairment (−36*.*0%; *t(*13*)* = 5*.*37, *p <* 0*.*001). The SCZ group showed an entirely different pattern of behavior from hospital admission and a more similar pattern to those of the other groups, with significant context sensitivity [*F(*1*,* <sup>13</sup>*)* = 8*.*29, *p* = 0*.*013, partial eta squared = 0.389], significant impairment [−33*.*2%; *t(*13*)* = −5*.*4, *p <* 0*.*001], and a non-significant, but positive, degree of facilitation [7.1%, *t(*13*)* = 0*.*812, *p* = 0*.*432] in contrast to Time 1, when it was negative.

When contextual sensitivity was compared between groups, the overall effect of group was no longer significant (see **Figure 4**): *F(*2*,* <sup>50</sup>*)* = 0*.*513, *p* = 0*.*60, partial eta squared = 0.02), and none of the pairwise between-group comparisons approached significance (all *p*s *>* 0*.*12). There were no statistically significant

SCZ, Schizophrenia (multiple episode); CON, Healthy Control.

correlations at Time 2 between PANSS symptom factor scores and context sensitivity for either the SCZ (all *p*s *>* 0.28) or FEP (all *p*s *>* 0.54) groups. Finally, for the patient group as a whole, poorer context sensitivity was marginally related to a higher level of cognitive/disorganized symptoms: *rho* = −0*.*32, *p* = 0*.*10, but there were no other symptom correlates (*p*s *>* 0*.*25). The comparison between disorganized and non-disorganized patients conducted at Time 1 was not conducted at Time 2 because only 3 patients had PANSS Conceptual Disorganization item scores greater than 3.

#### **COMPARISON BETWEEN TIME 1 AND TIME 2**

We next conducted a 2 (context) × 2 (time point) × 3 (group) ANOVA, to examine whether group differences in context sensitivity became smaller over time. There was a main effect of context [*F(*1*,* <sup>50</sup>*)* = 124*.*62, *p <* 0*.*001, partial eta squared = 0.714], and a significant group × context interaction [F*(*2*,* <sup>50</sup>*)* = 4*.*38, *p* = 0*.*018, partial eta squared = 0.149] but no other effects (*p*s *>* 0*.*17), including the group × time × context interaction (*p* = 0*.*19). The significant two-way interaction of group × context is interesting because it shows that even with the loss of context-insensitive SCZ patients who were present at Time 1, and even when the data are collapsed across the two time points, the group difference on context sensitivity remains. To probe more sensitively for longitudinal group differences, we compared only the two extreme groups, CON and SCZ. Here, the group × context × time interaction depended marginally on the time point [F*(*1*,* <sup>37</sup>*)* = 2*.*90, *p* = 0*.*097, partial eta squared = 0.073], indicating that—across time points—the schizophrenia group became more like controls.

Next, we considered changes in symptoms and how those changes related to context sensitivity. We first note that there was a significant drop in all 5 PANSS factor scores from admission to discharge (all *p*s *<* 0.05) for both patient groups, with the exception of Excitement among the SCZ group, which remained stable over the 2 weeks in the hospital. This provides evidence that the PANSS scores were providing an accurate index of illness state and were sensitive to treatment effects in this study. More relevantly, changes in the PANSS scores and changes in context sensitivity were uncorrelated for the SCZ group (all *p*s *>* 0.16). For the FEP group, context sensitivity changes were positively correlated with changes in negative symptoms (*rho* = 0*.*56, *p* = 0*.*048) but not with other symptoms (*p*s *>*43). When all patients were combined, there were no statistically significant correlations between changes in symptoms and context sensitivity (*p*s *>* 0.51). We also considered whether context sensitivity at Time 1 could predict symptom changes across time points. It was found that—at Time 1—higher context sensitivity in the SCZ group predicted a greater reduction in positive symptoms from admission to discharge (*rho* = 0*.*58, *p* = 0*.*03), but not other symptom types (*p*s *>* 0.29). An important caveat is that none of the symptom correlates described in this paragraph were specifically predicted, and none would remain significant if corrected for multiple comparisons. Of note, the smaller sample sizes at Time 2 limit the ability to detect statistical significance. Thus, it will be interesting to observe whether two notable *rho* values for the SCZ group–between increases in context sensitivity over time and reductions in positive (−0*.*39) and excitement (−0*.*36) symptoms–hold up with continued data collection (see below).

# **DISCUSSION**

Extending past findings, we found that—compared to healthy controls and people with a first episode of psychosis—persons with schizophrenia exhibit markedly less size contrast sensitivity at hospital admission (Uhlhaas et al., 2006a,b; Tibber et al., 2013). Consistent with other perception studies, (Silverstein et al., 1996; Uhlhaas et al., 2005; Silverstein and Keane, 2009; Keane et al., 2013), we also uncovered a state effect wherein the three groups demonstrated comparable context sensitivity by hospital discharge. Finally, we found that lessened context sensitivity may arise to some extent by the first episode of psychosis. It remains to be seen whether this is due to abnormal scores among a subgroup of FEP patients who go on to have schizophrenia as opposed to an affective psychosis. This question will be addressed in a subsequent report after longitudinal data are collected on this study sample.

A potential confound in the Time 1 findings is that SCZ patients may have demonstrated little context sensitivity not because of visual deficits but simply because they were not engaging in the task. It is possible that their advanced illness state caused them to randomly guess more often or occasionally press the wrong keys, leading to higher accuracy compared to other groups in the misleading condition and lower accuracy in the helpful condition. This explanation, if true, implies that the SCZ group should have performed worse than the FEP and CON groups in the no-context condition. It was found that the SCZ group's accuracy (86%) was about the *same* as that of the FEP group (91.4%, *p* = 0*.*17) but lower than that of the CON group (94.2%, *p* = 0*.*003). The direction of the difference is consistent with a generalized deficit, but the magnitude is not: a significant 8.2% dip in overall baseline accuracy doubtfully can explain a dramatic 64.9% group difference in context sensitivity (see **Figure 3**). Moreover, increased guessing among schizophrenia patients cannot explain why they performed far above chance in the misleading condition (69%) whereas controls and FEs performed right around chance (46% and 52%, respectively).

Another possibility is that SCZ patients became confused on the task precisely when a context was presented along with the targets: in these trials, subjects may have inadvertently judged the sizes of the surround circles rather than the central circles. However, if this were true, SCZ patients would have performed significantly better on the misleading than the helpful context trials, which did not occur (see **Table 2**). These results, taken together, indicate that there is a legitimate reduction in context sensitivity among people with schizophrenia.

# **SYMPTOM CORRELATES**

A paradoxical finding was that, in the SCZ group, more severe positive, excitement, negative and depression symptoms were associated with higher—and hence more normal—context sensitivity. Yang et al. (2013) found a similar effect with orientation and motion suppression tasks in which a central target is perceived to be more oriented or moving more in one direction when the surround contains elements are moving or oriented in the opposite direction. In that study, more pronounced Brief Psychiatric Rating Scale (BPRS) positive and negative symptom scores correlated with a greater effect of the surround (Yang et al., 2013). The effect sizes were not small (*r* = 0*.*67, *p <* 0*.*001 for motion; *r* = 0*.*49, *p* = 0*.*01 for orientation) and so cannot be dismissed as Type I errors. Why would more symptomatic patients behave more like healthy controls? Our finding that more normal context sensitivity is associated with increased positive, excitement, and depression symptoms may reflect both the typical co-occurrence of these symptoms, as well as their associated cognitive features. For example, positive symptoms and depression (which often includes agitation and excitement) often co-occur (Lindenmayer et al., 1991), and are related to a recent relapse as opposed to a chronically disabled state (Mulholland and Cooper, 2000; Hartley et al., 2013)—in other words, they are associated with patients who are higher functioning at their baseline. In terms of cognitive style, schizophrenia patients with more positive symptoms may allocate greater attentional resources to processing of the (irrelevant) contextual surrounds, leading to greater context sensitivity compared to the other groups. This is consistent with evidence of greater attentiveness to irrelevant cues being significantly correlated with positive symptoms (Morris et al., 2013). Yet another possibility is that some subjects, to varying degrees, are either unwilling or unable to be forthright about the true level of their symptoms. These subjects may have higher levels of impairment, poorer insight and prognosis, and more impaired visual processing. All of these explanations are speculative and, at this point, these unpredicted symptom correlates remain in need of further investigation.

Two symptom correlates that were predicted in our study were those between context sensitivity and cognitive disorganization in general and conceptual disorganization in particular. These effects replicate three earlier studies (Uhlhaas et al., 2006a,b; Horton and Silverstein, 2011), but not two recent ones (Tibber et al., 2013; Yang et al., 2013), which found no effect. Note however, that the Yang et al. (2013) study did not find evidence of a reduced Ebbinghaus illusion in schizophrenia, and studied clinically stable patients with little or no disorganization. A further difference between the Yang et al. and other studies, as noted by Tibber et al. (2013), is that the former study used unlimited stimulus presentation times, and this may diminish the illusion effect among all subjects, making significant correlations harder to detect. In addition, the task used by Yang et al. required comparing the size of a single circle to that of a circle with surrounds, and this method has been found to produce an illusion that is only about half as strong as when the sizes of two circles, each with surrounds, are compared (Franz et al., 2000), as was the case in the present study and other studies that found a reduced illusion effect in schizophrenia. The differences in patient samples between our study and the one by Tibber et al. (2013) may also explain why we found correlations between poorer context sensitivity and disorganization and they did not. Specifically, the patient sample in Tibber et al. (2013) was 79% outpatient, and 50% paranoid subtype, and only 3 of 24 patients scored greater than 3 on the PANSS Conceptual Disorganization item. This was, in general, a higher functioning sample than the one we studied (which was 100% inpatient), and the ones used in past studies where reduced context sensitivity and links to disorganization have been observed. In addition, paranoid subtype patients are generally higher functioning than disorganized patients and only typically begin to demonstrate disorganization after years of clinical deterioration, during which paranoid symptoms are reduced (McGlashan and Fenton, 1993). Moreover, disorganized symptoms are typically related to a poorer prognosis (Salokangas et al., 2002), and we have previously observed that reduced context sensitivity on an Ebbinghaus illusion task was more common in long-stay state hospital patients than in short-stay community hospital patients with schizophrenia (Uhlhaas et al., 2006b). In this study, positive and cognitive/disorganized symptom levels were independent of each other (*r* = −0*.*05, *p* = 0*.*83 at Time 1, and *r* = 0*.*07, *p* = 0*.*81 at Time 2), which was expected on the basis of prior factor analytic work (Lindenmayer et al., 1994a,b, 1995a,b) and the different relationships of these two factors with prognosis (Salokangas et al., 2002; Schennach-Wolff et al., 2011).

# **LIMITATIONS**

The data reported here have at least two important limitations. One is that the sample sizes are relatively small for the second time point, especially for the patient groups. Because the data reported here are preliminary findings from an ongoing study, later evidence will establish whether these findings are robust. The second limitation is that 1/3 of the SCZ patients who enrolled in the study were unavailable for testing at Time 2, as noted above (Results). This may have reduced the observed degree of change in context sensitivity in the SCZ group across time points. This suggestion is based on the findings that the SCZ patient subgroup that was unavailable at Time 2 was the least context sensitive at Time 1, and because patients (and controls) with the least context sensitivity at Time 1 demonstrated the greatest degree of change (improvement) across time points. It should be noted, however, that if our reasoning is incorrect and if those 7 SCZ patients' impairment at Time 1 was so severe that they would have *not* improved from hospital admission to discharge, then our available results would be biased in the direction of overestimating normalization of task performance in the SCZ group. Regardless, the available data make it clear that at least for the majority of SCZ patients tested (14/21 in this case), a clear perceptual impairment that is

#### **REFERENCES**


present at hospital admission, and that can not be accounted for by a generalized deficit, is not present at hospital discharge.

# **CONCLUSION**

This paper is the first report from an ongoing longitudinal study that investigates whether perceptual measures can predict symptom severity and/or level of functioning across multiple time points. To our knowledge, this is only the second study to examine longitudinal and treatment-related change in visual perception in schizophrenia (after Uhlhaas et al., 2005) and it is the first study of any kind to examine surround suppression in persons with a first episode of psychosis. In this preliminary report, we show that context sensitivity declines during acute phases of the illness, and—among schizophrenia patients—normalizes with short-term inpatient treatment. We also show that persons with first-episode psychosis exhibit marginally reduced context sensitivity at inpatient admission. These data suggest that performance on this Ebbinghaus illusion task may serve as a biomarker of relapse and recovery in people with schizophrenia. In later papers, we will report on how task performance, either at hospital discharge or at later time points, predicts short-term, post-hospital, prognosis and likelihood of relapse. We are especially interested in determining if first episode patients who demonstrate abnormal performance at either hospital-based testing have poorer outcomes, or if emergence of abnormal task performance over the next 1.25 years of illness is associated with a decline in functioning and/or a final diagnosis of schizophrenia versus a form of affective psychosis. Our other goal is to compare the predictive validity of perceptual task performance for established SCZ patients, relative to other putative biomarkers of treatment response for this population.

# **ACKNOWLEDGMENTS**

This study was supported by NIMH grant R01MH093439, Perceptual organization dysfunction as a biomarker of schizophrenia, to the first author. We thank the patients and staff at Rutgers University Behavioral Health Care for their help with this study, and Drs. Martin Doherty and William A. Phillips for providing us with the Ebbinghaus illusion task.


NY: New York State Psychiatric Institute.


deterioration in early schizophrenia. *Schizophr. Bull.* 19, 71–84. doi: 10.1093/schbul/19.1.71


Results of the DSP project. *Nord. J. Psychiatry* 56, 319–327. doi: 10.1080/080394802760322079


review of research from 2005 to 2010. *Schizophr. Bull.* 37, 690–699. doi: 10.1093/schbul/sbr052


schizophrenia. *Front. Psychol.* 4:88. doi: 10.3389/fpsyg.2013.00088


*Psychol. Res.* 72, 304–310. doi: 10.1007/s00426-007-0114-6


Diminished orientation-specific surround suppression of visual processing in schizophrenia. *Schizophr. Bull.* 35, 1078–1084. doi: 10.1093/schbul/sbp064

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 02 May 2013; accepted: 03 July 2013; published online: 24 July 2013.*

*Citation: Silverstein SM, Keane BP, Wang Y, Mikkilineni D, Paterno D, Papathomas TV and Feigenson K (2013) Effects of short-term inpatient treatment on sensitivity to a size contrast illusion in first-episode psychosis and multipleepisode schizophrenia. Front. Psychol. 4:466. doi: 10.3389/fpsyg.2013.00466 This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers*

*in Psychology. Copyright © 2013 Silverstein, Keane, Wang, Mikkilineni, Paterno, Papathomas and Feigenson. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Object substitution masking in schizophrenia: an event-related potential analysis

**Jonathan K.Wynn1,2\*, Kristopher I. Mathis <sup>1</sup> , Judith Ford3,4, Bruno G. Breitmeyer <sup>5</sup> and Michael F. Green1,2**

<sup>1</sup> Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA, USA

<sup>2</sup> Semel Institute for Neuroscience and Human Behavior, University of California, Los Angeles, Los Angeles, USA

<sup>3</sup> San Francisco Veterans Affairs Medical Center, San Francisco, CA, USA

<sup>4</sup> Department of Psychiatry, University of California San Francisco, San Francisco, CA, USA

<sup>5</sup> Department of Psychology, University of Houston, Houston, TX, USA

#### **Edited by:**

Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA

#### **Reviewed by:**

Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA Anne Giersch, Institut National de la Santé et de la Recherche Médicale, France

#### **\*Correspondence:**

Jonathan K. Wynn, Veterans Affairs Greater Los Angeles Healthcare System, Mental Illness Research, Education, and Clinical Center, Building 210, 11301 Wilshire Boulevard, Los Angeles, CA 90073, USA.

e-mail: jkwynn@ucla.edu

Schizophrenia patients exhibit deficits on visual processing tasks, including visual backward masking, and these impairments are related to deficits in higher-level processes. In the current study we used electroencephalography techniques to examine successive stages and pathways of visual processing in a specialized masking paradigm, four-dot masking, which involves masking by object substitution. Seventy-six schizophrenia patients and 66 healthy controls had event-related potentials (ERPs) recorded during four-dot masking.Target visibility was manipulated by changing stimulus onset asynchrony (SOA) between the target and mask, such that performance decreased with increasing SOA. Three SOAs were used: 0, 50, and 100 ms.The P100 and N100 perceptual ERPs were examined. Additionally, the visual awareness negativity (VAN) to correct vs. incorrect responses, an index of reentrant processing, was examined for SOAs 50 and 100 ms. Results showed that patients performed worse than controls on the behavioral task across all SOAs. The ERP results revealed that patients had significantly smaller P100 and N100 amplitudes, though there was no effect of SOA on either component in either group. In healthy controls, but not patients, N100 amplitude correlated significantly with behavioral performance at SOAs where masking occurred, such that higher accuracy correlated with a larger N100. Healthy controls, but not patients, exhibited a larger VAN to correct vs. incorrect responses. The results indicate that the N100 appears to be related to attentional effort in the task in controls, but not patients. Considering that the VAN is thought to reflect reentrant processing, one interpretation of the findings is that patients' lack of VAN response and poorer performance may be related to dysfunctional reentrant processing.

**Keywords: visual processing, schizophrenia, ERP, visual backward masking, reentrant processing**

# **INTRODUCTION**

Patients with schizophrenia exhibit several visual processing impairments (Green et al., 1994; Butler et al., 2003; Butler and Javitt, 2005; Rassovsky et al., 2005; Wynn et al., 2005; Silverstein and Keane, 2011), and these impairments have been tied to specific neural abnormalities, such as magnocellular pathway dysfunction, NMDA functioning, and activation in the lateral occipital complex (Butler et al., 2005; Green et al., 2009; Javitt, 2009). However, it is unclear whether visual processing impairment is isolated to early or later processing stages (Javitt, 2009; Dias et al., 2011; Rassovsky et al., 2011). Visual processing divides roughly into two stages: (1) initial formation of a percept, and (2) subsequent processing of the percept until it reaches awareness (Di Lollo et al., 2000; Enns and Di Lollo, 2000; Ro et al., 2003; Enns, 2004; Chen and Treisman, 2009; Dux et al., 2010). The first stage is thought to involve a neural feedforward sweep from retina to primary visual cortex. The second stage involves reentrant processes (i.e., recurrent cortico-cortical circuits) that refine the initially ambiguous percept until it becomes recognizable. These reentrant processes have even been shown to feedback

to very early neural areas such as the lateral geniculate nucleus (Sillito et al., 1994).

Visual backward masking can be used to assess both of these stages. Masking that disrupts feedforward processing usually occurs between 0 and 100 ms after target onset. Masking at the reentrant stage occurs slightly later (starting approximately 100 ms after stimulus onset;Woodman and Luck, 2003; Prime et al., 2011), and results in the mask displacing the neural representation of the target, a phenomenon termed object substitution masking (Di Lollo et al., 2000; Enns and Di Lollo, 2000; Chen and Treisman, 2009;Dux et al., 2010). The premise underlying object substitution is that the feedforward sweep of the target must be refined at higher cortical levels through reentrant processes to the lower input levels. If the mask is presented before the target is identified at this higher-level, there will be a mismatch between the feedforward and reentrant sweeps, and processing will switch to the mask. In essence, the target may be correctly identified at some level within the primary visual processing stream but accurate information of that target is not passed on to higher-level visual processing systems. Object substitution masking, therefore, has provided a

method to examine reentrant visual processing (Woodman and Luck, 2003).

Four-dot masking is a paradigm that is thought to act through object substitution (Enns and Di Lollo, 1997; Di Lollo et al., 2000; Enns, 2004). In four-dot masking, performance is greatest when the target and the mask have the same onset and offset (effectively cuing the observer to the target). Four-dot masking can occur when both target and mask have the same onset (termed "common-onset masking") and the mask remains visible longer than the target and can also occur when the onset of the mask appears at varying stimulus onset asynchronies (SOAs) relative to the target (Enns and Di Lollo, 2000). Using this second type of four-dot masking in a behavioral study we found impairments in schizophrenia patients (Green et al., 2011). Moreover, it did not appear that the patients' deficit was due to issues in initial processing of the target because when using the same SOAs and using a single dot to indicate which target to identify, we found: (1) that there was no group difference in behavior; and (2) iconic decay of the target was not different between groups. Therefore, the increased masking effects in schizophrenia patients appeared to be due to the presence of the mask and not other factors. However, we could not determine with a behavioral task if deficits in schizophrenia patients in the object formation stage of visual processing contributed to the deficits assessed with four-dot masking. Indeed, the version of four-dot masking that we used (with delayed-onset mask) could have involved some early masking of object formation, in addition to its characteristic object substitution masking. With the excellent temporal resolution of electroencephalography (EEG) and event-related potentials (ERPs) it is possible to evaluate perceptual and post-perceptual stages separately.

Specific ERP waveform components track with stages of visual information processing, including the visual P100 and N100 and the visual awareness negativity (VAN). The P100 is a positive wave peaking approximately 100 ms post stimulus onset. Several neural generators of the P100 have been identified, with activity largest over dorsal stream sites, with extrastriate and striate contributions (Maier et al., 1987; Aine et al., 1995; Di Russo et al., 2001; Vanni et al., 2004). The N100 is a negative wave peaking approximately 150 ms post stimulus onset. Multiple generators have also been identified for the N100, with the largest activity seen in ventral stream structures such as the object-sensitive lateral occipital complex (LOC; e.g., Bentin et al., 1999; Doniger et al., 2000, 2001, 2002). TheVAN has been reported to appear as a relative increase in negativity to conscious vs. unconscious (e.g., correct vs. incorrect) stimuli appearing approximately 200–300 ms in occipito-temporal sites, though there is considerable variability as to when this component peaks (for a review, see (Railo et al., 2011). Moreover, this ERP component has been proposed as a measure of reentrant cortical activity (Koivisto et al., 2006; Wilenius and Revonsuo, 2007) because of its regional distribution, it is associated with awareness of a visual target and its latency falls within the time frame of when reentrant cortical activity is thought to occur (Lamme and Roelfsema, 2000; Di Russo et al., 2001).

Several ERP studies have examined backward masking in healthy participants. Fahrenfort et al. (2007) found that both masked and unmasked targets produced a strong bilateral anterior occipito-temporal activation that occurred prior to 110 ms (likely consistent with the visual P100), which the authors attributed to feedforward processing of visual stimuli. However, only unmasked trials resulted in activity between 110–140 ms at posterior occipital sites, with no activity for masked trials seen at this time frame. The authors interpreted these findings as reflecting reentrant processing being interrupted by the presence of the mask.With a backward masking procedure similar to Fahrenfort et al; Van Loon et al. (2012) found that masking had no effect on the earliest ERP components (i.e., <120 ms, corresponding to the P100) but effectively decreased later ERP components (i.e., >150 ms, corresponding to the N100). Moreover, only the later, N100-like response correlated with behavior, such that greater accuracy correlated with a larger N100.

To date, only a handful of ERP studies of four-dot masking have been published (Woodman and Luck, 2003; Reiss and Hoffman, 2006; Kotsoni et al., 2007; Prime et al., 2011) and these studies evaluated whether four-dot masking is consistent with object substitution. In summary, these studies found that feedforward processing of an object is left intact, but later reentrant processing of the target is interrupted due to the mask substituting the target during this stage.

In the current study, we tracked the time course of visual information processing during object substitution masking in a large sample of schizophrenia patients and healthy controls. We manipulated visibility of the targets using various SOAs, such that performance is best at an SOA of 0 ms, with performance decreasing at higher SOAs. We chose this method of delayed-onset masking, rather than common-onset masking, to more directly compare the results of the current study to other masking paradigms commonly used in schizophrenia research that have delayed-onset masks (e.g., see Green et al., 2011). We used ERPs to assess early and later perceptual stages (P100 and N100) and theVAN to examine activity thought to be directly related to reentrant processing. We used the exact same paradigm as in our behavioral study that showed patient–control differences, though sampled a fewer number of SOAs. Based on prior EEG studies of backward masking, we hypothesized that the VAN would be the first component sensitive to object substitution masking. In particular, we expected that in healthy controls a larger VAN to correct vs. incorrect responses would be seen, whereas patients would not exhibit this effect.

#### **MATERIALS AND METHODS PARTICIPANTS**

Most participants also participated in a behavioral study of object substitution in a separate session and using a somewhat different procedure from the one presented in the current paper (Green et al., 2011). Seventy-seven stabilized outpatients with schizophrenia and 66 healthy control subjects participated in the study. One patient was excluded from analysis due to having an insufficient amount of usable EEG data (see below). Thus, the final patient sample size was *n* = 76. Patients were recruited from outpatient treatment clinics at the Veterans Affairs Greater Los Angeles Healthcare System (VAGLAHS) and through presentations at community residences. Patients met criteria based on the Structured Clinical Interview for DSM-IV Axis I Disorders (First et al., 1997). Sixty-two patients were receiving atypical antipsychotic medications, seven were receiving typical antipsychotic medications, three were receiving both types of medication, and four were not taking an antipsychotic medication at time of assessment.

Healthy control participants were recruited through internet and newspaper advertisements. Control participants were screened with the SCID and SCID–II (First et al., 1996) and were excluded if they met criteria for any lifetime psychotic disorder; bipolar mood disorder; recurrent depression; substance dependence; paranoid, schizotypal, or schizoid personality disorder; or any evidence (according to participant report) of a history of psychotic disorder among their first-degree relatives.

Additional exclusion criteria for both groups included being younger than 18 or older than 60 years, diagnosed with an active substance use disorder in the past 6 months, any identifiable neurological disorder, mental retardation, history of loss of consciousness for more than 1 h, or insufficient fluency in English. All participants had the capacity to give informed consent and provided written informed consent after all procedures were explained in accordance with procedures approved by the Institutional Review Boards at UCLA and VAGLAHS.

#### **CLINICAL RATINGS**

Psychiatric symptoms during the previous month were rated using the 24-item UCLA version of the Brief Psychiatric Rating Scale (BPRS; Overall and Gorham, 1962; Lukoff et al., 1986) and the Scalefor theAssessment of Negative Symptoms (SANS;Andreasen, 1984) by a trained rater. For the BPRS we report the "positive symptom" and "depression/anxiety" factors (Kopelowicz et al., 2008); for the SANS we report the global scores for affective flattening, alogia, anhedonia, and avolition (**Table 1**). All clinical assessments were conducted by interviewers trained to reliability through the Treatment Unit of the Department of Veterans Affairs VISN 22 Mental Illness Research, Education, and Clinical Center (MIRECC) based on previously reported procedures (Ventura et al., 1993, 1998).

#### **BACKWARD MASKING TASK**

All stimuli were presented using E-Prime 1.1 (Psychological Software Tools, Pittsburgh, PA) on a 17<sup>00</sup> cathode ray tube monitor running at a 160 Hz refresh rate. Participants sat 1 m away from the monitor.

The four-dot masking procedure was modified from similar tasks described elsewhere (Enns and Di Lollo, 2000; Enns, 2004) and was based on a larger behavioral study of four-dot masking from our laboratory (Green et al., 2011). In the task, four potential targets, consisting of four squares with a notch missing from the top, bottom, or side, appeared in a notional square on the monitor. Following the target stimuli, a mask was presented which comprised four-dots arranged in a square that surrounded, but did not touch, one of the potential targets (see **Figure 1**). The mask cued the location of the target. Each potential target measured 1.55˚ × 1.55˚ of visual angle and was arranged in a square of 4.58˚ × 4.58˚ of visual angle. The four-dot mask measured 2.23˚ × 2.23˚ of visual angle and each dot in the mask subtended 0.23˚ × 0.23˚ of visual angle. The target array was presented for 25 ms and the mask was presented for 37.5 ms. All stimuli were suprathreshold, with stimuli presented in black

**Table 1 | Demographic information and symptom ratings.**


\*p < 0.01, for difference between controls and patients.

presented on a white background. The luminance of the background was 260 lx while that of the stimuli was 9 lx, resulting in a contrast of ∼93%, defined by Michelson's contrast: contrast = (*L*max−*L*min)/(*L*max + *L*min). We collected data on four target-mask stimulus onset asynchronies (SOAs: 0, 50, 100, and 150 ms). However, we only analyzed data from the first three SOAs, as the mask terminated on or before the ERP components of interest for these SOAs, and there was little difference in performance between SOAs of 100 and 150. Fifty-four trials per SOA were presented in quasi-randomized fashion (18 trials for each of the three potential sides where the notch appears).

Each trial started with a fixation cross presented for 450 ms followed by a blank screen for 500 ms. Target and mask stimuli, separated by the SOAs mentioned above, were then presented. A 1 s blank screen was then presented, followed by a prompt for the participant to make a response. Participants verbally reported the direction of the notch in the target and the experimenter entered the response and initiated the next trial. The total number of correct responses (out of 54) at each SOA was analyzed.

#### **EEG RECORDING**

Electroencephalography was continuously recorded using a 64 channel Neuroscan SynAmps2 amplifier and a Neuroscan 64 channel QuickCap (Compumedics USA, Charlotte, NC). Data were sampled at 500 Hz with a bandpass of 1–100 Hz. Horizontal electrooculogram (EOG; placed on the outer canthus of the left and right eye) and vertical EOG (placed above and below the left eye) was also recorded. The reference during recording was a point halfway between electrodes Cz and CPz, and all sites were re-referenced offline to the average of the left and right mastoids. An electrode affixed to the forehead served to ground the array.

#### **ERP DATA ANALYSIS**

Data were processed offline using Neuroscan Scan 4.3 and Brain-Vision Analyzer 2 software (Brain Products, Gilching, Germany).

Vertical eye blinks were removed using a regression-based algorithm (Semlitsch et al., 1986). Data were low-pass filtered at 30 Hz, with a 24 dB roll off zero phase shift filter. Data were epoched to 100 ms pre- and 700 ms post-target onset. Baseline correctionfrom the 100 ms prior to stimulus presentation was applied. Epochs containing activity that exceeded ±100µV were rejected at the sites that were used for data analysis (P7, P5, P6, P8, PO7, PO8, O1, O2). Visual inspection of trials was then performed to eliminate any remaining abnormal EEG responses. The mean number of valid trials per SOA (out of a total of 54) included in subsequent statistical analyses was 51.7 (range for individual participants, 29– 54) and 52.4 (range for individual participants, 44–54) for patients and controls, respectively.

Event-related potentials were created by averaging together all accepted trials (regardless of accuracy), separately for each SOA. A time window was defined for each ERP component based on the peak activity observed by inspection of the mean global field power averaged across all subjects and SOAs. The width of the time window was selected to ensure coverage of each component and the mean activity within each window was the main dependent measure. The time windows were 68–108 and 134–174 ms for the P100 and N100, respectively. The P100 and N100 were examined in six parieto-occipital electrode sites where visual ERPs were largest based on visual inspection of the waveforms (P7, PO7, P5, P8, PO8, and P6). These sites were also chosen as they overlap with sites used in previous examinations of the N100 and P100 response in visual processing studies in schizophrenia (e.g., Butler et al., 2007). Activity was examined separately for each hemisphere by averaging the three left and the three right electrodes.

The VAN was analyzed in 50 patients and 43 healthy controls that had at least 15 artifact-free epochs for both correct and incorrect responses at each SOA of 50 and 100 ms. For patients, there were a mean (SD) 24.1 (5.6) and 28.0 (5.5) trials for correct and incorrect, respectively, averaged over the two SOAs; for controls there were 28.6 (6.2) and 24.2 (6.1) trials for correct and incorrect, respectively. Activity at electrodes PO7, PO8, O1, and O2 were examined for the VAN. The VAN was measured as the mean activity within the time window of 250–310 ms. Additionally, the P100

and N100 (using the same time windows and electrodes described above) to correct and incorrect responses were examined.

# **DATA ANALYSIS**

For demographic data, group differences were evaluated with *t*tests and chi-square tests. 2 (group) × 2 (hemisphere) × 3 (SOA) repeated measures ANOVAs were run separately for P100 and N100 for the entire sample, disregarding accuracy. To examine the VAN and accuracy effects on the P100 and N100 in the subsample, separate 2 (group) × 2 (accuracy) × 2 (SOA) repeated measures ANOVAs were run. In cases of repeated measures with more than one degree of freedom, we used Greenhouse–Geisser correction factors (ε). We report the uncorrected degrees of freedom, the corrected *p*-value, partial eta-square effect sizes (η 2 *p* ), and ε. All statistical analyses used a two-tailed significance level of 0.05. Bonferroni-corrected multiple comparisons were used to ensure a family wise significance level of *p* < 0.05.

# **RESULTS**

# **DEMOGRAPHICS**

Demographic information is presented in **Table 1**. Patients were clinically stable with relatively low levels of symptoms. They were significantly older than controls but did not differ in terms of gender or parental education. As we have shown in our previous studies, age is significantly correlated with backward masking performance (Green et al., 2011). Age is also largely correlated with performance in the current study for both groups, with *r*'s between −0.30 and −0.38 for each SOA. Therefore, we included age as a covariate in our analyses. We report age-corrected means (standard deviations) where appropriate.

# **BEHAVIOR**

Results of the analysis revealed a significant main effect of group (F1, 139 = 19.76, *p* < 0.001, η 2 *<sup>p</sup>* = 0.12) and a significant main effect of SOA (F2, 278 = 8.48, *p* < 0.001, ε = 0.92, η 2 *<sup>p</sup>* = 0.06). The group by SOA interaction was not significant (*p* > 0.17). Overall, patients showed poorer performance compared to controls and both groups showed the expected decline in accuracy as SOA increased (see **Figure 2**).

# **EEG**

Means and SD for the two ERP components can be seen in **Table 2**. **Figure 3** shows the ERP waveforms (averaged over the six parietooccipital electrodes mentioned above) at each SOA for each group. **Figure 4** shows the VAN waveforms (averaged over the two SOAs and the four parieto-occipital electrodes mentioned above) for each group.

#### **P100**

There was only a significant effect for group (F1, 139 = 3.94, *p* < 0.05, η 2 *<sup>p</sup>* = 0.03). There were no other significant main effects or interactions. Controls had a larger P100 compared to patients [1.92 (1.26) versus 1.41 (1.25)µV, respectively].

#### **N100**

Results of the analysis revealed a significant main effect of group (F1,139 = 5.42, *p* < 0.03, η 2 *<sup>p</sup>* = 0.04) and a significant group × SOA interaction (F2, 278 = 7.81, *p* < 0.001,ε = 0.98, η 2 *<sup>p</sup>* = 0.05). Controls had a significantly larger N100 compared to patients [−2.86(2.13) versus −1.77(2.01)µV]. Controls showed significantly larger N100 amplitudes at all SOAs, except at SOA 50, compared to the patients. The group × SOA interaction was due to patients showing an increase in amplitude at SOA 50 whereas controls show a decrease, a pattern difference that was not predicted.

#### **EFFECTS OF ACCURACY ON ERPs AT SOAs 50 AND 100 ms P100**

The main effects of group, accuracy, and SOA were not significant; there were also no significant interactions, all *F*'s < 1.40,*p*'s > 0.24.

#### **N100**

The main effects of group, accuracy, and SOA were not significant. There was only a significant group × SOA interaction

**Table 2 | Mean (SD) amplitudes for each group and each ERP component.**


L, left; R, right.

(F1, 90 = 11.25, *p* < 0.001, η 2 *<sup>p</sup>* = 0.11). The group × SOA interaction was due to patients and controls having similar amplitudes at an SOA of 50, whereas controls had a larger amplitude than patients at an SOA of 100 ms.

#### **VAN**

As the VAN appears as a greater difference between correct vs. incorrect responses, we expected to see a significant difference in the healthy controls, and we did. For controls, the difference between correct and incorrect reached significance (t<sup>42</sup> = 2.47, *p* < 0.02), 0.27 (1.25) vs. 0.59 (1.15) µV, correct vs. incorrect respectively (effect size = 0.27). In contrast, for the patients the *t*-tests showed no significant difference for correct vs. incorrect responses in patients (t<sup>49</sup> = 0.55, *p* < 0.60),−0.15 (0.94) vs. −0.08 (1.03) µV, correct vs. incorrect respectively (effect size = 0.07). The group × accuracy interaction failed to reach significance (F1, 90 = 2.18, *p* < 0.15, η 2 *<sup>p</sup>* = 0.02). Hence, although group differences in VAN were predicted, they need to be interpreted with caution given the non-significant group by accuracy interaction.

#### **RELATIONSHIP BETWEEN ERPs AND BEHAVIOR**

Correlations were examined between the P100 and N100 and behavioral performance at each SOA for each group, controlling for age. P100 and N100 activity was averaged across hemisphere for each SOA. As can be seen in **Table 3**, P100 amplitude was largely uncorrelated with performance in both groups (only one of the eight correlations reached significance). N100 amplitude correlated significantly with performance at the two later SOAs (where masking effects occurred), but only in the healthy controls. P100 was significantly correlated with N100 at each respective SOA in both groups, *r*'s ranging from −0.50 to −0.33.

#### **AGE EFFECTS**

Because analyses of covariance can sometimes misrepresent relationships among variables, we considered the potential effect of age on the behavioral and ERP data, without including a covariance analysis. We performed a median split, creating a "young" patient group (*n* = 39, mean age = 38.6) and an "old" patient

are noted by arrows. SOA 0 ms, dark blue; SOA 50 ms, red; SOA 100 ms, green.

**Table 3 | Correlations between backward masking performance (# targets correctly identified) and ERP components.**


\*p < 0.05.

group (*n* = 37, mean age = 54.4). The "young" patient group more closely matched the age of the healthy control group,with no statistical difference in age between these two groups. Two analyses were conducted. First, differences in ERPs and behavior between the "young" and "old" schizophrenia patient groups were examined;

no significant group effects or interactions with group were seen (all *F*'s < 1.0, *p*'s > 0.4). Second, differences in ERPs and behavior between only the "young" patients and the healthy controls were examined. Similarly, we performed this same split for the VAN analyses, with a "young" patient group (*n* = 25, mean age = 40.0) and an "old" patient group (*n* = 25, mean age = 50.5) being created. Again, there were no difference between the "young" and "old" patient groups and the pattern of results remained the same comparing the "young" patients to the healthy control sample. These analyses revealed the exact same pattern of results as when using the entire patient sample, indicating the presence of behavioral and ERP differences between schizophrenia patients and healthy controls regardless of age.

#### **DISCUSSION**

Using ERPs, we were able to clarify the nature of the patients' behavioral deficit in four-dot object substitution masking. Several key findings emerged from this study. First, patients exhibited

significantly reduced P100 and N100 amplitudes compared to controls across all SOAs, indicating impairment in perceptual processes across levels of visibility. Second, N100 correlated with accuracy in controls, but not patients, at SOAs where masking occurs. Third, controls, but not patients, exhibited a VAN response, though the group by accuracy interaction did not reach significance. These results provide further evidence that schizophrenia patients likely have a dysfunction in reentrant visual processing.

The reduced amplitude of P100 and N100 is consistent with previous reports of EEG abnormalities during early visual perception in schizophrenia (Foxe et al., 2001; Doniger et al., 2002; Schechter et al., 2005), suggesting dysfunction at the earliest stages of object formation. However, it seems unlikely that the P100 component directly affected object substitution for either group (i.e., amplitudes did not change with changing visibility). As mentioned in the introduction, ERP studies of backward masking show no effect on the P100, indicating that feedforward processing is left relatively intact. Furthermore, in the accuracy analysis of the ERP data, there was no effect of accuracy on the P100 and the subgroup of patients and controls included in the accuracy analyses did not differ in P100 amplitude. We found a somewhat mixed pattern for the effects of object substitution masking on the N100, in that there were no effects of SOA (i.e., visibility) and no effect of accuracy on N100 amplitude. On the other hand, there was a correlation between behavior and amplitude in the controls that may simply reflect greater attentional effort devoted to processing stimuli that were harder to accurately detect.

The finding of the lack of a VAN to correct vs. incorrect trials in schizophrenia patients points to dysfunctional reentrant processing. As mentioned in the introduction, the VAN appears to be the first ERP correlate of visual consciousness and likely is a direct measure of reentrant processing. Moreover, there was no effect of accuracy on either the P100 or N100, reflecting that these components are unlikely related to reentrant processing. The healthy controls exhibited aVAN, indicating that effective four-dot masking in this study interrupted reentrant visual processing. Given the lack of this finding in the schizophrenia patients, these results imply that schizophrenia patients' deficit during object substitution may be due to a type of dysfunction (e.g., lack of neural synchronization, coherence, etc.) in reentrant processing along this pathway. Regardless of how patients correctly identified stimuli in the current study, the VAN results suggest a dysfunction in reentrant processing. However, the lack of a significant group × accuracy interaction in the VAN data means that we interpret these findings with caution.

There also remains the possibility that other processes, such as consciousness of stimuli and attentional effort, may be accounting for our behavioral and EEG findings. For example, Del Cul et al. (2006) found that schizophrenia patients' ability to consciously process targets in the presence of a mask was diminished in comparison to healthy controls; however, their ability to subliminally process the targets was intact. The authors interpreted their findings as showing that patients' feedforward processing during masking was intact whereas conscious processing of those stimuli are dysfunctional. Also, Lalanne et al. (2012) suggested that backward masking impairments in schizophrenia patients may be due to a lack of focused attention on the target. Finally, using pupillometry during a backward masking task, Granholm et al. (2009) found that impaired performance in patients was attributable to abnormalities in attentional resource availability. These other factors may be alternative possibilities for the findings, or constitute explanations at other levels of processing.

As with our behavioral study (Green et al., 2011), patients' behavioral performance was worse than controls across all SOAs, including at the earliest SOA that has minimal masking. This finding limits interpretation in that it could be that patients are unable to correctly identify the target regardless of the effectiveness of the mask. We were able to address this possibility in our previous behavioral study which included a cuing procedure (see Introduction). For example,four-dot masking tasks depend on iconic decay (the visible persistence of a target) and group differences in this process could account for the masking deficit. However, we examined iconic decay in our behavioral study using a simple cuing task and found no differences between the patients and controls on iconic decay. Thus, differences in decay rates are unlikely to account for the performance differences. The current paradigm had no similar control procedure, and that is one limitation of the study.

The current study has other limitations. First, nearly all patients were medicated at the time of testing, potentially affecting the results. However, there is evidence that antipsychotic medication does not affect masking performance (Cadenhead et al., 1997; Green et al., 1999; Butler et al., 2003) or visual ERPs (Butler et al., 2001; Schechter et al., 2005), making medication effects unlikely. Second, our delayed-onset paradigm might not have completely eliminated the interrupting effect of the onset of the transient channels elicited by the mask (Jannati et al., 2011). While a common-onset masking procedure (Enns and Di Lollo, 2000) might be a more complete way to isolate reentrant processing during masking, there remains considerable evidence that even in metacontrast masking reentrant processing is being interrupted (Di Lollo et al., 2000; Breitmeyer, 2007; Fahrenfort et al., 2007). Therefore it is likely that our behavioral and VAN results reflect in a dysfunction in reentrant processing in patients with schizophrenia.

Another limitation is that our design was not optimized for eliciting P300 or N2pc that have been examined in the four-dot backward masking studies conducted in healthy controls (Woodman and Luck, 2003; Prime et al., 2011). Specifically, we did not have a rare stimulus, we had SOAs that were closely spaced, and we did not have lateralized presentation of the stimuli that would have enabled us to collect these other components. A small P300-like component is visible for the shortest SOA in **Figure 3**, but due to our parameters we did not consider it to be a valid P300. Finally, we did not equate the groups for their "unmasked" target performance (i.e., performance at an SOA of 0 ms), as we have done in our previous backward masking studies (Green et al., 2002, 2003, 2006).

The results from this study help clarify findings from our previous behavioral study on object substitution masking (Green et al., 2011). It was not possible to determine from the behavioral study (particularly given our choice of a delayed-onset mask) whether the deficits in the patients' identification of targets were due to impairment at the earliest stage of object formation or to impairment in a later stage where reentrant processes are necessary to refine the visual percept. The current study offers tentative support for the idea that impairment of visual processing was attributable to deficits associated with reentrant processing after the initial feedforward sweep. While there is abundant evidence, from our laboratory and others, that dysfunctional bottom-up processes in schizophrenia affect downstream processing of higher-level tasks (Sergi and Green, 2003; Leitman et al., 2005; Dias et al., 2011; Pinheiro et al., 2012), our findings with this paradigm implicate a slightly later visual processing stage.

#### **REFERENCES**


*Arch. Gen. Psychiatry* 62, 495–504.


**ACKNOWLEDGMENTS**

Support for this study came from National Institute of Mental Health grants MH43292 and MH065707 (principal investigator, Dr. Green). Jonathan Wynn was supported by a Veterans Affairs Career Development Award. The funding agencies had no other role in the study. None of the authors have any financial disclosures or conflicts of interest to report. Dr. Wynn takes responsibility for the integrity of the data and the accuracy of the data analyses. All authors had full access to all of the data in the study. The authors thank Daniel Mathalon, M.D., Ph.D. and Brian Roach for helpful comments on previous drafts of the manuscript. The authors thank Cory Tripp, Poorang Nori, and Mark McGee for help in data collection.


schizophrenia. *Psychophysiology* 46, 510–520.


Impairments of early-stage transient visual evoked potentials to magnoand parvocellular-selective stimuli in schizophrenia. *Clin. Neurophysiol.* 116, 2204–2215.


Training and quality assurance with the structured clinical interview for DSM-IV. *Psychiatry Res* 79, 163–173.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 30 August 2012; accepted: 14 January 2013; published online: 04 February 2013.*

*Citation: Wynn JK, Mathis KI, Ford J, Breitmeyer BG and Green MF (2013) Object substitution masking in schizophrenia: an event-related potential analysis. Front. Psychology 4:30. doi: 10.3389/fpsyg.2013.00030*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Wynn, Mathis, Ford, Breitmeyer and Green. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Schizophrenia and visual backward masking: a general deficit of target enhancement

#### *Michael H. Herzog1 \*, Maya Roinishvili 2,3, Eka Chkonia3,4 and Andreas Brand5*

*<sup>1</sup> Laboratory of Psychophysics, Brain Mind Institute, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland*


#### *Edited by:*

*Anne Giersch, Institut National de la Santé et de la Recherche Médicale, France*

#### *Reviewed by:*

*Muriel Boucart, CNRS-University Lille, France Szabolcs Kéri, University of Szeged, Hungary*

#### *\*Correspondence:*

*Michael H. Herzog, Laboratory of Psychophysics, Brain Mind Institute, Ecole Polytechnique Fédérale de Lausanne, Station 19, Plaza 1, EPFL, 1015 Lausanne, Switzerland. e-mail: michael.herzog@epfl.ch*

The obvious symptoms of schizophrenia are of cognitive and psychopathological nature. However, schizophrenia affects also visual processing which becomes particularly evident when stimuli are presented for short durations and are followed by a masking stimulus. Visual deficits are of great interest because they might be related to the genetic variations underlying the disease (endophenotype concept). Visual masking deficits are usually attributed to specific dysfunctions of the visual system such as a hypo- or hyper-active magnocellular system. Here, we propose that visual deficits are a manifestation of a general deficit related to the enhancement of weak neural signals as occurring in all other sorts of information processing. We summarize previous findings with the shine-through masking paradigm where a shortly presented vernier target is followed by a masking grating. The mask deteriorates visual processing of schizophrenic patients by almost an order of magnitude compared to healthy controls. We propose that these deficits are caused by dysfunctions of attention and the cholinergic system leading to weak neural activity corresponding to the vernier. High density electrophysiological recordings (EEG) show that indeed neural activity is strongly reduced in schizophrenic patients which we attribute to the lack of vernier enhancement. When only the masking grating is presented, EEG responses are roughly comparable between patients and control. Our hypothesis is supported by findings relating visual masking to genetic deviants of the nicotinic α7 receptor (CHRNA7).

**Keywords: schizophrenia, vision, acetylcholine receptor, vernier acuity, attention**

# **VISION AND SCHIZOPHRENIA: FINDINGS AND A FRAMEWORK**

The most obvious symptoms of schizophrenia are related to psychopathology (e.g., delusions, hallucinations, negative symptoms, and disorganization syndrome) and cognitive deficits such as working memory and attention. In addition, schizophrenic show strong perceptual deficits. The causes of schizophrenia may be found on various levels including ones which are not directly linked to phenomenology. Here, we propose that visual deficits in schizophrenic patients are not related to visual deficits *per se* but are a manifestation of a general dysfunction of enhancing and stabilizing neural activity.

#### **VISUAL BACKWARD MASKING DEFICITS**

To understand visual deficits in schizophrenia, we are using a backward masking procedure, called the shine-through effect (Herzog and Koch, 2001; Herzog and Fahle, 2002). This masking technique comes with a two step procedure. First, a vernier stimulus is presented which comprises two vertical bars slightly offset in the horizontal direction. Vernier offset discrimination is a challenging task. Offsets are often smaller than the diameter of a photoreceptor in the retina and, therefore, offset discrimination needs information integration across nearby neurons. Per trial, the vernier is randomly offset either to the left or right. Observers indicate the offset direction. For long vernier durations, schizophrenic patients show no or only very weak deficits. We presented the vernier with different durations starting off with a duration of 150 ms. For this duration, we found thresholds of 41.1 arc sec in patients and 33.0 arc sec in controls. Block by block, we reduced the vernier duration further until observers could not reliably perceive a vernier offset size of 0.66 arc min. This critical vernier duration was 43.2 ms for patients and 27.4 ms for controls (Chkonia et al., 2010). These results are in line with previous studies in which also significant, but moderate, visual deficits were found with different types of visual stimuli (e.g., Saccuzzo et al., 1974; Braff and Saccuzzo, 1981; Saccuzzo and Schubert, 1981).

To challenge the visual system, we added to the spatially demanding vernier offset discrimination task a temporal challenge by presenting a mask after the vernier. For each observer, we presented the vernier with his/her individual duration as determined in the first step. After the vernier, a blank screen followed, i.e., an inter-stimulus-interval (ISI), and then a mask. In the basic masking conditions, the mask comprised 5 or 25 aligned verniers, i.e., verniers without an offset (**Figure 1**).

**FIGURE 1 | Step 1.** We presented a vernier without a mask (not shown). In a block of 80 trials, we presented the vernier with a certain duration and determined adaptively the offset size for which 75% correct responses occurred. In successive blocks, we reduced the vernier duration (VD) from 150 ms to the VD for which the vernier offset size was about 0.66- (arc min). These critical vernier durations were 43.2 ms for patients and 27.4 ms for controls on average. **Step 2.** The vernier was presented with the individual vernier duration of each observer, as determined in step 1. Next a blank screen, i.e., an inter-stimulus-interval (ISI), followed and then a masking grating. The vernier offset was fixed and we adaptively varied the ISI between vernier disappearance and the grating onset. We determined the SOA for which 75% correct responses occurred. We plot results as SOA = VD + ISI. Patients need much longer SOAs than controls. The main effect is due to the prolonged ISIs rather than the only slightly longer vernier durations. Reprinted with permission from Herzog et al. (2004).

Whereas we varied the vernier offset in the previous condition, here, we kept the vernier offset constant and varied the ISI between the vernier and the mask. We determined the critical ISI for which observers reached 75% correct responses. We found a tremendous deterioration of performance for the schizophrenia patients. With the 25 element grating mask, schizophrenia patients needed SOAs of about 150 ms whereas controls needed only about 30 ms (Herzog et al., 2004; Chkonia et al., 2010). With the 5 element grating mask, patients needed SOAs of about 240 ms while controls needed only about 80 ms (Herzog et al., 2004). Hence, schizophrenic patients show very strong visual deficits when their visual system is challenged both spatially and temporally. These results are well in agreement with previous studies on masking where patients showed strong masking deficits even when target duration was adjusted for each observer individually (e.g., Braff and Saccuzzo, 1981; Green et al., 1994a; Cadenhead et al., 1997).

#### **VISUAL BACKWARD MASKING DEFICITS ARE TARGET ENHANCEMENT DEFICITS**

In the next, third step, we used for each observer his or her individual vernier duration and the individual ISI of the 25 element mask, i.e., we used on average an SOA of 150 ms for the patients and an SOA of 30 ms for the controls. In this condition, we determined the vernier offset size adaptively, i.e., we determined the offset for which 75% correct responses occurred. Because of these normalized, individual values, performance was comparable across all observers for the 25 element grating, i.e., as aimed for.

Next, we removed 2 elements from the 25 element grating creating two gaps, thus, singling out a central grating with 5 elements from two peripheral gratings with 9 elements each (**Figure 2**). In healthy controls, we showed previously that this gap grating leads to much stronger masking than the homogenous 25 element grating (Herzog and Koch, 2001). We proposed that complex, spatial processing causes the deterioration of performance because nearby elements group together and ungroup from the peripheral elements. Our computer simulations showed how vernier related activity is dynamically suppressed with the gap grating but not with the homogeneous 25 element grating (Herzog et al., 2003).

The rationale of step 3 is as follows. If the gaps, for example, are blurred, performance should, paradoxically, be better than if vision is intact and the gaps are clearly processed. Hence, deterioration of performance with the gap grating indicates intact spatial processing. We found that, indeed, patients showed as strong performance deficits as controls (**Figure 2**). The very same holds true for temporal manipulations. We presented a 5 element grating for only 20 ms before the 25 element grating lasting for 280 ms. Because of the short duration of 20 ms, the 5 element grating is invisible. Still, it exerts strong performance deficits by a factor of about 5 compared to the 25 element grating. Schizophrenic patients show very similar deteriorations. Hence, it seems that spatial and temporal processing of the masking gratings is largely intact in schizophrenic patients.

In a series of further experiments, we supported these results. For example, we first deteriorated performance by adding small lines to the 25 element grating. Then we added additional horizontal lines which undid the deleterious effects of the vertical lines. Again, performance in patients and controls was rather similar (Schütze et al., 2007; Roinishvili et al., 2008). Hence, it seems that spatio-temporal processing is largely intact in the schizophrenic patients—after vernier duration and SOA were adjusted individually. For this reason, we propose that schizophrenic patients are just two steps from normal (in visual processing). There is a moderate deficit for the unmasked vernier, which could partially reflect long term suffering from the disease. Adolescents with schizophrenia do not show such deficits (Rund et al., 1996; Holzer et al., 2009; see also Saccuzzo et al., 1974). Such deficits may also likely occur when people are suffering, for example, from fever and other diseases affecting attention and concentration. To compensate for these

**FIGURE 2 | A vernier was followed by a variety of gratings (the vernier is not shown, only the masking gratings). (A)** Because we used for each observer his/her individual vernier duration and ISI, performance is roughly the same in patients and controls when the 25 element grating follows the vernier. Performance deteriorates significantly for the patients and healthy controls if either a "gap grating" or a "5–25" element grating follows the vernier. In the 5–25 grating, a 5 element grating is presented for 20 ms followed immediately by a 25 element grating for 280 ms. Because of the short duration, the 5 element grating is invisible. Performance deficits are similar in patients and controls indicating, paradoxically, intact spatial and temporal processing of the schizophrenic patients, i.e., patients seem to process the gaps and the briefly presented 5 element grating "carefully." Were the gaps fully blurred or the 20 ms 5 element grating smeared out,

deficits, we provided individually adjusted vernier durations in the masking conditions. Here, we found dramatically deteriorated performance which we attribute to a reduced target enhancement.

#### **MULTIFACTORIAL TARGET ENHANCEMENT**

It seems that visual processing of the masks is intact in schizophrenic patients. We propose what is deficient in schizophrenia is the processing of the *target* as a *target*, particularly, in demanding situations when, for example, targets are presented briefly, masked, or their contrast is low (e.g., Slaghuis, 2004). The target is the element of a visual scene which is task relevant. In many everyday situations, there is no target, for example, when observers just passively watch a scene or a movie. In the laboratory, many features or elements can be defined as the target for the very same stimulus. For example, we may have varied the vernier offset direction and, in addition, the length of the grating elements being a bit longer or shorter. In this example, we can ask observers to attend to the vernier offset or to the length of the grating elements. The stimuli are the same, just the tasks differ. Likewise, in a soccer game, you may attend to the actions of the players of team A, or B, or the behavior of other fans.

performance would be on the level of the homogeneous 25 element grating (#25). However, this is not the case. **(B)** (a) As in **(A)**, performance of patients and controls is roughly identical with the homogenous 25 element grating because we used the individual vernier durations and ISIs. (b) Performance strongly deteriorates by adding single collinear lines. (c) Horizontal contextual lines yield a performance level comparable to the standard condition (a). (c) Combining vertical and horizontal lines from the conditions (b) and (c) improves performance compared to (b). The horizontal contextual lines "counteract" the vertical lines. Again, it seems that patients "carefully" process the task irrelevant vertical lines, reflected in the strong performance deficits. This deterioration and the recovery from it by horizontal lines, is very similar to the one of controls. Reprinted with permission from Herzog et al. (2004) and Roinishvili et al. (2008).

Vernier offset discrimination is a demanding task when the vernier offset is small and its duration short. In this situation, neural responses are weak and it may be necessary to boost the corresponding weak neural signals to reach good performance. One option is to use recurrent processing (**Figure 3**). This can lead to strong and persistent neural signals even when the vernier is not presented anymore on the screen. It may be that such recurrent processing is deteriorated in schizophrenia. Because there is no mask, patients can reach the same performance level as controls, after longer recurrent processing. Hence, there are no strong performance deficits. When, however, a mask is presented, the mask may override and interrupt the recurrent vernier processing. To counteract, at least partially, the effects of such masks (and other adverse effects), the human brain is equipped with further mechanisms to enhance weak neural signals.

#### *Neuromodulation and attention*

We can only speculate on the mechanisms of target enhancement. Next to recurrent processing, one obvious candidate is attention. Attention deficits are core deficits in schizophrenia (e.g., Cornblatt and Keilp, 1994; Nestor and O'Donnell, 1998; Green, 2006). There are many types of attention. One important type

is selective visual attention. Visual attention can improve performance substantially. For example, in a classical Posner cueing paradigm, observers fixate a dot in the center of a computer screen. A square appears randomly either on the left or right of the fixation dot. Reaction times are much faster when the cue is presented just before the square at the same location, i.e., when attention can move to the target location before target presentation (Posner and Petersen, 1990). Reaction times are longer when there is no cue or when the cue appears to the "incorrect" side. EEG recordings show higher amplitudes when the target square is correctly cued compared to when not (Hillyard and Anllo-Vento, 1998). Similarly, monkey studies have shown that neural signals can be strongly enhanced when an element in a visual scene is attended compared to when it is not attended (Treue and Martínez-Trujillo, 1999; Maunsell and Cook, 2002). Importantly, only weak, low contrast stimuli benefit from attention (Treue and Maunsell, 1996; Reynolds and Heeger, 2009). We propose that a similar scenario applies to masking situations where target stimuli are weak because of brief presentation times (**Figure 3**).

responses may be present because of recurrent amplification. If a masking

Neuromodulation by the cholinergic system might be another mechanism for target enhancement (Cullum et al., 1993; Parasuraman and Greenwood, 2004; Deco and Thiele, 2009). The cholinergic system projects to layer IV of the primary visual cortex where retinal information first enters the visual cortex (Gil et al., 1997; Disney et al., 2007). It is often proposed that acetylcholine controls the influx of "external information" into the cortex. For example, when concentrating on a difficult mathematical problem in a noisy environment, acetylcholine may "shut down" disturbing visual signals, e.g., from a TV screen. In contrast, acetylcholine boosts weak but important information. In psychophysical experiments, targets are presented over and over again in a stereotyped manner and thus target occurrence is predictable, i.e., acetylcholine release can be initiated before target presentation (Sarter et al., 2009). Hence, a hypofunction of the cholinergic system may explain deteriorated vision in schizophrenia (**Figure 3**).

these may be dysfunctional in schizophrenic patients.

This proposal is in line with several studies showing that the nicotinic ACh system has significant impact on the mismatch negativity (MMN) (Knott et al., 2012) and sensory gating in schizophrenic patients (Freedman et al., 1997; Adler et al., 1998). In addition, deficits of schizophrenic patients in P50 gating could be improved by nicotine agonists (Adler et al., 2005; Zhang et al., 2012). Hence, a dysfunction of the cholinergic system may be involved in a large variety of dysfunctions including visual ones.

For simple tasks with large and high contrast stimuli, no target enhancement is needed. Accordingly, in these conditions performance is usually quite good in schizophrenic patients. When the verniers are not the target they are likely to go unnoticed. This makes sense because, in this case, vernier enhancement would just lead to amplification of a "disturbing" signal. Hence, target enhancement is activated only when an element is weak and the task-relevant target. Particularly at this stage, deficits of schizophrenia become evident.

However, at present, these considerations remain speculative. There may be more mechanisms and systems that can up- or down-regulate neural activity. In addition, the cholinergic system is a complex system with combinations of several sub-systems (nicotinic vs. muscarinic) which all (or some of them) may be altered in schizophrenia in various combinations. It is likely that not only one enhancing system is deficient in schizophrenia but certain combinations of them.

In addition, very little is known on the effects of acetycholine and attention on vision and how these systems relate to each other and to other systems and mechanisms, including recurrent and top-down neural activation. In addition, it is unclear to which extent acetylcholine is involved in attention itself. Whatever the neural mechanisms are, we propose that there are systems that can enhance targets and that some of them are dysfunctional in schizophrenia.

#### *Genetics*

Schizophrenic patients are usually heavy smokers (Dickerson et al., 2013) and smoking is often considered a type of selfmedication which is very much in accordance with deficits in the cholinergic system (Moran et al., 2012). In schizophrenia, the gene for the alpha 7 receptor subunit was found to be related to deficits in sensory gating (Raux et al., 2002; Houy et al., 2004; Martin et al., 2007). In addition, visual acuity was found to be low in CHRNA7 knockout mice (Origlia et al., 2012). We investigated five single nucleotide polymorphism (SNPs) of CHRNA7 and found that one SNP correlated well with the diagnosis of schizophrenia. Moreover, this SNP showed a high correlation with backward masking deficits in the shine-through effect (Bakanidze et al., 2009).

# **EEG**

Our model of target enhancement is corroborated by our EEG recordings (Plomp et al., 2013). We recorded brain activity with 64 electrodes and computed the global field power (GFP). The GFP reflects overall brain activity. We determined performance in four conditions (**Figure 4A**). First, only the vernier was presented. Second, the vernier was followed immediately by the 25 element grating (mean SOA of controls), or, third by an ISI and then by the grating (mean SOA of patients). Fourth, only the grating was presented (no preceding vernier; observers were not aware that there is no vernier because masking is so strong that the vernier is often invisible also when presented). The GFP of patients was much lower than the GFP of controls in the first three conditions where the vernier was presented (**Figure 4B**). As a speculation, we like to argue that the reduced GFP reflects a lack of target enhancement of the vernier. The reduced neural activity becomes behaviorally relevant only in the two masked conditions because, as we argue, recurrent processing is disturbed by the mask and, thus, target amplification is reduced. In the vernier only condition, longer recurrent processing may compensate the reduced neural activity (see Neuromodulation and Attention). Reduced EEG traces may also be caused by other factors such as diminished neural excitation. However, interestingly, EEG traces of patients and controls were very similar in the mask only condition arguing against a permanent deficit (**Figure 4B**).

# **DISCUSSION**

We view visual masking deficits in schizophrenia as a manifestation of a general deficit of target enhancement rather than as a specific visual deficit such as a dysfunction of the magnocellular system (see below). In this sense, visual dysfunctions in schizophrenia allow a *general* view into the basic "mechanisms of madness." We suggest that mechanisms similar to target enhancement may by deficient also in other visual and cognitive processes of schizophrenic patients.

For example, schizophrenic patients show strong deficits in the detection of low contrast stimuli (Slaghuis, 1998, 2004; Kéri et al., 2004; Calderone et al., 2013). It remains an open question whether these visual deficits are restricted to patients with negative symptoms (e.g., Slaghuis, 1998) and to certain stimuli, such as low spatial frequency gratings (Kéri et al., 2004). Delord et al. (2006) found that contrast detection thresholds of schizophrenic patients were significantly poorer than those of controls for all frequencies, arguing rather for a general visual deficit. Patients show also strongly deteriorated performance for high contrast stimuli when these stimuli are fragmented, i.e., where only parts of the contour of an object are visible. We like to argue that all these tasks require strong attention and some sort of task stabilization because information has to be integrated, for example, across contour elements in a time consuming manner.

In a similar fashion, we think that target enhancement is crucial in cognition. For the basic continuous performance test (CPT), performance is only slightly deteriorated in patients compared to controls (Nuechterlein et al., 1992). However, when stimuli are embedded in noise and, thus, contrast reduces, performance of patients deteriorates much more strongly than performance in controls. Effects are more pronounced when a memory component is added (Nuechterlein et al., 1992; Chkonia et al., 2010). We like to argue that these tasks, as the masking paradigm, require mechanisms to stabilize information across space and time, and that this ability is deteriorated in schizophrenic patients. The cholinergic system makes projections throughout the entire brain and can thus stabilize all sorts of processing including the ones mentioned above (Nelson et al., 2005; Deco and Thiele, 2009; Bentley et al., 2011; Yakel, 2013). The same holds true for attention.

Schizophrenic patients show also deficits in passive tests such as the MMN (Olincy and Freedman, 2012), prepulse inhibition (PPI), and P50 gating (Adler et al., 1991; Olincy et al., 2010). We will argue below that we do not propose that dysfunctional target enhancement is the only deficit in schizophrenic patients. To the contrary, we suggest that there are many deficits necessary to create the full blown symptoms of schizophrenia. Hence, MMN and P50 gating deficits may be of a different nature than masking deficits. In this sense, it is surprising that nicotinic agonists can compensate P50 deficits (Adler et al., 2005; Zhang et al., 2012) and MMN deficits (Knott et al., 2012). In addition, we like to

the second and third conditions, the 30 ms vernier was followed by a grating either immediately (short SOA) or after a blank screen of 120 ms (long SOA).

disorganization but not with high scores for the positive and negative dimensions (Cappe et al., 2012).

200 ms [reprinted with permission from Plomp et al. (2013)].

mention that schizophrenic patients show also immune deficits such as a diminished niacin skin response (Puri et al., 2001; Messamore, 2012). Also these deficits can, surprisingly, be linked to the cholinergic system (Gallowitsch-Puerta and Tracey, 2005). Hence, cholinergic deficits may be much more fundamental and not be related to target enhancement only.

Schizophrenic patients show often a strong degree of cognitive disorganization affecting performance in a large variety of everyday tasks. For this reason, cognitive disorganization is, next to positive and negative symptoms a key aspect in the psychopathology of schizophrenia. One century ago, Bleuler proposed that loose associations are even the basic symptoms of schizophrenia and can cause delusions and hallucinations (Bleuler, 1911). We found that persons with high scores of cognitive disorganization of schizotypy have problems focusing on tasks which may be related to stabilizing task related information. Accordingly, we found backward masking deficits in the shine-through paradigm in healthy psychology students with high scores on cognitive

We do not propose that attention and cholinergic dysfunctions are necessary and sufficient for schizophrenia. Likewise, we do not propose that target enhancement is necessary and sufficient for schizophrenia. Quite to the contrary, we believe that dysfunctional target enhancement is only one factor that can contribute to schizophrenia. First, there are patients without target enhancement deficits since not all patients have masking deficits. On the other hand, healthy student observers can have (moderate) masking deficits without an indication of the disease. Second, there can be many causes for dysfunctional neural enhancements. Attention and cholinergic modulation are just two examples. Within the cholinergic system, there may be many deficits leading to a similar phenotype since the cholinergic system itself is a complex system with two subsystems and various receptor types. In addition, there may be many mutations that all can lead to one dysfunction. For this reasons, we suggest that the study of endophenoytpes can be a successful tool to characterize subpopulations of schizophrenia. Visual masking is only one potential endophenotype.

#### **ENDOPHENOTYPE CONCEPT**

The shine-through masking paradigm is a potential endophenotype of schizophrenia. We found masking deficits not only in schizophrenic patients but also in their unaffected relatives (Chkonia et al., 2010). In addition, masking deficits were stable for 1 year (Chkonia et al., 2010) and adolescents with psychosis show masking deficits even before the manifestation of the disease (Holzer et al., 2009). Masking deficits are specific for the spectrum of functional psychoses (bipolar, schizoaffective, and schizophrenic patients) and were not found in depressed patients and abstinent alcoholics (Chkonia et al., 2012). We like to mention that visual backward masking can easily be controlled and has a much better signal to noise ratio than most cognitive tests (Chkonia et al., 2012).

#### **THE DYSFUNCTION HYPOTHESIS OF THE MAGNOCELLULAR SYSTEM**

Almost all current approaches on visual masking deficits in schizophrenia propose a dysfunction of the magno-cellular system based on the dual channel model of Breitmeyer and Ganz (1976). Enhanced masking in schizophrenic patients is attributed to either a hyperactive or hypoactive magno-cellular visual system (e.g., Green et al., 1994a,b; Schechter et al., 2003; Slaghuis, 2004; Butler et al., 2007). A full review is beyond the scope and goal of this article [for critical reviews, see Skottun and Skoyles (2007, 2009)]. We just like to mention that most masking research in schizophrenic patients investigated A-type masking whereas the dual channel model is mainly proposed to explain B-type masking (non-monotonic masking functions with strongest masking for SOAs of around 50 ms). Also the shine-through paradigm reveals A-type masking characteristics and stimuli are biased rather toward the parvo-cellular than the magnocellular system.

Our proposal shares many similarities with early models on visual masking deficits in schizophrenia where a key component is target storage in an iconic memory. The iconic memory protects target related activity to be overwritten by the mask. Masking deficits in schizophrenia patients were proposed to occur by a dysfunctional storage mechanism or slow information transfer to short term memory (Saccuzzo et al., 1974; Braff, 1981; Saccuzzo and Braff, 1981; Schwartz et al., 1983; for a review see Schuck and Lee, 1989).

## **SUMMARY**

We found strong deficits with the shine-through masking paradigm in schizophrenic patients (Herzog et al., 2004), in

#### **REFERENCES**


accordance with many other studies on visual masking (Saccuzzo et al., 1974; Braff, 1981; Saccuzzo and Braff, 1981; Green et al., 1994a,b). Also patients with other functional psychoses than schizophrenia show strong masking deficits; however, there are no deficits in unipolar depressive patients (Chkonia et al., 2012). The shine-through masking paradigm is a potential endophenotype of schizophrenia because, amongst other findings, relatives of schizophrenic patients show masking deficits (Chkonia et al., 2010), masking deficits occur before the onset of schizophrenia in adolescents with psychosis (Holzer et al., 2009), and in healthy students with high schizotypy scores (Cappe et al., 2012). On the neural level, we found reduced EEG traces (Plomp et al., 2013) and, on the genetic level, an abnormal SNPs related to the cholinergic system (Bakanidze et al., 2009).

We propose the following speculative model which closes the loop linking phenomenological findings (masking deficits) to genetic abnormalities which, in turn, cause deviant neural processing as evident in the EEG. First, visual processing *per se* is largely intact in schizophrenic patients as evidenced in good performance in visual acuity tests and the vernier task with long vernier durations. Even for short vernier durations, only mild deficits occur. We related these deteriorations partly to unspecific effects as they may occur from suffering from a chronic disease including severe medication effects. Second, there are pronounced masking deficits which we relate to specific deteriorations of target enhancement and stabilization. Third, we propose that a deficient cholinergic system is one out of possibly more deficient enhancement mechanisms. Heavy smoking of patients may be an attempt to compensate for the deficient cholinergic system. Fourth, reduced GFP in the EEG may be an indication of deficient target enhancement. Fifth, we view masking deficits as a manifestation of similar enhancement deficits as they may occur in cognition, emotion, and personality. Sixth, we do *not* propose that deficient target enhancement is necessary and sufficient for schizophrenia. To the contrary, there are schizophrenic patients without masking deficits. On the other hand, enhancement deficits are not sufficient for schizophrenia because there are healthy students with masking deficits. In addition, there are potentially many systems that contribute to information enhancement and likely a combination of these mechanisms has to be deviant to cause schizophrenia.

#### **ACKNOWLEDGMENTS**

This work was supported by the NCCR "Synapsy" of the Swiss National Science foundation (SNF). We like to thank Julia Veit for useful discussions and Christine Mohr for comments on the manuscript.

relationship of auditory gating defects to negative symptoms in schizophrenia. *Schizophr. Res.* 3, 131–138.

Bakanidze, G., Puls, I., Brand, A., Herzog, M. H., Roinishvili, M., and Chkonia, E. (2009). The influence of genetic variants of the CHRNA7 gene on backward masking in schizophrenic patients. *Eur. Arch. Psychiatry. Clin. Neurosci.* 259(Suppl. 1), S101.

Bentley, P., Driver, J., and Dolan, R. J. (2011). Cholinergic modulation of cognition: insights from human pharmacological functional neuroimaging. *Prog. Neurobiol*. 94, 360–388.


dysfunction in schizophrenia. *Schizophr. Res.* 10, 131–141.


shine-through. *Proc. Natl. Acad. Sci. U.S.A.* 98, 4271–4275.


Contextual suppression and protection in schizophrenic patients. *Eur. Arch. Psychiatry Clin. Neurosci.* 258, 210–216.


backward masking? *Int. J. Neurosci.* 119, 88–104.


schizophrenia. *Am. J. Psychiatry* 169, 974–981.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; accepted: 16 April 2013; published online: 14 May 2013.*

*Citation: Herzog MH, Roinishvili M, Chkonia E and Brand A (2013) Schizophrenia and visual backward masking: a general deficit of target enhancement. Front. Psychol. 4:254. doi: 10.3389/fpsyg.2013.00254*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Herzog, Roinishvili, Chkonia and Brand. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Sex, symptom, and premorbid social functioning associated with perceptual organization dysfunction in schizophrenia

# *Jamie Joseph1,2†, Grace Bae2 and Steven M. Silverstein2,3\**

*<sup>1</sup> Rutgers Biomedical and Health Sciences, Rutgers University, Graduate Program in Neuroscience, Piscataway, NJ, USA*

*<sup>2</sup> Rutgers Biomedical and Health Sciences, Rutgers University Behavioral Health Care, Division of Schizophrenia Research, Piscataway, NJ, USA*

*<sup>3</sup> Rutgers Biomedical and Health Sciences, Robert Wood Johnson Medical School, Department of Psychiatry, Piscataway, NJ, USA*

#### *Edited by:*

*Anne Giersch, Institut National de la Santé et de la Recherche Médicale, France*

#### *Reviewed by:*

*Wolfgang Tschacher, Universität Bern, Switzerland Bruce K. Christensen, McMaster University, Canada*

#### *\*Correspondence:*

*Steven M. Silverstein, Division of Schizophrenia Research, Rutgers Biomedical and Health Sciences, Department of Psychiatry, University Behavioral Health Care and Robert Wood Johnson Medical School, 671 Hoes Lane, Piscataway, NJ 08854, USA e-mail: silvers1@ubhc.rutgers.edu †This work is part of the doctoral*

*dissertation of the first author.*

Impairments in visual perceptual organization abilities are a repeatedly observed cognitive deficit in schizophrenia. These impairments have been found to be most prominent among patients with histories of poor premorbid social functioning, disorganized symptoms, and poor clinical outcomes. Despite the demonstration of significant sex differences for these clinical factors in schizophrenia, the extent of sex differences for visual perceptual organization in schizophrenia is unknown. Therefore, we investigated the extent to which previously known correlates (premorbid social sexual functioning and disorganized symptoms) and a novel factor (participant sex) accounted for performance on two perceptual organization tasks (contour integration and Ebbinghaus illusion) that have previously demonstrated sensitivity to schizophrenia. We also determined the relative degree to which each of these factors predicted task scores over and above the others. Schizophrenia patients (*N* = 109, 43 females) from different levels of care were ascertained. Female patients demonstrated higher contour integration scores, but lower performance on the context sensitivity index of the Ebbinghaus illusion, compared to males. Contour integration performance was significantly associated with poorer premorbid adolescent social sexual functioning and higher levels of disorganized symptoms, supporting past results that indicate a relationship among poor premorbid social sexual functioning, disorganized symptoms, and visual perceptual abnormalities in schizophrenia. However, analyses of Ebbinghaus illusion performance suggests there is a complex relationship among patient sex, clinical factors and perceptual abilities with relatively intact bottom–up grouping processes in females, but greater problems, compared to males with more top–down mediated context sensitivity. Therefore, sex differences may be an important consideration for future studies of visual perceptual organization in schizophrenia.

**Keywords: schizophrenia, visual perceptual organization, sex differences, premorbid social sexual functioning, disorganized symptoms**

# **INTRODUCTION**

Schizophrenia is a serious psychiatric condition characterized by significant cognitive and perceptual impairments. However, there is a great deal of heterogeneity across patients in terms of types of impairments, levels of severity across impairments (Heinrichs and Zakzanis, 1998; Heinrichs, 2001) and functional outcomes (Green, 1996; Silverstein et al., 1998; Green et al., 2000). One phenomenon that has shown potential for reliably identifying subgroups of patients is reduced visual perceptual organization. This is defined as the processes involved in binding stimulus features into meaningful patterns, groupings or object representations. Perceptual organization impairments have been proposed to be part of a widespread impairment in binding related features, and coordinating cognitive activity, across space and time, in schizophrenia (Phillips and Silverstein, 2003, 2013). Impaired cognitive coordination is also thought to be the core deficit in the disorganized syndrome of schizophrenia (Phillips and Silverstein, 2003), based on past findings of significant relationships between perceptual organizations deficits, and increased cognitive and/or behavioral disorganization but not positive or negative symptoms (Place and Gilmore, 1980; Knight, 1984; Uhlhaas and Silverstein, 2005; Silverstein and Keane, 2011).

Perceptual organization impairments may also be relevant for understanding heterogeneity in the developmental course of schizophrenia as they have been found in patients with histories of poor, but not those with good, premorbid social functioning. Poor premorbid social functioning has been found to increase the risk for the emergence of disorganized symptoms (Wickham et al., 2001; Schenkel et al., 2005) and both premorbid functioning and disorganized symptoms are related to poorer prognosis, which is also a clinical correlate of impaired perceptual organization (Silverstein et al., 1998, 2006; Uhlhaas et al., 2006b).

A limitation of many perceptual studies in schizophrenia is that few account for sex differences in relation to the significant heterogeneity. However, wide-ranging clinical findings among schizophrenia patients suggest that sex differences should be considered in studies of the disorder. A number of reports theorize an influence of sex in the development of schizophrenia pathogenesis (Goldstein et al., 1990; Seeman and Lang, 1990; Hoff and Kremen, 2002; Walker et al., 2002; Goldberg et al., 2011; Walder et al., 2012, 2013). Specifically, sex differences in schizophrenia have been associated with premorbid functioning (Goldberg et al., 2011; Ochoa et al., 2012), age of onset (Walker et al., 2002; Goldberg et al., 2011; Zhang et al., 2012), symptomatology (Walker et al., 2002; Zhang et al., 2012), relapse rate (Ochoa et al., 2012), cognitive ability (Goldstein et al., 1998; Weiser et al., 2000; Hoff and Kremen, 2002), treatment response (Angermeyer et al., 1990; Buchsbaum, 1992; Buchsbaum et al., 1992; Carpiniello et al., 2012) and substance abuse (Mahoney et al., 2010).

Although perceptual organization abnormalities have been linked to factors that are suggested to vary by sex in schizophrenia (e.g., premorbid social functioning, treatment response), there have been no reports of sex differences in perceptual organization impairments in schizophrenia. There is reason to believe that sex differences might exist, since sex differences have been reported in non-clinical samples on tests of perceptual organization (Phillips et al., 2004). Therefore, the goal of this study was to determine whether sex differences on tests of perceptual organization exist in a sample of schizophrenia patients, and the extent to which these are related to other aspects of heterogeneity that have been previously linked to both variables (e.g., premorbid social functioning, disorganized symptoms).

# **MATERIALS AND METHODS PARTICIPANTS**

The study was approved by the University of Medicine and Dentistry (UMDNJ)—Robert Wood Johnson Medical School Institutional Review Board and written consent was obtained for all study participants. The study sample ascertained consisted of 66 male and 43 female patients who met DSM IV-TR (American Psychiatric Association, 2000) criteria for schizophrenia or schizoaffective disorder. Individuals with current substance abuse, mental retardation, neurological disorders, or other primary psychiatric disorders were excluded from the study. To avoid confounds due to moderate attentional deficits, patients with poor catch trial performance, reflecting significant inattention to the tasks (*n* = 22 see Section Data Analyses), were excluded. Included patients (52 males, 35 females) ranged in age from 24 to 64 (*M* = 46*.*62, *SD* = 10*.*70).

Patients were recruited from three levels of care within a vertically integrated system (Smith et al., 1999): (1) acute partial hospital (most recent inpatient discharge was within the past 6 months and includes full daily structure and treatment); (2) extended partial hospital (where the last inpatient discharge occurred over 6 months ago, but daily treatment and structure are still required); and (3) outpatient (where last inpatient discharge was over 2 years ago and patients are considered to be clinically stable requiring bi-weekly or monthly visits to psychiatric care providers).

# **CLINICAL ASSESSMENTS**

Psychiatric diagnoses for all participants were determined with the Diagnostic Interview for Genetic Studies (DIGS) (Nurnberger et al., 1994) and confirmed via review of UMDNJ-UBHC medical records in order to determine study eligibility. Symptoms occurring 2 weeks prior to testing were assessed using the Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987). Syndromes were then analyzed based on a five factor model (Lindenmayer et al., 1994) which includes positive, negative, cognitive, excitement, and depression factors.

As disorganization has been shown to be a strong correlate of both perceptual disorganization and premorbid social functioning, disorganized symptoms were characterized in two ways. The first method is similar to a strategy used in one of our previous studies where significant differences in perceptual organization ability were observed when patients were dichotomized, based on the PANSS conceptual disorganization item score (Uhlhaas et al., 2006a), into either a disorganized or non-disorganized group. For the present study, participants with a score of 2 or lower (i.e., within the normal range) were considered the nondisorganized schizophrenia group (*n* = 32) and participants with a score of 3 (mild symptoms) or greater comprised the disorganized schizophrenia group (*n* = 53). These groups were then compared using *t*-tests. For all other analyses, disorganized symptoms were analyzed based on a separate disorganization factor identified by Cuesta and Peralta (1995) that is not part of the original PANSS. This factor includes the PANSS items of conceptual disorganization, and poor attention, as well as the inappropriate affect item developed by Cuesta and Peralta.

Psychosocial adjustment and academic development were evaluated using the Premorbid Adjustment Scale (PAS) (Cannon-Spoor et al., 1982). Each item was scored on a scale of 0 (good) to 6 (poor) from early childhood until 1 year prior to the onset of first psychotic symptoms, or 1 year prior to first psychiatric hospitalization if exact age of onset could not be determined. Scores on 5 domains of functioning were calculated: social withdrawal, peer relationships, scholastic performance, school adaptation, socialsexual aspects of life and an overall mean score (Cannon-Spoor et al., 1982). Much prior literature has assessed the association of perceptual deficits and premorbid functioning by using the Phillips or Zigler-Phillips scale of premorbid adjustment (Zigler, 1961). These scales primarily comprise questions about friendships, dating, and marriage. They correspond closely to the premorbid social sexual functioning factor of the PAS, a more current and widely used scale. Because: (1) our primary interest was in premorbid social-sexual functioning, which we have assessed in past studies and which has been found to correlate inversely with perceptual organization ability; and (2) there are no data to suggest that other aspects of premorbid functioning (e.g., academic functioning) would or should be related to perceptual organization, only scores on the social sexual subscale of the PAS were prioritized for analyses with perceptual indices.

Study participants were taking atypical and/or typical antipsychotic medications with stable medication dosages. Antipsychotic medication dosages were converted to chlorpromazine equivalents based on published standards (Andreasen et al., 2010). Recent and typical tobacco use for was assessed with the Fagerström Test for Nicotine Dependence (Heatherton et al., 1991) to determine the current effects of nicotine use or dependence on perceptual task performance.

The vocabulary subtest of the Shipley Institute of Living Scale (Zachary, 1991) was administered to participants in order to estimate IQ by converting raw scores into a WAIS-R IQ score. A Snellen chart was used to ascertain visual acuity estimates for each eye and then both eyes. Handedness was ascertained with a questionnaire about hand preference for different daily tasks including: writing, throwing, using scissors, etc. Participants were noted as right handed or left handed based on these responses.

# **PERCEPTUAL ORGANIZATION TASKS**

#### *Contour integration-JOVI task*

The Jittered Orientation Visual Integration (JOVI) task is a test of contour integration that determines a participant's ability to integrate Gabor elements into a perceptual whole. Gabor elements are sinusoidal luminance distributions that are Gaussian modulated (Silverstein et al., 2000; Uhlhaas et al., 2005). That is, Gabor elements show lower contrast at the edges compared to the center, and luminance varies from white to black in a gradually alternating fashion (**Figure 1**). Gabor elements are considered to activate orientation-selective feature detectors in the primary visual cortex (Angelucci and Bullier, 2003), and are therefore a useful means to study their integration in early visual processing.

The stimuli presented for this task are based on our recent study of contour integration in schizophrenia (Silverstein et al., 2012b). Participants were shown static Gabor elements arranged in an oblong shape forming a contour embedded in a display of randomly oriented Gabor elements. The degree of orientation jitter of contour Gabor elements varied across six conditions (±0◦, 7–8◦, 9–10◦, 11–12◦, 13–14◦, 15–16◦), and this manipulation placed increasing degrees of burden on perceptual organization processes (i.e., at higher jitter levels, the correlations between adjacent element orientations become weaker, contour smoothness is increasingly disrupted, and contour perception becomes more difficult). For all stimuli, the ratio of the density of adjacent

the JOVI task. The top **top left** panel of the figure is an example of a lower jitter degree condition presented to participants (7–8◦ ). The **top right** panel of the figure shows the highest jitter degree presented (15–16◦ ). The **bottom left** and **right** panels show the catch stimuli included in each block to account for participant attention level (see text for description of these stimuli).

background elements to the density of adjacent contour elements was 0.9. Because adjacent background elements were, on average, closer together than adjacent contour elements, perceptual organization processes used to process contours are independent of density cues for this task (Silverstein et al., 2000).

All JOVI stimuli trials were presented for 2 s followed by a 1 s inter stimulus interval during which responses were no longer recorded. There were 48 stimulus trials per jitter condition which were presented in blocks of 12 trials by condition. In addition, two types of catch stimuli (no errors expected) using 0◦ jitter were administered during each block to assess for attention lapses. One catch trial type had curved lines drawn through the contours (to eliminate the need for perceptual organization), and the other contained contour elements without any background elements (to remove effects of noise). The JOVI is a symmetric 1-alternative forced choice task where subjects responded whether the narrow end of the oblong contour was pointing left or right (**Figure 1**). The task and stimuli patterns were created using E-prime (Psychology Software Tools, Pittsburgh, PA).

# *Ebbinghaus illusion task*

than the right center circle.

The Ebbinghaus illusion assesses integration of non-target information during target perception, and the resulting change in size perception of the target is thought to result from size constancy (Doherty et al., 2008, 2010). In this task (**Figure 2**), the perceived size of center circles is modified by the presence of outer context circles: typically, larger context circles make the center circle appear smaller than its actual size, while smaller context circles lead to an enlarged appearance of the center circle (Uhlhaas et al., 2006a).

The stimuli used for this task were developed by Phillips et al. (2004), Doherty et al. (2010). Participants were shown two black

circles presented on a white background with one center circle 100 pixels in diameter and the second center circle varying by 2, 6, 10, 14, or 18 pixels (This corresponded to one center circle having 2.67◦ of visual angle in diameter, with the other center circle 0.05, 0.16, 0.27, 0.37, or 0.48◦ larger or smaller). These center circles were shown with three different surrounding contexts: no outer context, misleading context (outer context circles should impair accurate inner circle size discrimination) and helpful context (outer context should aid accurate inner circle size discrimination). The no context condition had 96 trials: 32 at the smallest size difference and 16 at all other size differences. The misleading context condition had 80 trials: 16 at each level of inner circle size difference. The Helpful context condition had 16 trials, all at a 2 pixel difference. Stimuli were presented for 2 s with 200 ms inter stimulus interval, in a random order. Subjects responded whether the left center circle or right center circle was larger (see **Figure 2**). The task and stimuli were created using C++.

#### **DATA ANALYSES**

All data were analyzed using IBM SPSS Statistics 19. The demographic, clinical and perceptual data were compared between male and female participants using independent samples *t*-tests or χ<sup>2</sup> square tests to analyze categorical variables (**Table 1**). Spearman correlations were performed to examine the associations among demographic factors, PAS factors, PANSS factors and perceptual scores (**Tables 2**–**4**).

For multiple regression analyses, in cases where correlational analyses identified significant predictors of task performance that were predicted *a priori* (e.g., poor premorbid social functioning, disorganized symptoms), hierarchical regression analyses were employed in order to determine the extent to which each significant predictor accounted for variance in the dependent variable over and above that of other predictors, in addition to which variable alone accounted for the greatest proportion of variance in test score. In cases where correlational analyses indicated significant predictors that were not predicted *a priori*, stepwise regression was used to determine the descending order in which these predictors accounted for variance in the dependent variable.

For the contour integration (JOVI) task, the mean score across all jitter conditions was used as the performance index since a previous study suggested higher test-retest reliability for the overall mean score compared to threshold values (Silverstein et al., 2012a). Study participants with JOVI catch trial scores below 90% were excluded from all data analyses (*n* = 22). For the Ebbinghaus illusion task, a difference score index was computed: [(Helpful Context 2 Pixel Condition-No Outer Context 2 Pixel Condition)—(Misleading Context 2 Pixel Condition-No Outer Context 2 Pixel Condition)], which reflected overall context sensitivity.

#### **RESULTS**

#### **DEMOGRAPHIC CLINICAL AND PERCEPTUAL DIFFERENCES BASED ON PARTICIPANT SEX**

Female and male participant demographic, PAS, PANSS and perceptual task score differences are shown in **Table 1**.



<sup>χ</sup>*chi square test; \*p < 0.05; \*\*\*p < 0.001;* <sup>ϕ</sup>*phi correlation coefficient.*



*\*p < 0.05; \*\*p < 0.01.*

**Table 3 | Spearman correlations of PANSS factors and perceptual organization task indices.**


*\*p < 0.05; \*\*p < 0.01; \*\*\*p < 0.001.*

#### **DEMOGRAPHIC FACTORS AND PERCEPTUAL TASK PERFORMANCE**

The correlations between demographic factors and perceptual task indices are shown in **Table 2**.

#### **PANSS SYMPTOM FACTORS AND PERCEPTUAL TASK PERFORMANCE**

Spearman correlations between PANSS symptom factor scores and perceptual task performance are listed in **Table 3**. In terms of the dichotomy between disorganized and non-disorganized patients; only Total JOVI scores [non-disorganized group: *M* = 182*.*88, *SD* = 36*.*00, disorganized group: *M* = 162*.*07, *SD* = 38*.*67; *t(*83*)* = 2*.*47, *p* = 0*.*016, *d* = 0*.*56] and not



*\*p < 0.05.*

Ebbinghaus index scores [non-disorganized group: *M* = 6*.*06, *SD* = 11*.*88; disorganized group: *M* = 5*.*17, *SD* = 11*.*95; *t(*84*)* = 0*.*34, *p* = 0*.*737, *d* = 0*.*07] were significantly different.

#### **PAS FACTORS AND PERCEPTUAL TASK PERFORMANCE**

Spearman correlations between PANSS symptom factor scores and perceptual task performance are listed in **Table 4**.

#### **INTER TASK CORRELATIONS ARE NOT SIGNIFICANT**

The correlation between the JOVI and Ebbinghaus performance indices was not statistically significant: *r(*87*)* = 0*.*024, *p* = 0*.*822.

#### **HIERARCHICAL REGRESSION EXAMINING SIGNIFICANT CORRELATES OF CONTOUR INTEGRATION PERFORMANCE**

The set of factors that were significantly correlated with lower JOVI scores included: male sex, poor premorbid adolescent social sexual functioning, higher PANSS cognitive factor scores and higher Cuesta and Peralta disorganized factor scores. Two of these factors, poor premorbid social functioning and disorganized symptoms, were previously associated with contour integration performance (Uhlhaas et al., 2006a). The PANSS cognitive factor was excluded to avoid multicollinearity (i.e., *r >* 0*.*795) due to a strong correlation with disorganized factor symptoms *r(*86*)* = 0*.*887, *p <* 0*.*001.

Hierarchical regression analyses were conducted to answer 3 distinct questions: (1) which variable, alone, accounted for the most variance in JOVI score; (2) which variable, after all other variables were entered at prior steps, accounted for the most additional variance in JOVI score; and (3) given the study's focus on sex differences, to what extent did sex differences account for additional variance in JOVI score after all other variables were entered at a prior step. The factor that, alone, explained the most variance in JOVI performance was disorganized symptoms: *<sup>B</sup>* = −5*.*37, *SE B* <sup>=</sup> <sup>1</sup>*.*89, <sup>β</sup> = −0*.*30, *<sup>R</sup>*<sup>2</sup> <sup>=</sup> 0*.*092, *F* = 8*.*01, *p* = 0*.*006. While participant sex (*B* = −12*.*19, *SE B* = 9*.*27, β = −0*.*15, *t* = −1*.*32, *p* = 0*.*192) premorbid adolescent social sexual functioning (*B* = −2*.*98, *SE B* = 2*.*43, β = −0*.*14, *t* = −1*.*23, *p* = 0*.*224) and disorganized symptom factor scores (*B* = −4*.*32, *SE B* = 1*.*91, β = −0*.*24, *t* = −2*.*26, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*026) contributed, as a set, to total JOVI score (*R*<sup>2</sup> <sup>=</sup> 0*.*148, *F* = 4*.*46, *p* = 0*.*006), disorganized factor scores still accounted for unique variance over and above these other variables: *<sup>R</sup>*<sup>2</sup> *change* <sup>=</sup> <sup>0</sup>*.*055, *<sup>F</sup>(*1*,* <sup>77</sup>*)* <sup>=</sup> <sup>4</sup>*.*96, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*029, and this was the largest increase of one variable over and above any of the others. Examination of the effect of sex (entered at step 2) over and above other variables (entered as a set at step 1) indicated a non-significant additive effect: *<sup>R</sup>*<sup>2</sup> *change* <sup>=</sup> <sup>0</sup>*.*019, *F* = 1*.*73, *p* = 0*.*192 (*B*, *SE B*, and β-values are the same as above).

#### **STEPWISE REGRESSION EXAMINING SIGNIFICANT CORRELATES OF EBBINGHAUS ILLUSION PERFORMANCE**

The set of factors that were correlated with Ebbinghaus illusion performance were patient sex, PANSS negative symptom factor score, Shipley subtest score, premorbid social sexual adolescent functioning, premorbid scholastic factor performance, premorbid school adaptation factor performance, age of onset and current smoking status. Since, except for the PAS social sexual functioning variable, there were no *a priori* hypotheses as to how these factors would contribute to variance in Ebbinghaus illusion performance, a stepwise regression was performed. The factors were entered, by the stepwise regression algorithm, in the following order: sex, age of onset, premorbid social sexual adolescent functioning, PANSS negative symptom factor, PAS school adaptation factor, PAS scholastic factor and current smoking status. The stepwise analysis indicated that the combination of these factors accounted for significant variance in the Ebbinghaus performance index: *<sup>R</sup>*<sup>2</sup> <sup>=</sup> <sup>0</sup>*.*33, *<sup>F</sup>* <sup>=</sup> <sup>9</sup>*.*04, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001. Participant sex alone made a significant contribution to variance in Ebbinghaus performance: *<sup>B</sup>* = −9*.*11, *SE B* <sup>=</sup> <sup>2</sup>*.*56, <sup>β</sup> = −0*.*37, *<sup>R</sup>*<sup>2</sup> <sup>=</sup> <sup>0</sup>*.*14, *F* = 12*.*26, *p* = 0*.*001. In addition, even after all of the other statistically significant predictor variables were entered as a set in block 1 of a second regression analysis, sex continued to be a significant predictor when entered alone in block 2: *<sup>B</sup>* <sup>=</sup> <sup>9</sup>*.*34, *SE B* <sup>=</sup> <sup>2</sup>*.*59, <sup>β</sup> <sup>=</sup> <sup>0</sup>*.*38, *<sup>R</sup>*<sup>2</sup> *change* <sup>=</sup> <sup>0</sup>*.*11, *<sup>F</sup>* <sup>=</sup> <sup>12</sup>*.*97, *p <* 0*.*001.

# **DISCUSSION**

This study replicated previously reported findings that impaired contour integration was associated with poorer premorbid social functioning and increased disorganization symptoms (Silverstein et al., 2000; Schenkel et al., 2005; Uhlhaas et al., 2005). In the current study, female gender was associated with higher contour integration scores compared to males in a chronic schizophrenia population that included many patients who were sufficiently disabled so as to require daily partial hospital treatment. However, our hierarchical regression results indicated that level of clinical disorganization was a better predictor of impaired contour integration performance than participant sex or premorbid social sexual adolescent functioning. Of note, using a 70% JOVI catch trial cutoff as an exclusion criterion, as opposed to 90%, premorbid social sexual functioning was the stronger predictor. This suggests that the sex differences observed on contour integration performance are mostly in agreement with the hypothesis of a greater neurodevelopmental basis in males with schizophrenia (Goldstein et al., 1998), and with gender differences in premorbid functioning (Goldberg et al., 2011; Ochoa et al., 2012), and neurocognitive task performance (Hoff and Kremen, 2002; Johnson et al., 2005) which tends to be poorer in males. The slight difference in results between analyses using a 70 vs. 90% threshold for JOVI catch trial performance (i.e., 30 vs. 10% attention lapse errors) may also account for the very recent finding of a lack of a sex difference among schizophrenia patients on the JOVI (Strauss et al., 2013). The sample in the Strauss et al. study consisted mostly of clinically stable outpatients who would be expected to have relatively few problems with attention lapses (i.e., they would be more similar to our patients who had fewer than 10% lapse errors than to those with up to 30% lapse errors). Therefore, it may be that sex differences are stronger when more severely ill patients are included in the sample and visual processing impairment is expressed to a greater degree, suggesting an interaction between sex, illness phase, and perception that is in need of further examination.

Factors such as premorbid social functioning and disorganized symptoms have been shown to predict a poor outcome in schizophrenia spectrum populations (Zigler, 1961; Strauss and Carpenter, 1972; Knight et al., 1979; Bearden et al., 2011). Whether these relationships reflect a specific patient subtype or simply strong relationships between selected variables remains an open question. However, our data support past hypotheses that a group of patients characterized by these factors and abnormal perceptual organization exists, and this suggests the importance of including perceptual, developmental, symptom, and gender variables in future studies to characterize heterogeneity in perceptual functioning, and perhaps other aspects of cognition, in schizophrenia.

The gender differences on Ebbinghaus illusion task performance have not been previously reported. Moreover, unlike in our past studies with this illusion, task performance was not related to disorganization symptoms. One possible explanation for the different findings is that acutely ill patients were not included in the study, and some perceptual functions are state sensitive in schizophrenia (e.g., Uhlhaas et al., 2005), including performance on the Ebbinghaus illusion (Silverstein et al., 2013). Therefore, it is possible that our outpatient and partial hospital sample did not have sufficient range in Ebbinghaus scores to detect relationships observed in past studies, which included relatively large numbers of inpatients, or wholly inpatient samples. Future studies of the Ebbinghaus illusion in schizophrenia can clarify the extent to which gender is related to task performance as a function of phase of illness by including acutely ill inpatient samples.

The non-significant correlation between the contour integration task and the Ebbinghaus illusion may indicate that these tasks assess different stages of perceptual organization in schizophrenia. A possible interpretation of these findings is that contour integration reflects relatively more bottom–up than top–down contributions to perception [i.e., more stimulus-driven and lower level aspects of perception, including basic binding processes in V1–V4 (Silverstein et al., 2009)], whereas the Ebbinghaus illusion, by incorporating both grouping and size constancy principles, involves a relatively greater influence of top–down factors including memory and past experience interpreting size based on 2D depth cues (Doherty et al., 2010) [see Purves and Lotto, 2011 for an account of the illusion consistent with this hypothesis, and Dima et al. (2009) for data indicating reduced top-down involvement in another task involving experience-based modulation of perception in schizophrenia]. It is possible that the different mechanisms involved in performance on each task can account to some degree for the differences in gender differences in patient performance across the two tasks. However, at present, there is no obvious explanation for why females with schizophrenia would be superior to males in lower-level vision, while males would be more normal on a higher-level visual task. It is also important to note that the direction of gender differences in our study on the Ebbinghaus illusion, with females with schizophrenia being less context sensitive than males, is in contrast to a previously reported finding in a non-psychiatric population (Phillips et al., 2004). Therefore, this suggests a possible interaction of gender, Ebbinghaus illusion task performance and schizophrenia spectrum diagnosis that requires further exploration.

This study has two significant limitations. The first is that multiple comparisons were not controlled for, and therefore the results should be considered as preliminary and require replication. Another limitation is that psychiatric and non-psychiatric control subjects were not included in the study. The tasks used in this study have been used in other studies that included patient and non-psychiatric controls and these groups have shown significantly superior performance compared to schizophrenia, suggesting specificity for the latter (Silverstein and Keane, 2011). However, the inclusion of non-psychiatric controls and other psychiatric populations in future studies will be necessary to

#### **REFERENCES**


outcome in youth at clinical high risk for psychosis. *J. Am. Acad. Child Adolesc. Psychiatry* 50, 669–680. doi: 10.1016/j.jaac.2011.03.021


determine whether the correlates of perceptual organization performance are specific to schizophrenia.

The results of our study appear to be relevant for future studies assessing gender, premorbid function and perceptual organization in prodromal, at risk, and first episode populations. Since the perceptual organization tasks employed are considered to have well-defined neurodevelopmental trajectories (Kovacs et al., 1999; Kaldy and Kovacs, 2003; Doherty et al., 2010), it would be useful in future studies to compare and consider perceptual organization in early vs. later stages of schizophrenia to determine if impairments are primarily accounted for by altered maturation of perceptual organization circuitry or illness-related perceptual organization deterioration.

Overall, the present study's findings for contour integration performance support those of earlier smaller studies in indicating that impairment at an early stage of perceptual organization is associated with poorer premorbid social functioning and a tendency toward development of disorganized symptoms. Because other studies indicate that patients with these characteristics also have a poorer response to treatment and a more chronic illness course (Silverstein et al., 1998; Uhlhaas et al., 2006b), the development of greater insight into this group, and the development of new treatment strategies for it, is especially important.

# **ACKNOWLEDGMENTS**

We thank Dr. Bill Phillips and Dr. Martin Doherty for providing us with the software for the Ebbinghaus illusion task. We thank UMDNJ-UBHC staff and patients for their participation in and cooperation with this research. This research was partially supported by an American Psychiatric Foundation award to Dr. Steven M. Silverstein.

cohort study. *Schizophr. Res. Treat.* 2012:576369. doi: 10.1155/2012/ 576369


M. J., Vossen, H., et al. (2011). Neurodevelopmental liability to schizophrenia: the complex mediating role of age at onset and premorbid adjustment. *Schizophr. Res.* 133, 143–149. doi: 10.1016/j.schres.2011.09.014


stuff." *Schizophr. Bull.* 26, 119–136. doi: 10.1093/oxfordjournals.schbul.a033430


in schizophrenia gender differences. *Schizophr. Bull.* 16, 185–194. doi: 10.1093/schbul/16.2.185


Psychiatric Publishing Inc), 193–214.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 05 February 2013; accepted: 02 August 2013; published online: 27 August 2013.*

*Citation: Joseph J, Bae G and Silverstein SM (2013) Sex, symptom, and premorbid social functioning associated with perceptual organization dysfunction in schizophrenia. Front. Psychol. 4:547. doi: 10.3389/fpsyg.2013.00547*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Joseph, Bae and Silverstein. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# An event-related potential examination of contour integration deficits in schizophrenia

# *Pamela D. Butler 1,2,3\*, Ilana Y. Abeles 1,2,3, Steven M. Silverstein4,5, Elisa C. Dias 1,2, Nicole G. Weiskopf 1,6, Daniel J. Calderone1,2,3 and Pejman Sehatpour 1,7*

*<sup>1</sup> Schizophrenia Research Division, Nathan S. Kline Institute for Psychiatric Research, Orangeburg, NY, USA*

*<sup>2</sup> Department of Psychiatry, NYU School of Medicine, New York, NY, USA*

*<sup>3</sup> Department of Psychology, City University of New York, New York, NY, USA*

*<sup>4</sup> Division of Schizophrenia Research, University of Medicine and Dentistry of New Jersey - University Behavioral HealthCare, Piscataway, NJ, USA*

*<sup>5</sup> Department of Psychiatry, University of Medicine and Dentistry of New Jersey - Robert Wood Johnson Medical School, Piscataway, NJ, USA*

*<sup>6</sup> Department of Biomedical Informatics, Columbia University, New York, NY, USA*

*<sup>7</sup> Department of Psychiatry, Columbia University College of Physicians and Surgeons, New York, NY, USA*

#### *Edited by:*

*Michael Green, University of California, Los Angeles, USA*

#### *Reviewed by:*

*Jeffrey Bedwell, University of Central Florida, USA Jonathan K. Wynn, University of California, Los Angeles, USA*

#### *\*Correspondence:*

*Pamela D. Butler, Schizophrenia Research Division, Nathan S. Kline Institute for Psychiatric Research, 140 Old Orangeburg Rd., Orangeburg, NY 10962, USA. e-mail: butler@nki.rfmh.org*

Perceptual organization, which refers to the ability to integrate fragments of stimuli to form a representation of a whole edge, part, or object, is impaired in schizophrenia. A contour integration paradigm, involving detection of a set of Gabor patches forming an oval contour pointing to the right or left embedded in a field of randomly oriented Gabors, has been developed for use in clinical trials of schizophrenia. The purpose of the present study was to assess contributions of early and later stages of processing to deficits in contour integration, as well as to develop an event-related potential (ERP) analog of this task. Twenty-one patients with schizophrenia and 28 controls participated. The Gabor elements forming the contours were given a low or high degree of orientational jitter, making it either easy or difficult to identify the direction in which the contour was pointing. ERP results showed greater negative peaks at ∼165 (N1 component) and ∼270 ms for the low-jitter versus the high-jitter contours, with a much greater difference between jitter conditions at 270 ms. This later ERP component was previously termed Ncl for closure negativity. Source localization identified the Ncl in the lateral occipital object recognition area. Patients showed a significant decrease in the Ncl, but not N1, compared to controls, and this was associated with impaired behavioral ability to identify contours. In addition, an earlier negative peak was found at ∼120 ms (termed N120) that differentiated jitter conditions, had a dorsal stream source, and differed between patients and controls. Patients also showed a deficit in the dorsal stream sensory P1 component. These results are in accord with impairments in distributed circuitry contributing to perceptual organization deficits and provide an ERP analog to the behavioral contour integration task.

**Keywords: schizophrenia, perception, cognition, contour integration, electrophysiology, vision**

# **INTRODUCTION**

Visual integration, also referred to as "perceptual organization," is impaired in schizophrenia (Silverstein and Keane, 2011). Visual integration is defined as the processes linking the output of neurons, which individually code local (typically small) attributes of a scene, into a global (typically larger) complex structure more suitable for guidance of behavior (Butler et al., 2008). Integration is important for gestalt grouping and object recognition. The importance of visual integration impairments is underscored by inclusion of this domain as a core construct in the Cognitive Neuroscience Treatment Research to Improve Cognition in Schizophrenia (CNTRICS) initiative (Green et al., 2009; Butler et al., 2012; Silverstein et al., 2012).

Visual integration deficits are seen on a number of tasks in schizophrenia including contour integration (Silverstein et al., 2000, 2006, 2012; Uhlhaas and Silverstein, 2005; Kozma-Wiebe et al., 2006; Silverstein and Keane, 2011), coherent motion (Chen, 2011), object recognition from fragmented line drawings (Doniger et al., 2002; Sehatpour et al., 2010), grouping according to proximity or color similarity (Kurylo et al., 2007), and configural processing of faces (Silverstein et al., 2010). A salient aspect of these studies is that they do not appear to be due to a "general deficit," in that patients perform more accurately than controls when the task relies on judgments about individual features or when grouping interferes with isolating or processing single features (Place and Gilmore, 1980; Silverstein and Keane, 2011).

The present paper focuses on a visual integration paradigm that has been widely used in schizophrenia—contour integration (Field et al., 1993; Silverstein et al., 2000, 2006, 2012; Uhlhaas and Silverstein, 2005; Kozma-Wiebe et al., 2006; Silverstein and Keane, 2011). This task involves viewing Gabor patches that are arranged to form an oval that points to the right or left within a field of randomly-oriented noise Gabor patches (**Figure 1**). The Gabor high-jitter stimuli.

signals roughly model the receptive field properties of cells in the primary visual cortex (V1). Importantly, the embedded figure cannot be detected by purely local filters or by neurons with large receptive fields corresponding to the size of the contour (Dakin and Hess, 1998). Thus, the contours are thought to be detected by relying on long-range connections and/or reentrant feedback from V2 or higher areas to produce grouping, and to enhance representation of global shape, especially in the presence of noise (Silverstein and Keane, 2011). There is also some, albeit more inconsistent, evidence for bottom-up contributions to perceptual integration deficits in schizophrenia from studies of contour linking over short distances (Spencer et al., 2003; Keri et al., 2005). Indeed, an fMRI study of contour integration found impairment in a distributed network that includes occipital areas as well as prefrontal, parietal, and ventral temporal areas in patients with schizophrenia (Silverstein et al., 2009).

There is, however, a paucity of information regarding the contributions of early and later stages of visual processing to contour integration in controls and to their impairments in patients with schizophrenia. Event-related potentials (ERPs) are ideally suited to assess different stages of processing due to their high temporal resolution. Previous ERP studies have looked at visual integration in other paradigms such as perceptual closure, which involves identification of fragmented objects (Doniger et al., 2002; Sehatpour et al., 2010). Patients with schizophrenia showed impairment of early-stage sensory processing in this task, as seen by decreased amplitude of dorsal visual stream P1, which occurs at ∼100 ms. There is extensive cross-connectivity between brain regions including dorsal and ventral stream areas (Rosa et al., 2009). Information conveyed by the dorsal stream P1 contributes to later stages of processing in the ventral stream lateral occipital complex (LOC) involved in perceptual closure. The ERP signature of perceptual closure, which has been termed closure negativity (Ncl), occurs at ∼300 ms, is impaired in schizophrenia, and has been found to have a source in LOC (Doniger et al., 2002; Sehatpour et al., 2010).

An ERP study, to our knowledge, has not been previously carried out to examine contour integration as described here. The present study assessed the P1, N1, and the closure potential Ncl in a contour integration task. While the P1 and N1 occur close in time to each other, peaking at ∼100 and ∼170 ms, respectively, they reflect very different processes. The P1 has dorsal and ventral stream extrastriate visual cortex sources (Martinez et al., 1999; Di Russo et al., 2003), whereas the N1 appears to reflect primarily ventral stream sources (Allison et al., 1999; Bentin et al., 1999; Doniger et al., 2000). Though additional dorsal stream sources have been demonstrated for the N1 (Sehatpour et al., 2006; Novitskiy et al., 2011), dorsal involvement of N1 is less pronounced than for P1. The P1 is a sensory component, whereas the N1 reflects an initial stage of perceptual processing involving general feature discrimination. For instance, the N1 is larger in response to easy versus difficult to identify shapes, unlike the P1 (Foxe et al., 2005).

Sensory and feature discrimination processes both contribute to later stages of object recognition processing in LOC. This study provides information about contributions of different stages of processing to contour integration impairment in schizophrenia. In addition to assessing these components, a negative component was found that peaked at ∼120 ms, which we termed N120. While a behavioral version of this task has recently been optimized and validated for use in clinical trials and fMRI studies as part of the CNTRICS initiative (Silverstein et al., 2012), this study provides an initial exploration of an ERP analog of this task that could be further developed for use in clinical trials.

We hypothesized that the P1 component would not be sensitive to easy versus difficult to identify contours (i.e., low versus high jitter) but that the later components, particularly the Ncl, would show a differential response to these conditions. In addition, we hypothesized that patients would show a decrease in P1 and Ncl components, but not N1, compared to controls.

# **MATERIALS AND METHODS PARTICIPANTS**

Participants consisted of 21 patients meeting Diagnostic and Statistical Manual of Mental Disorder (DSM-IV) criteria for schizophrenia (*n* = 15) or schizoaffective disorder (*n* = 6) and 28 healthy volunteers. Patients were recruited from inpatient and outpatient facilities associated with the Nathan Kline Institute for Psychiatric Research. Diagnoses were obtained using the Structured Clinical Interview for *DSM-IV* (SCID) (First et al., 1997) and all available clinical information. Controls were recruited through the Volunteer Recruitment Pool at the Nathan Kline Institute. Healthy volunteers with a history of SCID-defined Axis I psychiatric disorders were excluded. Patients and controls were excluded if they had any neurological or ophthalmologic disorders that might affect performance or met criteria for alcohol or substance dependence within the last six months or abuse within the last month. All participants had 20/32 or better corrected visual acuity on the Logarithmic Visual Acuity Chart (Precision Vision, LaSalle, IL). This study was approved by the Nathan Kline Institute for Psychiatric Research/Rockland Psychiatric Center and Rockland County Department of Mental Health Institutional Review Board and all participants provided informed consent according to the Declaration of Helsinki.

Groups did not differ significantly in age (patients, 36.8 ± 10.5 y; controls, 35.8 ± 11.8 y; *p* = 0.76) or gender (patients, 20 males, 1 female; controls, 22 males, 6 females, *p* = 0.21). As expected, patients had significantly lower IQ (patients, 92.5 ± 8.1; controls, 105.6 ± 12.9, *p* < 0.001) (Ammons and Ammons, 1962) and education (patients, 11.3 ± 1.4 y; controls, 14.3 ± 2.0 y, *p* < 0.001). Individual IQ data were missing for one patient. Patients were ill for a mean of 13.5 ± 10.2 years and were receiving a mean antipsychotic dose equivalent to 988.59 ± 460.4 mg CPZ per day. Duration of illness data were missing for two patients. Mean Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987) 5-factor scores (Lindenmayer et al., 1994) for Positive, Negative, and Cognitive (Disorganized) factors were 3.2 (mild to moderate) 2.5 (minimal to mild), and 2.2 (minimal to mild), respectively. PANSS data could not be obtained for one patient.

# **STIMULI**

Stimuli were obtained from Silverstein and colleagues (Kozma-Wiebe et al., 2006; Silverstein et al., 2009). The carrier spatial frequency of the Gabor patches was 5 cycles/degree and their contrast was approximately 95%. The stimuli consisted of a closed chain of Gabor patches forming an egg-like shape within a background of randomly oriented Gabor stimuli. The spacing between the contour elements was kept constant (8λ; where λ is the wavelength of the Gabor stimulus) as was the average spacing between the background elements. The value (average adjacent background element spacing: contour element spacing) of each image was 0.9. By keeping the signal-to-noise ratio at a constant level below 1.0, participants' performance was a function of the adequacy of long-range interactions between spatial filters; density cues could not be used to detect contours, as is possible with values above 1.0 when contour elements are closer together than background elements. Prior studies indicated that at - = 0.9, chronic, state hospitalized schizophrenia patients are able to detect contours accurately in the absence of orientational jitter (Silverstein et al., 2000, 2006). Two orientation levels were used: 7–8◦ (low jitter) and 27–28◦ (high jitter) (**Figure 1**). This resulted in a small misalignment of the contours in the low-jitter condition and a much greater misalignment in the high-jitter condition. There were 40 stimuli for each jitter level, and these were divided evenly between left- and right- pointing egg-shaped ovals, for a total of 80 stimuli. The egg-shaped ovals were always roughly in the center of the image, although each contour varied slightly in size, local and global curvature, and spatial location. Stimuli were presented on a Phillips CRT monitor located 114 cm from the participants. Visual angle was 9.5 × 7 degrees. The mean luminance of the monitor was 65 cd/m2.

# **PROCEDURE**

The 80 stimuli were intermixed randomly and shown for 250 ms each with an interstimulus interval (ISI) of 2000 ms. Participants were instructed to look at the center of the screen, and to press the left or right mouse button if the contour pointed toward the left or right, respectively. Practice trials were given with the low-jitter stimuli in blocks of twenty trials until participants scored above chance on at least one block. During the ERP experiment, all 80 unique stimuli were viewed once in each block, and the blocks were approximately 3 min long. Eight blocks were conducted overall, allowing participants to view 320 trials each of low- and high-jitter stimuli. Participants were periodically prompted to stay focused. The behavioral outcome measure was the percent of correct responses.

#### **DATA ACQUISITION**

High-density continuous EEG was acquired from 64 surface electrode sites that were arranged equidistant from each other, using the ANT/Duke layout and EEProbe acquisition system (ANT, Enschede, The Netherlands), along with digital stimulus timingtags. Data were digitized online at 512 Hz. All data were recorded relative to a common average reference (i.e., average of all electrodes) online. Epochs (−100 to 400 ms) were created off-line. Data were baseline corrected from −100 ms to stimulus onset and an artifact rejection of ±120μv was applied to all electrodes. Trials containing eye-movements, identified as deflections of >10μV lasting >25 ms appearing on both eye channels, were rejected offline. The number of trials that survived artifact rejection for patients was 206 ± 68.9 and for controls was 255 ± 51.8 and was significantly different between groups (*t*(47) = 2.9, *p* = 0.006). Epochs were averaged for each participant for each jitter condition. A spherical spline algorithm was implemented to compute the current source density (CSD) of the EEG. The CSD represents an improvement over more commonly used voltage measures as it decreases the effects of volume conduction and acts as a high-pass spatial filter, which reduces the overlap between ERPs at different sites (Saron et al., 2003; Whitford et al., 2011) and has been used in previous schizophrenia studies (Whitford et al., 2011; Kayser et al., 2012).

## **ERP COMPONENTS**

CSD topography maps were used to select the electrode sites with most prominent P1, N120, N1, and Ncl components. P1 was maximal over dorsal-occipital electrodes (right hemisphere: P4, P6, PO8; left hemisphere: P3, P5, PO7), N120 over central electrodes (P0z and Oz), and N1 and Ncl over lateral, ventral-occipital sites (right hemisphere: P8, PO8, PO10; left hemisphere: P7, PO7, and PO9). The mean CSD was obtained for each component over specific latency windows (P1: 85–115 ms; N120: 105–135 ms; N1: 150–180 ms; Ncl: 250–290 ms). Grand average waveforms were constructed separately for low and high jitter for patients and controls. Figures show filtered waveform data (50 Hz low-pass, 24 dB/octave roll-off).

#### **STATISTICAL ANALYSIS**

Between-group analyses for the behavioral measure (percent correct) and each ERP component (P1, N120, N1, or Ncl) were performed separately using mixed-model analyses of variance (ANOVAs) with group (patient, control) as the between-subjects factor, and with jitter (low, high) and hemisphere (right, left for ERP components) as within-subject factors. Relationships between measures were assessed using Pearson correlation coefficients. Behavioral data were not collected from 1 patient and 2 controls in the low-jitter condition and 2 patients and 2 controls in the high-jitter condition. Effect sizes for ANOVAs are reported as partial eta squared (η<sup>2</sup> *<sup>p</sup>*) to be consistent with past studies of contour integration (Keane et al., 2012; Silverstein et al., 2012).

# **RESULTS**

#### **BEHAVIORAL RESULTS**

Controls performed significantly better than patients [*F*(1, <sup>43</sup>) = <sup>12</sup>.7, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.228]. Both groups performed better on the low-jitter condition than on the high-jitter condition

[*F*(1, <sup>43</sup>) <sup>=</sup> 198, *<sup>p</sup>* <sup>&</sup>lt; <sup>0</sup>.001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.822] (**Figure 2**). There was also a significant group × jitter interaction [*F*(1, <sup>43</sup>) = 4.1, *p* = 0.049, η<sup>2</sup> *<sup>p</sup>* = 0.087], with a greater between-group difference in the low- than high-jitter condition.

# **ELECTROPHYSIOLOGY RESULTS** *P1*

P1 CSD was significantly reduced in patients compared to controls [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>8</sup>.7, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.005, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.156] (**Figure 3**). Group × jitter [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>1</sup>.5, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.23, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.031] and group × jitter × hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>1</sup>.96, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.16, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.04] interactions were not significant, reflecting similar P1 CSD to low- and highjitter stimuli. There was no significant group × hemisphere interaction [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.08, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.78, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.002] or main effect of hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>2</sup>.1, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.15, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.043].

# *N120*

There was a significant main effect of jitter [*F*(1, <sup>47</sup>) = 4.4, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.04, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.085], indicating differential activity across groups to low- versus high-jitter stimuli (**Figure 4**). A significant group × jitter interaction was also found [*F*(1, <sup>47</sup>) = 4.1, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.05, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.081], indicating lack of differential CSD to low versus high jitter in patients [*F*(1, <sup>20</sup>) = 0.001, *p* = 0.97, η2 *<sup>p</sup>* < 0.001] but differential CSD to the jitter conditions in controls [*F*(1, <sup>27</sup>) <sup>=</sup> <sup>11</sup>.5, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.002, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.298, a large effect size; (Pallant, 2007)].

# *N1*

N1 amplitude was not significantly different between groups [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.9, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.34, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.019] (**Figure 5**). Group × hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.29, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.59, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.006], group × jitter [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.13, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.71, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.003], and group × jitter × hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.21, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.65, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.004] interactions were not significant. However, the low-jitter condition produced a more negative peak than the high-jitter condition [*F*(1, <sup>47</sup>) = 17.19, *p* < 0.001, η<sup>2</sup> *<sup>p</sup>* = 0.268].

# *Ncl*

There was a significant main effect of jitter [*F*(1, <sup>47</sup>) = 47.8, *p* < 0.001, η<sup>2</sup> *<sup>p</sup>* = 0.504], indicating differential activity across groups to low- versus high-jitter stimuli (**Figure 5**). A significant group × jitter interaction was also found [*F*(1, <sup>47</sup>) = 8.2, *p* = 0.006, η<sup>2</sup> *<sup>p</sup>* = 0.148], indicating reduced differential activity to low versus high jitter in patients [*F*(1, <sup>20</sup>) <sup>=</sup> <sup>8</sup>.1, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.01, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.288] compared to controls [*F*(1, <sup>27</sup>) <sup>=</sup> <sup>51</sup>.5, *<sup>p</sup>* <sup>&</sup>lt; <sup>0</sup>.001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.656]. Group <sup>×</sup> hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.02, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.884, <sup>η</sup><sup>2</sup> *<sup>p</sup>* < 0.001], jitter <sup>×</sup> hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.29, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.59, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.006], and group <sup>×</sup> jitter <sup>×</sup> hemisphere [*F*(1, <sup>47</sup>) <sup>=</sup> <sup>0</sup>.76, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.39, <sup>η</sup><sup>2</sup> *p* = 0.016] interactions were not significant.

#### **RESULTS FOR PARTICIPANTS PERFORMING AT 70 PERCENT CORRECT OR BETTER**

To assess whether the ERP results were due to difficulty of patients in identifying the contour and/or to a general deficit, analyses were also carried out with only those participants reaching at least

70% correct in the behavioral low-jitter condition. This resulted in the inclusion of 12 out of 21 patients and 22 out of 28 controls. The general pattern of results was similar to that seen with the entire cohort. For P1, CSD for patients was lower than controls, with a trend for a significant difference between groups [*F*(1, <sup>32</sup>) <sup>=</sup> <sup>3</sup>.9, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.057, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.109]. For N120, the significant group <sup>×</sup> jitter interaction [*F*(1, <sup>32</sup>) <sup>=</sup> <sup>4</sup>.6, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.04, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.127] remained, but the main effect of jitter was no longer significant [*F*(1, <sup>32</sup>) <sup>=</sup> <sup>0</sup>.5, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.5, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.015]. For N1, the main effect of jitter remained significant [*F*(1, <sup>32</sup>) = 12.2, *p* = 0.001,

η2 *<sup>p</sup>* = 0.276]. For Ncl, a significant main effect of jitter was still found [*F*(1, <sup>32</sup>) <sup>=</sup> <sup>47</sup>.8, *<sup>p</sup>* <sup>&</sup>lt; <sup>0</sup>.001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.599] as was a group <sup>×</sup> jitter interaction [*F*(1, <sup>32</sup>) <sup>=</sup> <sup>4</sup>.5, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>.04, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0.123]. This indicates that even when only participants who were clearly performing the low-jitter task above chance were included, there was still a large difference in Ncl response to low- versus high-jitter stimuli overall and patients showed a reduced difference between jitter conditions compared to controls. However, while the behavioral deficit was less pronounced than when all participants were included, patients still showed an impaired behavioral response to

**FIGURE 5 | Event-related potential CSD responses to low- and high-jitter stimuli from occipital electrodes (right hemisphere: P8, PO8, PO10; left hemisphere: P7, PO7, and PO9) in controls and patients.** CSD waveforms show a small difference in response to low- versus high-jitter for the N1 component (blue shaded window) and a much larger difference between conditions for the Ncl

component (purple shaded window). CSD maps at 270 ms show the observed difference in negativity between the low- and high- jitter conditions for each group in the time-frame of the Ncl component. The bar graph shows significant differences between the groups in the responses to low- vs high-jitter stimuli for the Ncl component. ∗*p* < 0.05, ∗∗*p* < 0.005.

low- [*t*(32) = 2.1, *p* = 0.045, effect size Cohen's *d* = 0.75 standard deviation units] and high-jitter [*t*(31) = 2.2, *p* = 0.046, effect size Cohen's *d* = 0.95 standard deviation units] conditions compared to controls.

#### **SOURCE ANALYSIS**

Inverse dipole modeling was carried out on the grand average waveforms using the brain electrical source algorithm (Scherg, 1990). Models were constructed with three sets of paired dipoles

with each set constrained to be symmetrical in location but not orientation, over restricted time intervals corresponding to the components of interest. Dipole locations were calculated based on control data and subsequently applied to patient results. The P1 component was fitted to a window of 85–115 ms accounting for 97% of the variance in controls and 96% in patients. The generator sources were localized to extrastriate Brodmann Area 18 (V2-V3) (Tailarach coordinates: *x*, ±18; *y*, −89; *z*, 13). The N120 component was fitted to a window of 115–135 ms to obtain a clear temporal separation from the P1 component, and accounted for 99% of the variance in controls and 98% of the variance in patients. The generator sources were localized to extrastriate Brodmann Area 18 (Tailarach coordinates: *x*, ±14; *y*, −74; *z*, 16). The N1 component was fitted to a window of 150– 180 ms accounting for 96% of the variance in controls and 98% in patients. The Ncl was fitted to a window of 250–290 ms accounting for 98% of the variance in controls and 92% in patients. These two components were co-localized to Brodmann Area 37 (fusiform gyrus) (Tailarach coordinates: *x*, ±31.4; *y*, −50;*z*, −13).

#### **CORRELATIONS BETWEEN ERP MEASURES**

ERP components were collapsed across hemisphere because there were no significant main effects or interactions with hemisphere. For P1, correlations were performed for low and high jitter separately. For N120, N1, and Ncl, high-jitter CSD was subtracted from low-jitter CSD because these components showed a differential response to jitter. P1 responses were significantly correlated with the Ncl CSD difference for controls (low jitter: *r* = −0.44, *p* = 0.02; high jitter: *r* = −0.41, *p* = 0.03) but not patients (low jitter: *r* = −0.29, *p* = 0.2; high jitter: *r* = −0.35, *p* = 0.12). In addition, controls showed a trend for (*r* = 0.36, *p* = 0.057) and patients showed a significant (*r* = 0.48, *p* = 0.03) correlation between N1 and Ncl CSD differences.

# **CORRELATIONS BETWEEN ERP/BEHAVIORAL AND CLINICAL MEASURES**

No significant relationships were found between any ERP component or results on the behavioral task and CPZ equivalents or duration of illness for patients. However, using the 5-factor PANSS, patients showed a significant relationship between the Cognitive (Disorganized) factor and both low-jitter behavior (*r* = −0.47, *n* = 19, *p* = 0.04) and Ncl CSD (*r* = 0.51, *n* = 20, *p* = 0.02), replicating past findings that poorer contour integration is related to greater disorganization in other domains (Silverstein et al., 2000; Uhlhaas et al., 2006; Silverstein and Keane, 2011). In addition, for patients, there was a significant relationship between percent correct on behavioral performance in the low-jitter condition and Ncl (*r* = −0.51, *p* = 0.02), but not other components.

# **DISCUSSION**

Patients with schizophrenia have impaired visual integration abilities that have been documented in more than 50 studies, spanning over 50 years, as demonstrated by several different research groups, using varied tasks, and in different cultures (Snyder et al., 1961; Izawa and Yamamoto, 2002; Chen et al., 2005; Sehatpour et al., 2010; Silverstein and Keane, 2011). The present study used a contour integration task that was used previously by Silverstein and colleagues (Silverstein et al., 2006, 2009, 2012) to investigate visual integration deficits in schizophrenia. While the contour integration task has been extensively used and optimized for behavioral (Silverstein and Keane, 2011; Silverstein et al., 2012), and fMRI (Silverstein et al., 2009) studies in schizophrenia, there have been no ERP studies that have examined the contributions of early and later stages of processing to these deficits. The present study examined ERP responses to low-jitter stimuli, in which contours were easier to identify, versus high-jitter stimuli, in which contours were more difficult to identify, to further understand contributions of different stages of processing to ability to integrate contours.

As expected, neither patients nor controls showed a differential P1 response to the low- versus high-jitter condition. The P1 component, which peaks at ∼100 ms, is an early sensory component and responds to low-level stimulus properties. Aside from the degree of jitter, the low-level stimulus properties of the two conditions were the same and so would not be expected to differentially affect the P1. However, patients showed a significant decrease in the P1 CSD response compared to controls. The current source localization findings of the P1 dipole in extrastriate Brodmann Area 18 are consistent with previous detailed source localization of the dorsal P1 (Di Russo et al., 2001) and with fMRI findings showing reduced activation to Gabor-defined contours in extrastriate visual areas in schizophrenia (Silverstein et al., 2009). The decreased P1 seen in the present study is also consistent with findings from perceptual closure and illusory contour tasks, in which P1 amplitude was similar to scrambled versus unscrambled objects and to recognizable versus nonrecognizable illusory shapes, but was reduced in patients versus controls (Foxe et al., 2005; Sehatpour et al., 2010). Decreased P1 has also been seen in a number of studies of schizophrenia (Yeap et al., 2006; Butler et al., 2007; Haenschel et al., 2007; Dias et al., 2011; Martinez et al., 2011), and is thought to reflect impaired magnocellular and/or early-stage cortical processing (Butler et al., 2007; Martinez et al., 2011). Bar and colleagues (Bar, 2003; Kveraga et al., 2007) have suggested that the magnocellular system provides a low-resolution template of stimuli that projects to frontal cortex, which in turn is involved in topdown contributions to object recognition. While a shape rather than an object was shown to participants in the present study, impairment at this early stage of processing, indexed by the P1, may contribute to impaired ability to form a "frame" of the contour.

The N120 component, which peaked at 120 ms, showed a small but significant difference between low- versus high-jitter stimuli. This was unexpected because a component that differentiates between levels of recognizability has not, to our knowledge, been reported earlier than the N1 component. Controls showed a significant effect of jitter whereas for patients, there was an almost total lack of differentiation between the jitter conditions at this point in processing. Thus, this component appears to represent an early cortical process that is sensitive to differing levels of prepotent stimulus organization, and is impaired in patients with schizophrenia. The time-window, orientation, and Tailarach positions of dipoles differed for P1 and N120, though a dorsal stream source was found for both. However, further studies are needed to evaluate this component.

The N1 component, which peaked at ∼165 ms, showed a small but significant difference in response to low- versus high-jitter stimuli. This is consistent with previous studies showing a larger N1 when classes of objects rather than presence or absence of objects are required to be detected (Mangun and Hillyard, 1991; Vogel and Luck, 2000) or when illusory contours versus noncontour control stimuli are presented (Murray et al., 2004; Foxe et al., 2005). However, the N1 was not diminished in patients with schizophrenia versus controls. The N1 indexes initial activation of ventral stream object recognition areas (Sehatpour et al., 2010). Thus, while this component appears to be sensitive to jitter conditions, the initial input to the ventral stream appears to be intact, in agreement with previous studies of visual integration in schizophrenia (Doniger et al., 2002; Foxe et al., 2005; Sehatpour et al., 2010).

For controls, the Ncl component, peaking at ∼270 ms, was highly divergent between the low- and high-jitter conditions, with a much greater negativity seen to the low-jitter than highjitter stimuli. However, this was not the case for patients, who showed a significantly smaller difference than controls to lowversus high-jitter stimuli. In addition, for patients, better behavioral performance in the low-jitter condition was significantly related to a greater Ncl (i.e., a greater difference between lowversus high-jitter CSD), suggesting that this component is important in the actual recognition of contours. The Ncl, like the N1, was source localized to ventral stream fusiform gyrus, and this region was shown to be involved in contour integration in a prior fMRI study (Silverstein et al., 2009). The intact N1 indicates that rather than intrinsic impairment in this ventral stream area, the impaired Ncl response in patients may be due to aberrant input from other brain areas. Indeed, the finding of a significant correlation between P1 and Ncl in controls, but not patients, suggests contributions of dorsal stream processes to ventral stream object recognition, which are dysfunctional in patients.

The present ERP results for Gabor-defined contours represent something of a hybrid of results found for perceptual closure, which involves actual objects (Doniger et al., 2002; Sehatpour et al., 2010), and for illusory contours, which involves more automatic shape processing (Murray et al., 2002; Foxe et al., 2005). Similar to the present study, perceptual closure results showed decreased P1 amplitude, a normal N1, and decreased Ncl to scrambled versus unscrambled line drawings of objects in patients versus controls (Doniger et al., 2002; Sehatpour et al., 2010). The Ncl was localized to ventral stream areas in these studies. A combined fMRI/ERP study of perceptual closure showed that a distributed network was involved in perceptual integration such that impaired activation of dorsal stream visual regions contributed significantly to impaired frontal activation, which in turn contributed to impaired activation of hippocampus and ventral stream regions (Sehatpour et al., 2010). This is consistent with a "frame and fill" model of object recognition (Bar, 2003) in which the low-resolution template generated by the fast magnocellular/dorsal stream provides the "frame" for the fine-detailed information provided by slower parvocellular visual projections to the ventral stream. However, like the illusory contour study (Murray et al., 2004; Foxe et al., 2005), the present study also produced a differential N1 to easy versus difficult to identify stimuli, but did not differ between groups, whereas the objects used in the perceptual closure study produced differential processing only in the later Ncl component and not the earlier N1. Thus, both early (i.e., N120, N1) and later (i.e., Ncl) processes appear to differentiate between recognizability of the Gabor-defined contours.

Using a contour integration task, which was the basis for the task used in the present study, Silverstein and colleagues (Silverstein et al., 2009) found a distributed network of brain activity in controls which included greater recruitment of visual areas V2/V3, and V4 as well as frontal and parietal areas compared to schizophrenia patients in an fMRI study. Silverstein and colleagues suggested that recruitment of anterior areas involved in attention is partly driven by the quality of form representations constructed in the occipital lobe, and, conversely, that extent of activation in the occipital lobe is affected by the amount of feedback from frontal and parietal areas to visual areas to increase the salience of the contours relative to the background noise. They also suggested that the lack of impairment in V1 is consistent with the hypothesis that this area is involved in smaller scale groupings (which can occur normally in schizophrenia). However, in regions progressively anterior to V1, grouping occurs over increasingly larger regions of space (Angelucci et al., 2002), and it is in these regions where processing breaks down in the disorder. Results are consistent with areas found to be activated in controls and non-human primates in contour integration (Kourtzi et al., 2003). While speculative, the present ERP results suggest that the impairment in the contour integration task, as in the perceptual closure task (Sehatpour et al., 2010), may involve impaired input to the frontal cortex from dorsal stream areas and subsequent impaired frontal input to ventral stream areas.

Behavioral performance was impaired in the low-jitter condition for patients versus controls in the present study. In one previous study, performance of patients was not impaired in a low-jitter 7–8◦ condition (Silverstein et al., 2009), whereas in several other studies it was impaired (Kozma-Wiebe et al., 2006; Silverstein et al., 2012). This may be partly due to prior exposure to a 0-degree jitter condition in some of the studies, but not others (see Silverstein and Keane, 2009; Silverstein et al., 2012). However, it is also likely to be due to differences between studies in stimulus exposure duration; in previous studies duration was 2000 ms whereas it was 250 ms in the present study. This resulted in lower levels of accuracy even for controls in the present study compared to previous studies (Kozma-Wiebe et al., 2006; Silverstein et al., 2009, 2012). Studies using a longer duration stimulus presentation as well as more levels of jitter would be helpful in optimizing the ERP version of this task.

Contour integration deficits were previously found to be related to increased disorganization and/or poor outcome (Uhlhaas et al., 2005, 2006; Silverstein et al., 2005; Silverstein and Keane, 2011). Using the 5-factor PANSS, the present study showed a significant correlation between the Cognitive (Disorganized) factor and low-jitter behavior, in agreement with previous findings, and extended these observations to show a significant relationship between the Cognitive factor and Ncl in patients. Indeed, Phillips and Silverstein (2003) suggested that perceptual organization, including contour integration, is a lowlevel manifestation of a general computational processing deficit involving binding features that are contextually related, which would explain links between impaired perceptual organization and disorganized thinking as well as other cognitive impairments.

It is important to address whether a general deficit and/or impaired motivation or attention could underlie the present results. This does not seem likely for several reasons. While patients performed more poorly than controls on both the lowand high-jitter behavioral conditions, when results were analyzed including only participants who were responding at much greater than chance levels (e.g., 70% correct on the low-jitter condition), patients still showed a significant impairment in Ncl compared to controls. In addition, previous studies of contour integration have addressed this issue. For instance, Silverstein and colleagues (Silverstein et al., 2009) found that patients still had decreased activation in areas V2/V3 compared to controls when they were matched to controls on accuracy. Further, as reviewed by Uhlhaas and Silverstein (2005), 10 schizophrenia studies found that patients performed better than controls on tasks in which ability to group stimuli interferes with ability to respond to single stimuli. The superior performance of patients in response to single stimuli is strong evidence for impaired integration independent of a general deficit. Finally, patients did not show a deficit in the N1 component, which would have been expected if results were due to a general deficit.

A further limitation of the study was that all patients were receiving antipsychotic medication. However, CPZ equivalents were not correlated with any ERP component or behavioral performance. In addition, lack of relationship with medication has been found in previous studies of contour integration and a deficit has also been found in un-medicated patients (Silverstein

#### **REFERENCES**


Pernier, J. (1999). ERP manifestations of processing printed words at different psycholinguistic levels: time course and scalp distribution. *J. Cogn. Neurosci.* 11, 235–260.


and Keane, 2011). In addition, a limitation is also that this was a relatively small sample size and was almost completely male, particularly in the patient group.

The present results extend previous behavioral and fMRI contour integration findings to the ERP domain. Further, the current contour integration paradigm demonstrated similar ERP signatures as were found in perceptual closure studies (Doniger et al., 2002; Sehatpour et al., 2010) as well as illusory contour studies (Foxe et al., 2005). This may reflect both a general pattern of ERP deficits for perceptual organization and a common impairment in overall perceptual organization in schizophrenia. Taken together, ERP and fMRI data from contour integration, perceptual closure, and illusory contour paradigms complement each other and suggest that impairment in contour integration in patients with schizophrenia occurs at early perceptual levels but also involves higher regions of cortex. This study also provides a pilot assessment of an ERP version of the contour integration task for use in clinical trials.

## **ACKNOWLEDGMENTS**

The authors would like to thank Gail Silipo for her critical contribution to study management, Joanna DiCostanza, Rachel Ziwich, and Jamie Sanchez for their critical contributions to patient recruitment, assessment, and data management, and Dr. Manuel Gomez-Ramirez for writing MATLAB scripts for determining mean CSDs within specific latency windows. The authors also thank the faculty and staff of the Clinical Research and Evaluation Facility and the Outpatient Research Service at the Nathan S. Kline Institute for Psychiatric Research. The authors thank the two reviewers for their helpful suggestions. We also thank all the people who participated in the study.

# **FUNDING**

This work was supported by the National Institutes of Health (RO1 MH084848 to Pamela D. Butler and Elisa C. Dias).


potential. *Hum. Brain Mapp.* 15, 95–111.


local "association field". *Vision Res.* 33, 173–193.


I., Feher, A., et al. (2006). Reduced top-down influences in contour detection in schizophrenia. *Cognit. Neuropsychiatry* 11, 112–132.


ERP and DTI study. *Neuroimage* 54, 2318–2329.

Yeap, S., Kelly, S. P., Sehatpour, P., Magno, E., Javitt, D. C., Garavan, H., et al. (2006). Early visual sensory deficits as endophenotypes for schizophrenia: high-density electrical mapping in clinically unaffected first-degree relatives. *Arch. Gen. Psychiatry* 63, 1180–1188.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 January 2013; accepted: 03 March 2013; published online: 21 March 2013.*

*Citation: Butler PD, Abeles IY, Silverstein SM, Dias EC, Weiskopf NG, Calderone DJ and Sehatpour P (2013) An eventrelated potential examination of contour integration deficits in schizophrenia. Front. Psychol. 4:132. doi: 10.3389/fpsyg. 2013.00132*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Butler, Abeles, Silverstein, Dias, Weiskopf, Calderone and Sehatpour. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Oscillatory dynamics of Gestalt perception in schizophrenia revisited

# *Kevin M. Spencer 1,2\* and Shahab Ghorashi 1,2*

*<sup>1</sup> Research Service, Veterans Affairs Boston Healthcare System, Boston, MA, USA*

*<sup>2</sup> Department of Psychiatry, Harvard Medical School, Boston, MA, USA*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Scott R. Sponheim, University of Minnesota, USA Peter Uhlhaas, University of Glasgow, UK William A. Phillips, University of Stirling, UK*

#### *\*Correspondence:*

*Kevin M. Spencer, Veterans Affairs Boston Healthcare System, Research 151C, 150 S. Huntington Ave., Boston, MA 02130, USA e-mail: kevin\_spencer@ hms.harvard.edu*

**Background:** Abnormalities in **γ** oscillations (30–100 Hz) in the scalp-recorded electroencephalogram (EEG) have been proposed to reflect neural circuitry abnormalities in schizophrenia. Oscillations in the γ band are thought to play an important role in visual perception, mediating the binding of visual features into coherent objects. However, there is relatively little evidence to date of deficits in γ-mediated processes associated with Gestalt perception in schizophrenia.

**Methods:** Fourteen healthy control subjects (HC) and 17 chronic schizophrenia patients (SZ) discriminated between illusory Kanisza Squares and No-Square control stimuli, indicating their judgment with a manual button press. Time-frequency decomposition of the EEG was computed with the Morlet wavelet transform. Time-frequency maps of phase locking factor (PLF) values were calculated for stimulus- and response-locked oscillations.

**Results:** HC and SZ did not differ in reaction time, error rate, an early ERP effect associated with Gestalt processing, nor an early visual-evoked γ oscillation. Two response-locked high γ effects had greater PLF for Square than No-Square stimuli in HC, and the reverse pattern in SZ. One of these effects was correlated with thought disorder symptom ratings in SZ.

**Conclusions:** SZ demonstrated abnormalities in γ oscillations associated with the perception of Gestalt objects, while their early visual-evoked γ activity was mostly normal, contrary to previous results. This study supports the hypothesis that high-frequency oscillations are sensitive to aspects of psychosis.

**Keywords: schizophrenia, visual perception, Gestalt, gamma oscillation, event-related potential**

# **INTRODUCTION**

Abnormalities in γ oscillations (30–100 Hz) in the scalp-recorded electroencephalogram (EEG) have been proposed to reflect cortical circuitry abnormalities in schizophrenia, particularly reduced inhibition of pyramidal cells by fast-spiking interneurons (Kwon et al., 1999; Gonzalez-Burgos and Lewis, 2008; Woo et al., 2010). Classically, γ oscillations have been thought to play an important role in visual perception, mediating the binding of visual features into coherent objects (Singer and Gray, 1995; Singer, 1999; but see Thiele and Stoner (2003), Palanca and DeAngelis (2005), Lima et al. (2010). Thus, it is perhaps surprising that while γ oscillation abnormalities have been reported in various domains in schizophrenia, such as auditory (e.g., Kwon et al., 1999; Leicht et al., 2010) and visual sensory function (e.g., Spencer et al., 2008a; Grützner et al., 2013), working memory (Cho et al., 2006; Haenschel et al., 2009), and motor control (Ford et al., 2008), to date relatively little evidence of γ abnormalities associated with visual object perception in schizophrenia has been reported. Rather, investigations of oscillatory activity during visual perception tasks involving feature binding have mainly reported abnormalities in β frequency oscillations (13–30 Hz), as we review below.

In the first study of γ oscillations related to visual featurebinding in schizophrenia, Spencer et al. (2003) found that the early visual-evoked γ oscillation (Vγ1; ∼25–40 Hz, 80–120 ms) was evoked for Gestalt (Kanisza square) but not control stimuli in control subjects. In schizophrenia patients neither stimulus type evoked Vγ1. While this deficit was suggestive of an abnormality in an oscillatory feature-binding process, Vγ1 has been associated with attentional and memory processes (Herrmann et al., 2004), rather than feature binding *per se* (Tallon-Baudry et al., 1996). A follow-up study examining Vγ1 in a simple oddball task also found a deficit in schizophrenia patients, suggesting that the early perceptual process manifested by this oscillation was impaired in a task—and stimulus-independent manner (Spencer et al., 2008a). Further studies have reported deficits in visual-evoked γ oscillations in schizophrenia (Krishnan et al., 2005; Wynn et al., 2005; Grützner et al., 2013; Sun et al., 2013).

In a second study utilizing the same Gestalt perception task with Kanisza squares, Spencer et al. (2004) investigated oscillations that preceded and were phase-locked to the time of the manual response, reasoning that such response-locked oscillations might be more related to perceptual decision-making processes than stimulus-locked oscillations. The authors found a response-locked β oscillation (∼24 Hz) in schizophrenia patients that was elicited when they detected the Gestalt stimulus, but not the control stimulus. A similar oscillation was found in the low γ band (∼37 Hz) of the control subjects. The Gestalt effect (Square minus No-Square) on the patients' β oscillation was positively correlated with particular psychotic symptoms including thought disorder, disorganization, and visual hallucinations. The correlation between disorganization symptoms and the β Gestalt effect was consistent with the relationships observed between disorganization and Gestalt perception by Silverstein et al. (2000), Uhlhaas et al. (2005, 2006a). While Spencer et al. (2004) proposed that the β oscillation in patients was equivalent to the γ oscillation in controls but at a lower frequency, the sparse electrode array used in that study did not allow for a detailed comparison of the scalp topographies of these oscillations to establish whether or not these oscillations might have similar neural generators.

Uhlhaas and colleagues have published a series of studies examining the oscillatory correlates of Gestalt perception in schizophrenia using "Mooney" face stimuli (Mooney and Ferguson, 1951). Uhlhaas et al. (2006b) reported that induced γ oscillatory activity in the 40–70 Hz band did not differ between patients and controls, but inter-electrode phase synchrony in the β band evoked by "Mooney" faces was reduced in patients. Face-evoked β synchrony was positively correlated with positive symptoms, similar to the correlations between the β Gestalt effect and positive symptoms in Spencer et al. (2004). Reductions in γ band phase synchrony in patients were found in the face but also the no-face conditions. However, in a study employing the same task and using magnetoencephalography, Grützner et al. (2013) found that γ power in the response to the offset of upright compared to inverted face stimuli was reduced in schizophrenia patients. Most recently, the clearest example of a deficit in γ activity related to Gestalt perception in schizophrenia was reported by Sun et al. (2013), who found reduced Gestalt effect on a high γ induced oscillation in first-episode patients.

Thus, the evidence for γ oscillation deficits related to visual feature-binding in schizophrenia is rather sparse, despite the evidence for an important role of γ in visual feature-binding (Singer and Gray, 1995; Singer, 1999), and the accumulation of behavioral findings pointing to impaired visual integrative processes in this disorder (reviewed in Uhlhaas and Silverstein, 2005; Butler et al., 2008). In the present study we revisited this issue by replicating our Gestalt studies utilizing Kanisza stimuli. As in our previous studies (Spencer et al., 2003, 2004), subjects discriminated between Kanisza square and control stimuli. Here we used a dense electrode array to obtain more detailed information about the spatial topography of the oscillations under study. We also made a modification to the experimental paradigm by presenting the stimuli briefly, instead of remaining present until the manual response was issued, to test whether responselocked oscillations might be driven in part by the physical persistence of the stimuli. We predicted that γ abnormalities in schizophrenia patients would be correlated with positive symptom ratings, specifically hallucinations, thought disorder, and/or disorganization, as in our previous findings (Spencer et al., 2004).

# **MATERIALS AND METHODS**

# **SUBJECTS**

This study was approved by the Institutional Review Boards of the Veterans Affairs Boston Healthcare System and Harvard Medical School. After a complete description of the study to the subjects, written informed consent was obtained. All subjects were paid for their participation.

Healthy control subjects (HC; *N* = 14, 2 females) and chronic schizophrenia patients (SZ; *N* = 17, 1 female) participated in the study. HC were recruited from the community. SZ were recruited from mental health services at the VA Boston Healthcare System, and were diagnosed according to DSM-IV criteria (First et al., 1995). All subjects were selected without regard for ethnicity, and met our standard inclusion/exclusion criteria: (1) age between 18 and 55 years; (2) right-handed as assessed by the Edinburgh handedness inventory (Oldfield, 1971) (so that possible hemispheric lateralization effects would not be obscured by left-handers with reduced or reversed functional laterality); (3) no history of electroconvulsive treatment; (4) no history of neurological illness, including epilepsy; (5) no history of alcohol or drug dependence, nor abuse within the last year, nor long duration (*>*1 year) of past abuse (DSM-IV criteria); (6) no present medication for medical disorders that would have deleterious EEG, neurological, or cognitive functioning consequences; (7) verbal IQ above 75; (8) no alcohol use in the 24 h prior to testing; and (9) English as a first language. In addition, HC were screened for the presence of an Axis-I disorder using the SCID-Non-Patient edition (First et al., 2002), and were also excluded if they reported having first-degree relative with an Axis I disorder.

Demographic and clinical data are presented in **Table 1.** Clinical symptoms were assessed using the Scale for the Assessment of Positive Symptoms (SAPS) (Andreasen, 1984) and the Scale for the Assessment of Negative Symptoms (SANS) (Andreasen, 1983). The final SZ and HC groups did not differ in sex proportions, age, or parental socio-economic status (PSES; Hollingshead, 1965). The diagnostic composition of the SZ group was 7 paranoid, 7 undifferentiated, and 3 schizoaffective. All of the patients were taking atypical antipsychotics at the time of the experiment. One patient was also receiving a typical antipsychotic. Antipsychotic medication dosages were converted to chlorpromazine equivalents (Stoll, 2001; Woods, 2003).

#### **STIMULI AND EXPERIMENTAL DESIGN**

Subjects were seated in a quiet, dimly-lit room, 1 m in front of a computer monitor upon which the visual stimuli were presented. Subjects were instructed to discriminate between Kanisza Square and control No-Square displays (90 trials per condition), pressing a button with one hand for square-present, and another button with the other hand for square-absent. The response hand assignment was counterbalanced across subjects.

The Square and No-Square stimuli were presented in white on a black background at fixation. A fixation cross was continuously present. The stimuli were 5◦ wide with a 0.4 support ratio (**Figure 1A**). Unlike our previous studies utilizing Kanisza displays in which the stimuli were presented until after the response had been made (Spencer et al., 2003, 2004), here the stimuli were presented for 106 ms.


**Table 1 | Demographic and clinical variables and between-group comparisons for the healthy control (HC) and schizophrenia patient (SZ) groups.**

*Mean* ± *SD are given for each variable.*

#### **EEG RECORDING AND PROCESSING**

The EEG was recorded with a Biosemi ActiveTwo system using active electrodes in an electrode cap at 71 standard scalp sites (DC–100 Hz bandpass filter, 512 Hz digitization rate). The DC offsets were kept below 25 mV. During data acquisition, all channels were referred to the system's internal loop (CMS/DRL sensors located in the parietal region), and were then re-referenced offline to the left mastoid electrode. The bipolar vertical electrooculogram (EOG) was derived from electrode Fp1 and an electrode below the left eye. The horizontal EOG was derived from electrodes on the left and right outer canthi.

The epoching and initial processing of the continuous EEG recording were performed with BrainVision Analyzer 2.0 (Brain Products GmbH). For each trial, a 2 s epoch was extracted from 400 ms pre-stimulus to 1598 ms post-stimulus. Further processing was performed using software in MATLAB (Mathworks, Inc.) and IDL (Exelis Visual Information Solutions, Inc.). Error trials were excluded from processing, and an initial artifact detection scan was run. The artifact exclusion criteria were: (1) *>* ± 90μV change in one time point; and (2) amplitude range within an epoch exceeding 200μV. Then independent component analysis [implemented in the *runica.m* program from EEGLAB (Delorme and Makeig, 2004)] was used to remove ocular and muscle artifacts. Independent components representing artifacts were identified based on their characteristic topographic, temporal, and spectral signatures (Keren et al., 2010; Shackman et al., 2010; Hipp and Siegel, 2013). Next, a second artifact detection scan was run. Finally, the retained correct-response, artifact-free epochs were re-referenced to the average reference (Dien, 1998), computed on all 68 scalp channels, excluding the EOG channels. The number of epochs retained per subject was (mean ± standard deviation) 158 ± 13 for HC and 157 ± 17 for SZ, and these numbers did not differ [*t(*29*)* = 0*.*300, *p* = 0*.*766].

Two data sets were created from the single trial epochs: stimulus-locked and response-locked. For the response-locked data set, the original stimulus-locked epochs were shifted according to the reaction time (RT) on each trial such that they encompassed the period from −1448 to 500 ms relative to RT.

Event-related brain potentials (ERPs) and spectral measures were computed from the single-trial epochs. Time-frequency (TF) decomposition was performed using the Morlet wavelet transform (Torrence and Compo, 1998), which was applied in 1 Hz steps from 7–100 Hz at each time point to yield TF maps. The wavelet frequency/duration ratio (f0/σf) was 6. Here we report results for the phase-locking factor (PLF) measure (Tallon-Baudry et al., 1996). PLF reflects the degree to which EEG phase is consistent across trials, and thus reflects event-locked synchronized activity. This measure is therefore highly redundant with the evoked power measure and tends to be more sensitive to effects on oscillations than evoked power. PLF is computed as one minus the circular variance of phases across trials and ranges from 0 (random distribution) to 1 (perfect phase locking).

Baseline activity was subtracted from each TF map. Baseline periods were from −150 to 0 ms relative to stimulus onset for the stimulus-locked maps, and 0 to 150 ms relative to reaction time (RT) for the response-locked maps.

#### **STATISTICAL ANALYSES**

Subjects' task performance was measured with error rate, median RT, and the signal detection measures *d* (discriminability) and log β (response bias) (Wickens, 2002). Effects on ERPs and oscillations were measured with average amplitude/PLF at electrodes and latency windows determined from the grand averages. Performance, ERP, and oscillation measures were analyzed with ANOVAs with the design Group (HC/SZ) X Stimulus (Square/No-Square) (X relevant electrode site factors). The Greenhouse-Geisser correction for inhomogeneity of variance (Keselman and Rogan, 1980) was applied for factors with more than two levels and is reflected in the reported *p-*values.

Correlations between task performance and electrophysiological effects of interest (averaged across electrode sites), demographic, and clinical variables (including symptom ratings) were calculated using the nonparametric Spearman's ρ (2-tailed, α = 0*.*05).

A statistical non-parametric mapping (SnPM) method based on the permutation test was used to find clusters of TF elements (time points in each frequency band) that reflected Group (HC/SZ) X Stimulus (Square/No-Square) interactions in the response-locked oscillations. The permutation test has several advantages over parametric statistical tests (Nichols and Holmes, 2001; Maris and Oostenveld, 2007). Most importantly for the analysis of TF data, the permutation test does not rely upon assumptions about the distribution of the data. Thus, it is more sensitive than parametric tests when the assumptions underlying the latter are not met (such as normality), which is likely for the PLF measure. Additionally, the permutation test provides control for multiple comparisons, since all the TF elements are permuted in parallel. In practice, however, we found it necessary to apply additional statistical criteria to control for multiple comparisons. Our statistical TF mapping method consisted of the following steps:


in the *t* distribution obtained by permutation, the *p*-value of that TF element would be 0.90.


# **RESULTS**

#### **TASK PERFORMANCE**

Error rate and RT data are displayed in **Figure 1B**. HC and SZ were in general highly accurate, with error rates of less than 3% per condition, and did not differ in overall error rate [*F(*1*,* <sup>29</sup>*)* = 1*.*31, *p* = 0*.*261]. HC made more errors for Square than No-Square stimuli, while SZ error rates did not differ between conditions [Group X Stimulus: *F(*1*,* <sup>29</sup>*)* = 6*.*20, *p <* 0*.*05; HC: *F(*1*,* <sup>13</sup>*)* = 8*.*02, *p <* 0*.*05; SZ: *F(*1*,* <sup>16</sup>*)* = 0*.*369, *p* = 0*.*552].

HC and SZ did not differ in overall RT [*F(*1*,* <sup>29</sup>*)* = 1*.*67, *p* = 0*.*207]. Both subject groups responded more quickly to Square than No-Square stimuli [Stimulus main effects: overall *F(*1*,* <sup>29</sup>*)* = 19*.*4, *p <* 0*.*001; HC: *F(*1*,* <sup>13</sup>*)* = 4*.*74, *p <* 0*.*05; SZ: *F(*1*,* <sup>16</sup>*)* = 16*.*7, *p <* 0*.*001]. This effect was larger for SZ than HC [Group X Stimulus: *F(*1*,* <sup>29</sup>*)* = 4*.*44, *p <* 0*.*05].

Analyses of *d*' indicated that both subject groups were able to easily discriminate between the Square and No-Square stimuli HC (mean ± s.e.m., 4.27 ± 0.16) and SZ (4.06 ± 0.22), with no difference between groups [*t(*29*)* = 0*.*719, *p* = 0*.*478]. Log β analyses revealed that HC (0.595 ± 0.176) were biased to make No-Square rather than Square responses [*t*-test vs. 0, *t(*13*)* = 3*.*38, *p <* 0*.*01], whereas SZ (−0*.*007 ± 0*.*114) were not biased to make either response. The difference in log β between groups was significant [*t(*29*)* = 2*.*97, *p <* 0*.*01].

The task performance data indicate that the HC and SZ groups performed the task equally well overall, and gave the Square stimuli priority in responding over the No-Square stimuli. However, HC made more misclassifications of Square stimuli. HC did not evince a speed/accuracy tradeoff, as there was no correlation between their RT and error rate Square minus No-Square effects (ρ = 0*.*433, *p* = 0*.*122).

#### **STIMULUS-LOCKED ERPs**

The grand average stimulus-locked ERPs are shown in **Figure 2A.** The average amplitude of the P1 component was measured at the electrodes PO9/10, PO7/8, PO3/4, and O1/2 in the 70–95 ms latency range. P1 amplitude did not differ between stimulus conditions [*F(*1*,* <sup>29</sup>*)* = 0*.*027, *p* = 0*.*871] but was reduced in SZ compared to HC [*F(*1*,* <sup>29</sup>*)* = 5*.*41, *p <* 0*.*05]. This P1 deficit varied across electrodes [Group X Electrode: *F(*3*,* <sup>87</sup>*)* = 3*.*53, *p <* 0*.*05], being significant at the electrode pairs PO3/4 [*F(*1*,* <sup>29</sup>*)* = 6*.*60, *p <* 0*.*05] and O1/2 [*F(*1*,* <sup>29</sup>*)* = 7*.*021, *p <* 0*.*05].

N1 component average amplitude was measured at the same electrodes as the P1 in the 110–200 ms latency range. N1 amplitude was greater for Square vs. No-Square stimuli overall [*F(*1*,* <sup>29</sup>*)* = 39*.*2, *p <* 0*.*000001] and in each group [HC: *F(*1*,* <sup>13</sup>*)* = 18*.*1, *p <* 0*.*001; SZ: *F(*1*,* <sup>16</sup>*)* = 21*.*1, *p <* 0*.*001] but did not differ between groups [*F(*1*,* <sup>29</sup>*)* = 0*.*135, *p* = 0*.*716]. Further examination revealed that the effect of stimulus on the N1 was actually due to an overlapping long-lasting negative shift in the 110–400 ms range at occipito-temporal electrodes [overall Stimulus: *F(*1*,* <sup>29</sup>*)* = 19*.*5, *p <* 0*.*001] that did not differ significantly between groups [Group X Stimulus: *F(*1*,* <sup>29</sup>*)* = 1*.*57, *p* = 0*.*220; HC: *F(*1*,* <sup>13</sup>*)* = 11*.*2, *p <* 0*.*01; SZ: *F(*1*,* <sup>16</sup>*)* = 7*.*36, *p <* 0*.*05]. The topography of this shift is displayed in the topographic maps in **Figure 2A.**

#### **RESPONSE-LOCKED ERPs**

A sample of the grand average response-locked ERPs are shown in **Figure 2B.** No effects of stimulus type or subject group were apparent at any electrode.

#### **STIMULUS-LOCKED OSCILLATIONS**

The grand average stimulus-locked TF maps averaged across the posterior channels (parietal, occipital, and occipito-temporal electrodes) are shown in **Figure 3A.** Vγ1 was evoked by Square and No-Square stimuli in both subject groups, with nonsignificantly higher PLF in SZ than HC. As has been reported

previously (e.g., Tallon-Baudry et al., 1997; Spencer et al., 2008a), Vγ1 displayed a bimodal topography (**Figure 3B**), with components at posterior and fronto-central electrode sites. Therefore, we analyzed the posterior and anterior components of Vγ1 separately.

The average PLF of the posterior Vγ1 component was measured at electrodes P1/2, P3/4, P5/6, P7/8, PO3/4, PO7/8, and O1/2 in the range of 50–130 ms and 25–58 Hz. There were no effects of Group or Stimulus, nor a Group X Stimulus interaction (*F*'s *<* 0.732, *p*'s *>* 0.399).

The average PLF of the anterior component of Vγ1 was measured at electrodes F1/z/2, FC3/1/z/2/4, C3/1/z/4, and CP1/z/2 in the same time and frequency ranges as the posterior Vγ1. There were no significant effects of Group, Stimulus, nor a Group X Stimulus interaction (*F*'s *<* 2.56, *p*'s *>* 0.121).

A second γ oscillation was observed following Vγ1 at posterior sites, which we termed "Vγ2." In HC, Vγ2 was evoked by Square stimuli in the range of 186–240 ms and 35–52 Hz. In SZ this oscillation appeared to occur in the nearly the same latency range (192–230 ms) but at a lower frequency (27–36 Hz) in the Square condition. The latency of this oscillation, and the similar topography to Vγ1, suggests that it could be an offset response. Measured in the above latency and frequency ranges at electrodes PO3/z/4, O1/z/2, and Iz, there was a trend for Vγ2 PLF to be increased in the Square relative to the No-Square condition over all subjects [*F(*1*,* <sup>29</sup>*)* = 3*.*95, *p* = 0*.*056]. The Group main effect and Group X Stimulus interaction were not significant (*F*'s *<* 0.918, *p*'s *>* 0.346). Since the Stimulus effect on Vγ2 in the SZ group might have been reduced by a small degree overlap with the trailing edge of Vγ1 in the No-Square condition, we analyzed Vγ2 separately in HC, for which there was no overlap with other oscillations. In HC, Vγ2 PLF was significantly increased for Square compared to No-Square stimuli [*F(*1*,* <sup>13</sup>*)* = 5*.*27, *p <* 0*.*05].

# **RESPONSE-LOCKED OSCILLATIONS**

The grand average response-locked TF maps, averaged across channels, are displayed in **Figure 4A.** The major effect that was apparent was a burst of high γ activity at ∼69–50 ms before the button press in the 74–99 Hz band in the Square condition for HC but not SZ.

SnPM was used to detect Group X Stimulus interactions in the response-locked oscillations (see Methods). In the positive (HC *>* SZ) Group X Stimulus TF map there were 2 clusters of significant TF elements (**Figure 4B**), both in the high γ band and shortly preceding the response. Cluster 1 spanned the time and frequency ranges of −65.2 to −53.5 ms and 95–97 Hz. Cluster 2 spanned the time and frequency ranges of −65.2 to −51.6 ms and 89–90 Hz. Both clusters had very similar fronto-central topographies (**Figure 4C**), suggesting that they might reflect the peaks of a broader high γ effect spanning a wider frequency range—namely, the high γ burst apparent in the HC Square data. In follow-up tests for each cluster (Bonferroni-corrected by 2 groups X 2 clusters), HC evinced significant Square *>* No-Square effects [Cluster 1: *t(*13*)* = 6*.*51, *p <* 0*.*0001; Cluster 2: *t(*13*)* = 5*.*92, *p <* 0*.*001], while SZ demonstrated significant No-Square *>* Square effects [Cluster 1: *t(*16*)* = −4*.*178, *p <* 0*.*01; Cluster 2: *t(*13*)* = −4*.*32,

*p <* 0*.*01] (see **Figure 5A**). No significant clusters were found in the negative (SZ *>* HC) Group X Stimulus map.

The predominantly fronto-central topographies of the response-locked γ clusters could be consistent with generators in motor and/or premotor cortex. In fact, decision-related activity has been found in premotor cortex (e.g., Hernández et al., 2002), and response hand-specific accumulation of decision information has been found in γ and β activity in premotor and motor cortex (Donner et al., 2009; de Lange et al., 2013). While the present response-locked γ effects do not seem to reflect pure motor-related activity, as they were modulated by stimulus condition, it is possible that they might be related to some degree to manual response generation. Response hand assignment was counterbalanced across subjects, which could have obscured any lateralization of the cluster topographies associated with particular response hand assignments. That is, approximately half of the subjects pressed the left/right buttons for Square/No-Square, and the other half of subjects had the opposite response hand assignment. Therefore, motor cortex activity in the grand average data should not be lateralized according to response hand.

To investigate the degree to which the response-locked γ effects might be related to motor activity, we tested whether the topography of the response-locked γ oscillation in the grand average data (**Figure 4A**) would be lateralized according to response hand assignment. We made new grand averages in which the electrodes were switched between the homologous sites in the hemispheres according to the subjects' response hand assignment, so that lateralized activity related to the responses would be made evident. That is, subjects who had the left/No-Square right/Square assignment had their electrodes switched (left hemisphere to right hemisphere and vice versa), so that lateralization of EEG activity related to response hand assignment would be identical across all the subjects. This resulted in only minor changes to the topography of the response-locked γ oscillation, and no increase in its lateralization (**Figure 4D**).

#### **CORRELATION ANALYSES**

To test our hypothesis that there would be correlations between response-locked γ effects and positive symptoms (particularly visual hallucinations, thought disorder, and/or disorganization),

condition is outlined (−69 to −50 ms, 74–99 Hz). **(B)** Group X Stimulus statistical map. The clusters of significant TF elements are indicated. **(C)** Electrodes contributing to each cluster. The percent area of each

greater thought disorder ratings had larger PLF for No-Square

hand assignment (see Results).

as in our previous study (Spencer et al., 2004), we investigated whether the response-locked γ Square minus No-Square effects in SZ were correlated with positive symptom rating scales. We focused on the 4 global ratings scales of the SAPS (Hallucinations, Delusions, Bizarre Behavior, and Formal Thought Disorder), and Bonferroni-corrected the correlations for 2 clusters X 4 symptom ratings. The Square minus No-Square PLF effects were averaged across the electrodes and TF windows contributing to the Group X Stimulus clusters. These analyses revealed that the Cluster 1 effect was correlated with the Global Rating of Formal Thought Disorder (ρ = −0*.*676, *p <* 0*.*05; **Figure 5B**). Since the direction of the Cluster 1 effect in SZ was for greater PLF in the No-Square than the Square condition, this correlation indicated that SZ with

than Square stimuli. To better understand the functional significance of the EEG effects (Square minus No-Square differences for ERP negative shift, Vγ2 in HC, and response-locked γ clusters), we conducted exploratory correlations between these effects and task performance effects (*d* , log β, and Square minus No-Square differences in RT and % error). The only correlations were found in HC, for which the RT advantage for Square vs. No-Square stimuli increased with the Vγ2 effect (ρ = −0*.*554, *p <* 0*.*05).

None of the significant Square minus No-Square effects in SZ was correlated with medication dosage (ρ's *<* |0.304|, *p*'s *>* 0.236).

**DISCUSSION**

scale in SZ.

Gestalt perception was associated with several electrophysiological effects: (1) a negative shift in the ERP that overlapped the N1 component for both HC and SZ; (2) enhanced PLF for Vγ2 in HC (with a possibly similar effect in SZ obscured by overlap with Vγ1); and (3) differences between Square and No-Square conditions in response-locked γ oscillations that differed in direction between HC and SZ. The subject groups' task performance patterns were largely similar. Consistent with our prior study (Spencer et al., 2004), the Gestalt effect on a response-locked γ oscillation was correlated with a positive symptom. Thus, while the exact results of Spencer et al. (2004) were not reproduced, the general pattern of results here is similar in several respects, especially with regard to the sensitivity of response-locked γ oscillations to positive symptoms in SZ.

Cluster 1 effect and the SAPS Global Rating of Formal Thought Disorder

#### **RESPONSE-LOCKED OSCILLATIONS**

We examined oscillatory activity that preceded and was phaselocked to RT because we reasoned that this approach would reveal oscillations that were more related to the perceptual decision, and hence to the perception of the Gestalt, than to the processing of stimulus features. In our previous study we used a sparse electrode array and analyzed oscillation effects that were apparent via visual inspection. Here we used a dense electrode array with a statistical mapping approach to find Gestalt effects.

In Spencer et al. (2004), illusory Kanisza squares elicited a response-locked γ oscillation in HC and a response-locked β oscillation in SZ. We hypothesized that the β oscillation in SZ was analogous to the γ oscillation in HC, as these oscillations had similar latencies and occipito-temporal topographies, but the oscillation in SZ occurred at a lower frequency. The Gestalt effect on the β oscillation was positively correlated with positive symptom ratings, specifically visual hallucinations, thought disorder, and conceptual organization. In the present study no such β effect was found. Instead, we found response-locked high γ effects that had increased PLF for Square compared to No-Square stimuli in HC, and the opposite sign (No-Square *>* Square PLF) in SZ. These effects had a fronto-central topography. In SZ, one of these high γ effects was correlated with thought disorder symptoms. Thus, the β oscillation in Spencer et al. (2004) and the high γ oscillation here seem to share a similar sensitivity to positive symptoms, particularly thought disorder.

The high γ response-locked oscillation occurred within 100 ms of the button press and had a mainly fronto-central topography. The latency of this oscillation indicates that it was elicited at a late stage of evidence accumulation, at approximately the same time that the motor response was issued by primary motor cortex (assuming a 50 ms interval between the onsets of electromyographic activity and manual response; Coles et al., 1985). The fronto-central topography of this oscillation is consistent with sources in motor or premotor cortex, where perceptual decisionrelated activity has been previously reported (e.g., Hernández et al., 2002).

In HC the high γ Gestalt effects consisted of increased PLF for Square compared to No-Square stimuli, while in SZ the effects were in the opposite direction. The increased γ synchronization to a Gestalt stimulus in HC would be consistent with the proposed role for γ oscillations in visual feature binding, but the topography (fronto-central as opposed to posterior) and latency (just preceding the manual response) of the response-locked γ effects do not support such a role. Likewise, the reversed pattern in SZ (No-Square *>* Square PLF) is not congruent with feature binding. The absence of correlations with task performance measures makes it difficult to posit the functional significance of these effects.

The failure to find the same γ/β effect as in Spencer et al. (2004) could be due to several factors. First, the effect might have been present on too few electrodes and/or was not significant enough to pass the statistical thresholds used in the SnPM procedure. Second, the brief stimulus duration in the present study could have resulted in a different pattern of activity in brain regions involved in evidence accumulation to make the Square/No-Square decision. Third, the use of the average reference here (as opposed to averaged mastoids in the previous study) could have altered the topography of the oscillation.

#### **ERPs AND VISUAL-EVOKED γ OSCILLATIONS**

Previous studies of illusory contour perception have found a negative ERP shift that overlaps the N1 component for illusory contours compared to control stimuli, which has been localized to the lateral occipital complex (e.g., Murray et al., 2002; Altschuler et al., 2012;. Here this shift was found for HC and SZ and did not differ between them, consistent with the findings of Foxe et al. (2005) and Spencer et al. (2004). In addition, we note that the amplitude of the P1 component was reduced in SZ irrespective of the stimulus, consistent with several prior findings (reviewed in Javitt, 2009).

Several prior studies have reported evidence that visual γ oscillations are generally impaired in schizophrenia (reviewed in Spencer, 2008 and Tan et al., 2013). In earlier studies our group found that Vγ1 was elicited by illusory Square but not No-Square stimuli in HC, whereas in SZ, neither stimulus elicited this oscillation (Spencer et al., 2003, 2004). We also reported that SZ had reduced Vγ1 PLF in a visual oddball task with letter stimuli (Spencer et al., 2008a). Krishnan et al. (2005) found that the power of the visual steady-state response (VSSR) to β and γ stimulation was reduced in SZ. Likewise, Grützner et al. (2013) reported reductions in SZ of the power of visual-evoked onset and offset γ responses, which are likely to have been analogous to the oscillations we have termed Vγ1 and Vγ2. (The power of a sustained induced γ oscillation was also reduced in SZ.) These findings implied that SZ have a basic deficit in generating γ oscillations to visual stimuli.

However, other evidence suggests that deficits in visual γ oscillations in schizophrenia may not be universal and may depend upon additional factors. In the present study, both types of stimulus in both subject groups evoked Vγ1 as well as Vγ2. Furthermore, Wynn et al. (2005) reported a SZ deficit in γ activity evoked by masked but not unmasked stimuli. In first-episode SZ, Sun et al. (2013) did not find a reduction in the power of a visual-evoked onset γ oscillation, although the power of an offset γ oscillation and sustained induced γ were reduced. Finally, Riecanský et al. (2010) ˇ found that the PLF of the 40 Hz VSSR was increased in SZ. Therefore, we must conclude that SZ do not have a basic deficit in generating γ oscillations to visual stimuli, and the visual γ deficits that we have previously reported may be due in part to other problems such as dysfunctional topdown processes. The reason for the difference in Vγ1 results between this study and the earlier ones is not clear. Stimulus duration is unlikely to be a factor, since Spencer et al. (2008a) also utilized a brief stimulus duration (100 ms). At the present we do not have any explanation for the difference in findings, but we will present more data on visual-evoked γ in SZ in future papers.

In the present study we also found a later stimulus-evoked γ oscillation, Vγ2, which had a similar topography as Vγ1, suggesting that it might be an offset response. In HC, Vγ2 PLF was greater in the Square compared to the No-Square condition. This effect was correlated with the difference in RT between stimulus conditions, suggesting that Vγ2 played a functional role in Gestalt perception. Both the Gestalt effect and the correlation with RT were found previously for Vγ1 (Spencer et al., 2003). A similar Gestalt effect might have been present in the SZ group but overlap with the trailing end of Vγ1 prevented us from determining this conclusively. We note that Grützner et al. (2013) reported that Gestalt stimuli (Mooney faces) evoked greater power compared to non-Gestalt stimuli in a visual γ offset response. This Gestalt effect was reduced in the SZ. It is possible that the same effect was observed in both Grützner et al. and the present study (although a reduced Gestalt effect in SZ was not found here).

Vγ2 occurred at the same latency in SZ as in HC but in a lower frequency range. As noted above, a similar difference in the frequency of a response-locked oscillation between HC and SZ was found in Spencer et al. (2004). In that study, Square stimuli elicited a response-locked γ oscillation in HC, whereas in SZ, Square stimuli elicited a response-locked β oscillation. The reason for the difference in frequency between subject groups is not clear.

## **CORRELATIONS BETWEEN γ/β OSCILLATIONS AND POSITIVE SYMPTOMS**

In several studies we have reported positive correlations between γ/β measures and positive symptoms of schizophrenia: (1) the β response-locked oscillation Gestalt effect in Spencer et al. (2004), (2) the 40 Hz harmonic of the 20 Hz ASSR (Spencer et al., 2008b), and (3) a left auditory cortex 40 Hz ASSR source (Spencer et al., 2009). While the particular oscillations showing these correlations differ, the general pattern we have found is that increased γ/β is associated with psychosis, even when SZ show a deficit in the oscillation measure compared to HC at the group level. Hence, increased symptomatology has been paradoxically associated with more "normal" oscillatory activity. One possible explanation for these findings is that high-frequency oscillatory activity is downregulated in schizophrenia as a response to neural circuit dysfunctions such as hyperexcitability (e.g., Hoffman and Cavus, 2002), for example as a result of antipsychotic treatment. The failure of this downregulation of γ/β activity is directly related to the persistence of psychotic symptoms, and manifests as across-patient correlations between psychotic symptoms and γ/β effects.

Here we found that a γ Gestalt effect in SZ was correlated with thought disorder symptom ratings, similar to the β Gestalt effect in Spencer et al. (2004). [In support of this finding, we note that Uhlhaas et al. (2004) found relationships between measures of perceptual integration and thought disorder in schizotypal individuals.] However, the direction of the correlation here was negative rather than positive, as the effect was measured as Square minus No-Square PLF, and the SZ had greater PLF in the No-Square than the Square condition. But since SZ had greater PLF in the No-Square condition, and this activity increased relative to PLF in the Square condition, the SZ did show an increase in γ related to positive symptomatology. However, in absolute terms, the degree of abnormality (No-Square *>* Square PLF) increased with thought disorder symptomatology, such that the patients with the highest thought disorder ratings had the largest No-Square *>* Square PLF difference. Thus, this effect differs from the above positive correlation findings, in which the most "normal" activity was associated with the most symptoms.

#### **CONCLUSIONS**

In this study we attempted to replicate previous findings of abnormal oscillatory indices of visual Gestalt perception in individuals with schizophrenia. While the previous findings were not exactly replicated, in particular the findings of visual-evoked γ deficits in SZ, several aspects were found again. Most notably, a response-locked high-frequency Gestalt effect was found that was correlated with thought disorder symptoms. This finding supports the hypothesis that high-frequency oscillations are sensitive to aspects of psychosis.

#### **ACKNOWLEDGMENTS**

This work was supported by a US Department of Veterans Affairs Merit Review grant (CX000154), US National Institutes of Health grants R03 MH076760 and R01 MH080187, and a NARSAD Young Investigator Award to Kevin M. Spencer.

#### **REFERENCES**


**Conflict of Interest Statement:** In the past 3 years, Kevin M. Spencer received consultation fees from Galenea Inc. and Bristol-Myers Squibb. Kevin M. Spencer also received a Research and Development Grant from Galenea Inc. that was administered through the Boston VA Research Institute. The other author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 11 May 2013; accepted: 17 January 2014; published online: 04 February 2014.*

*Citation: Spencer KM and Ghorashi S (2014) Oscillatory dynamics of Gestalt perception in schizophrenia revisited. Front. Psychol. 5:68. doi: 10.3389/fpsyg.2014.00068 This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2014 Spencer and Ghorashi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# High-frequency neural oscillations and visual processing deficits in schizophrenia

#### *Heng-Ru May Tan1 \*, Luiz Lana2,3,4 and Peter J. Uhlhaas 1,3,4*

*<sup>1</sup> Institute of Neuroscience and Psychology, College of Science and Engineering and College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK*

*<sup>2</sup> Brain Institute, Federal University of Rio Grande do Norte, Natal, Brazil*

*<sup>3</sup> Department of Neurophysiology, Max-Planck Institute for Brain Research, Frankfurt, Germany*

*<sup>4</sup> Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Frankfurt, Germany*

#### *Edited by:*

*Steven Silverstein, University of Medicine & Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Kevin M. Spencer, VA Boston Healthcare System/Harvard Medical School, USA Shahab Ghorashi, Harvard Medical School, USA*

#### *\*Correspondence:*

*Heng-Ru May Tan, Centre for Cognitive Neuroimaging, Institute of Neuroscience and Psychology, College of Science and Engineering and College of Medical, Veterinary and Life Sciences, University of Glasgow, 58 Hillhead Street, Glasgow G12 8QB, UK e-mail: Heng-RuMay.Tan@ glasgow.ac.uk*

Visual information is fundamental to how we understand our environment, make predictions, and interact with others. Recent research has underscored the importance of visuo-perceptual dysfunctions for cognitive deficits and pathophysiological processes in schizophrenia. In the current paper, we review evidence for the relevance of high frequency (beta/gamma) oscillations towards visuo-perceptual dysfunctions in schizophrenia. In the first part of the paper, we examine the relationship between beta/gamma band oscillations and visual processing during normal brain functioning. We then summarize EEG/MEG-studies which demonstrate reduced amplitude and synchrony of high-frequency activity during visual stimulation in schizophrenia. In the final part of the paper, we identify neurobiological correlates as well as offer perspectives for future research to stimulate further inquiry into the role of high-frequency oscillations in visual processing impairments in the disorder.

**Keywords: schizophrenia, high-frequency neural oscillations, visual perception, neurobiology, evoked and induced neural activity, neural synchrony**

# **DYSFUNCTIONS IN VISUAL PERCEPTION IN SCHIZOPHRENIA**

Disturbances in visual perception were for a long time considered relatively unimportant in the understanding of schizophrenia (ScZ) compared to the more striking clinical presentation of hallucinations and delusions. Bleuer summarized this view as follows: "Sensory responses to external stimulus are quite normal. To be sure, the patients will complain that everything appears to be different. However, this strangeness is usually attributable to a deficit in customary associations and particularly to an alteration of emotional emphasis." (Bleuler, 1969; page 76). Similarly, Kraepelin (1919) concurred that "[P]erception of external impressions in dementia praecox is not usually lessened to any great extent as far as a superficial examination goes." (Kraepelin, 1971; page 5).

Following evidence from phenomenological research which indicated profound alterations in perceptual experience (see Uhlhaas and Mishara, 2007 for a review), an increasing number of studies began to investigate sensory processing experimentally (Place and Gilmore, 1980). Since then, a large body of evidence has accumulated that has highlighted impaired visual processing as a core deficit in schizophrenia (Klosterkötter et al., 2001; Javitt, 2009). Such dysfunctions involve the discrimination of orientation, motion, and object size (e.g., Butler and Javitt, 2005; Butler et al., 2007, 2008; Chen, 2011), which have been related to the magnocellular pathway because of reduced sensitivity to stimuli with low spatial frequency (Schechter et al., 2003; Butler and Javitt, 2005).

Moreover, ScZ-patients show reduced contextual influences in relationship to contrast (Yang et al., 2013), motion (Tadin et al., 2006), orientation (Yoon et al., 2010) as well as during contourintegration (Uhlhaas et al., 2006b), which could underlie impairments in perceptual organization (Uhlhaas and Silverstein, 2005). Additional visual processing deficits in ScZ have been revealed by masking paradigms (Green et al., 2011) which have highlighted longer intervals between the target and mask stimuli for accurate identification of targets (e.g., Green et al., 1994, 1999). Importantly, visual dysfunctions have been linked to impairments in higher cognitive functions (Javitt, 2009), such as working memory (Haenschel et al., 2009). Moreover, there is evidence to suggest that changes in visual perception are related to more complex features of the disorder, such as the development of delusions and changes in self-experiences (Uhlhaas and Mishara, 2007).

Data on abnormal visual functions in ScZ is consistent with evidence on anatomical abnormalities as revealed by postmortem studies (Selemon et al., 1995; Selemon and Goldman-Rakic, 1999; Dorph-Petersen et al., 2007) as well as magnetic resonance (MR) and Diffusion Tensor Imaging (DTI) studies (Staal et al., 2000; Clasen et al., 2003; Arnone et al., 2009; White et al., 2011; Whitford et al., 2011b). These findings suggest that in addition to abnormalities in fronto-temporal regions, alterations in anatomical parameters extend to early visual areas. More recently, electro/magnetoencephalography (EEG/MEG) and functional magnetic resonance imaging (fMRI) have disclosed corresponding deficits in neural responses during visual stimulation (Spencer et al., 2003; Wynn et al., 2005; Uhlhaas and Singer, 2006; Uhlhaas et al., 2006a,b; Yeap et al., 2006). Specifically, studies assessing event-related potentials (ERPs) have demonstrated impairments during early and later visual processing stages in ScZ (**Box 1**).

Given that the visual system has been extensively explored through anatomy, electrophysiology and neuroimaging, detailed examination of visual dysfunctions in ScZ may allow insights into the underlying neurobiological correlates. In the following review, we will focus on the role of high-frequency neural oscillations because considerable evidence exists on the role of beta (13– 25 Hz)/gamma (25–200 Hz) band activity in visual processing as well as their potential involvement in the pathophysiology of ScZ. We will first examine the role of high-frequency neural activity

#### **Box 1 | Event-related potentials (ERPs), visual perception, and Schizophrenia**

*ERP* waveforms are commonly derived in EEG studies that investigate the neurophysiological mechanisms underlying sensory perception. ERP waveforms consist of a series of transient positive and negative voltage deflections that are time-locked to stimulus onset. These transient fluctuations in ERP waveform polarity are conventionally extracted as basic components and named by their polarity in conjunction with either their latency or ordinal position, relative to stimulus onset. In general, early components (e.g., those occurring before ∼200 ms) are thought to reflect early sensory processing while higher cognitive processes are related to later components.

*N80*—an initial negative deflection that peaks ∼70–90 ms post stimulus onset (Di Russo et al., 2007)—is considered the earliest visual ERP component and thought to be mainly driven by parvocellular (P) input but likely with small influence from the magnocellular (M) pathway in response to visual stimuli consisting of higher contrast levels (Foxe et al., 2008). In a recent investigation, using a range of visual stimuli that theoretically bias M or/and P pathways using systematic manipulation of stimulus luminance, chromatic contrasts and flicker, Núñez et al. (2013) demonstrated that the occipital N80 component in early-onset (but not adult-onset) ScZ was significantly lower in amplitude in response to stimuli that involve both M and P pathways. They also observed significantly prolonged onset of N80 latency in adult-onset ScZ. In contrast, the N80 amplitude in response to isolated P- and M-biased stimuli was comparable in both healthy controls and ScZ patients. The finding is suggestive of a deficit in M-priming on the P pathway in early-onset ScZ.

*P1*—a component with positive deflection ∼100 ms post visual stimulus onset—is known to involve the dorsal and ventral visual streams (Martinez et al., 1999; Di Russo et al., 2002). In response to visual stimuli (e.g., motion, spatial, low contrast) that bias the magnocellular pathway, the occipital P1 amplitude is commonly reported to be markedly decreased in patients with ScZ (e.g., Doniger et al., 2002; Schechter et al., 2005; Butler et al., 2007; Lalor et al., 2012; Núñez et al., 2013; however, see e.g., Johnson et al., 2005; Wynn et al., 2008, also mentioned below). Prominent P1 amplitude reduction was also observed in ScZ patients when they engaged in visual tasks involving illusory contour processing (Foxe et al., 2005) or fragmented object recognition (Doniger et al., 2002), and this reduction paralleled the weaker scalp activations over their lateral and posterior occipital areas.

*N1*—a negative component peak that manifests ∼150–200 ms—is thought to predominantly reflect ventral stream processing (e.g., Doniger et al., 2002). Many studies that have assessed the N1 component, e.g., using illusory contour or fragmented contoured stimuli for object recognition, have reported comparable N1 amplitudes between ScZ patients and controls (e.g., Foxe et al., 2001; Doniger et al., 2002; Foxe et al., 2005), suggesting that parvocellular-mediated ventral stream processing is largely unaffected. However, studies that investigated face processing in ScZ patients have demonstrated pronounced reduction in their N170 amplitude in response to face vs. building stimuli (Herrmann et al., 2004b; Turetsky et al., 2007). Similar prominent reduction in N150 amplitude was observed in ScZ patients engaged in local vs. global visual perceptual tasks, and the amount of amplitude decrease in response to global stimuli correlated with corresponding performance accuracy and response times (Johnson et al., 2005). Further evidence of N1 amplitude reduction (∼200 post stimulus onset) is recently demonstrated in the fine-grain visual-masking discrimination task employed by (Plomp et al., 2013). Intriguingly, the extra time required by ScZ to reach normal discrimination performance levels did not alleviate the pronounced N1 amplitude reduction. Instead, as source analysis revealed, the discrimination difficulties are likely to be related to the significantly weaker parietal and lateral occipital activity in ScZ patients.

*N250*—a negative component that peaks ∼250 ms over fronto-central electrode sites—is considered sensitive to the emotional content of faces (Streit et al., 1999, 2001). Some studies have found reduced N250 amplitude in ScZ patients (e.g., Streit et al., 2001; Wynn et al., 2008) with normal N170 suggesting emotional information decoding deficits, while others have found the opposite (e.g., Johnson et al., 2005; Turetsky et al., 2007) suggesting that facial feature encoding is impaired rather than emotional information decoding.

*NCL*—a negative component that manifests ∼270–320 ms observed during visual tasks involving perceptual closure—is characterized by bilateral occipito-temporal scalp topography (Doniger et al., 2002) and is thought to reflect effortful extraction of object identity (Foxe et al., 2005). Significantly reduced *NCL* amplitude has been shown to be preceded by a normal N1 but prominently reduced P1 amplitude in ScZ patients (Doniger et al., 2002; Foxe et al., 2005). This observation has led to a view that the initial stages of visual ventral stream processes are unaffected in ScZ patients but the later stages of ventral stream processing involving object recognition are likely affected by indirect magnocellular-mediated dorsal stream inputs (e.g., Merigan and Maunsell, 1993; Nowak and Bullier, 1997; Schroeder et al., 1998) into the visual areas along the ventral stream (e.g., lateral occipital cortex).

*P300*—a positive component that peaks ∼300–900 ms post stimulus onset. Unlike earlier potentials, it is supposed to be an endogenous component which reflects stimulus context and levels of attention and arousal. The auditory P300 has consistently been shown to be impaired both in amplitude and latency (Bramon et al., 2004) while evidence for a dysfunctions during visual processing are less consistent (Ford, 1999); but see recent findings by Oribe et al. (2013) on prodromal and first-episode ScZ-patients.

during normal visual perception emphasizing work from invasive and non-invasive electrophysiology followed by an overview of studies with EEG/MEG that have examined alterations in high-frequency oscillations in ScZ. In the final section, we will discuss potential mechanisms which could account for abnormal beta/gamma oscillations in ScZ as well as provide recommendation for future research.

# **HIGH-FREQUENCY OSCILLATIONS AND VISUAL PROCESSING**

#### **INVASIVE ELECTROPHYSIOLOGY**

The involvement of gamma-band oscillations in sensory processing was first described by Adrian and colleagues in the 1940s (Adrian, 1950). Local field potential recordings from the olfactory bulb of anesthetized cats, rabbits, and hedgehogs showed pronounced oscillations in the 40–60 Hz frequency range. Subsequently, Freeman and colleagues (Bressler and Freeman, 1980; Freeman and Skarda, 1985) reported correlations between 35 and 85 Hz activity and olfactory perception, suggesting that gamma-band oscillatory modulations are involved in information coding in the olfactory system (Freeman, 1991).

Crucial evidence for a mechanistic role of gamma-band activity in visual perception and cortical computations was obtained by Singer and colleagues in the late 1980s (Singer, 1999). Specifically, Gray et al. (1989) showed that action potentials generated by cortical cells are phase-locked to the oscillatory gamma rhythm and consequently neurons aligned their discharges with high temporal precision. In its original formulation, the "Binding by Synchrony hypothesis" (BBS; Singer, 1999) proposed that ensembles of neurons that preferentially respond to features of the same object should fire synchronously, whereas these same neurons should not synchronize their firing to features belonging to other objects or to the background. Over the years, this hypothesis has gained substantial attention (for critical reviews see Gray, 1999; Shadlen and Movshon, 1999; Singer, 1999; Uhlhaas et al., 2009).

There is, however, conflicting evidence for the BBS in the primate primary visual area (V1) with some studies failing to find evidence for a relationship between binding of stimulus features and synchronous gamma-band activity (e.g., Lima et al., 2010). Given the large number of visual areas in the primate brain (Van Essen and Gallant, 1994), it is conceivable that binding through oscillatory mechanisms occurs in higher visual areas. Candidate brain regions would be structures that have been shown to express strong gamma oscillations in response to visual stimulation, such as the middle temporal cortex (MT) and V4 areas (e.g., Kreiter and Singer, 1996; Fries et al., 2001; but see also Thiele and Stoner, 2003; Palanca and DeAngelis, 2005). Nonetheless, it is important to note that the temporal and spatial scales for binding might be smaller than previously assumed and therefore even V1 remains as a viable candidate for binding (Fries et al., 2007; Havenith et al., 2011; Nikolic et al., 2013 ´ ). These observations highlight the need to employ more sophisticated analysis techniques for the detection of transient signals that may be important for BBS.

In addition to stimulus parameters (see **Box 2**), the amplitude, and frequency of high-frequency oscillations in visual cortices can also be influenced by cognitive variables, such as attention. Initial evidence was provided by Fries and colleagues (Fries et al., 2001) who showed that 35–90 Hz activity in macaque visual area V4 strongly increased when behaviorally relevant stimuli were within the focus of attention. More recently, the same group demonstrated that spatial attention can also result in a shift to higher gamma-band frequencies in V1 (Bosman et al., 2012). Similarly, Lima et al. (2010) demonstrated using plaid stimuli that selective attention to one of the directional components of the plaid pattern affected gamma-band power in a manner that resembled the power (and frequency) modulation when the actual contrast of the stimulus was increased. Additionally, V1 gamma spectral power in macaques was shown to increase with temporal expectancy for behaviorally relevant events.

#### **EEG-MEG STUDIES**

Following the initial findings in both anaesthetized and awake animals on the potential relationship with visual processing (Singer and Gray, 1995), high-frequency oscillatory responses to visual stimuli have also been documented in EEG/MEG and electrocorticographic (ECoG) in humans (Sauvé, 1999; Tallon-Baudry and Bertrand, 1999; Lachaux et al., 2005; Tallon-Baudry, 2009; Martinovic and Busch, 2011). Broadly three different categories of high-frequency responses can be distinguished (**Box 3**).

Evoked high frequency oscillatory responses are typically observed ∼70–120 ms post stimulus with an occipital topography (e.g., Martinovic and Busch, 2011). Sources of evoked gamma activity during simple visual stimulus perception or object recognition have been localized to primary visual (Muthukumaraswamy et al., 2010), lateral occipital-temporal and inferior temporal cortical areas (Gruber et al., 2006). Amplitude and phase-locking of evoked high-frequency oscillations are modulated by stimulus properties. Corroborating invasive studies (**Box 2**), human neuroimaging research have also reported beta and gamma-band activity amplitude increases with contrast (Sannita et al., 1995; Schadow et al., 2007), stimulus duration, and size (Perry et al., 2013). In addition, spatial frequency modulates the power of high-frequency activity non-monotonically (Sannita et al., 1995; Tzelepi et al., 2000) and eccentricity decreases beta/gamma-band power (Busch et al., 2004; Fründ et al., 2007).

Due to their latency and topography, evoked high-frequency responses are likely to reflect feed forward driven responses (e.g., Butler and Javitt, 2005; Tobimatsu and Celesia, 2006; Martinovic and Busch, 2011). Early studies suggested that both amplitude and latency of evoked high-frequency activity were largely unaffected by experimental manipulations involving attention (e.g., Tallon et al., 1995; Tallon-Baudry et al., 1996, 1997). However, more recent findings (Herrmann et al., 1999; Frund et al., 2008) have challenged this view through demonstrating that top-down factors can impact on evoked gamma-band activity as well (e.g., Chaumon et al., 2009).

Following the link between binding of stimulus of elements into coherent representations and gamma-band oscillations in invasive recordings (Gray et al., 1989), several EEG

#### **Box 2 | Stimulus parameters and high-frequency neural oscillations**

High-frequency oscillations are modulated by several important parameters, such as color, contrast, presentation eccentricity, orientation, and speed. The complexity of natural images makes it difficult to systematically explore the influence of any particular features on brain activity. Thus, simplified stimuli that can be parametrically changed over a feature space are typically employed. In particular, gratings are commonly used because they produce strong responses in the gamma frequency range. A brief overview of these parameters and their observed influence on high frequency oscillations are listed below.

#### **SIZE**

The effect of grating size on the activity recorded from the macaque primary visual area (V1) is such that bigger gratings generate stronger oscillations, lower peak frequency, and decreased firing rates (Gieselmann and Thiele, 2008; Jia et al., 2013a). Human MEG studies replicated the positive relationship between stimulus size and gamma frequency power, but failed to reproduce the effect of size modulating peak gamma frequency (Perry et al., 2013).

#### **POSITION**

The frequency of gamma oscillatory response is dependent on the apparent eccentricity of the stimulus. Centrally presented gratings tend to generate higher frequencies than stimuli presented peripherally (Lima et al., 2010).

#### **CONTRAST**

Gratings with higher contrast are associated with higher firing rates and increased gamma peak-frequency (Ray and Maunsell, 2010). The strength of gamma-band oscillations initially increases with contrast but if the contrast is too high there is a tendency for the oscillations to reduce in power (Ray and Maunsell, 2010; Jia et al., 2013b; Roberts et al., 2013).

#### **SPEED**

Gamma-band oscillations vary consistently with stimulus-velocity of bars (Gray et al., 1990) and gratings (Friedman-Hill et al., 2000; Lima et al., 2011). In humans, static gratings generate lower peak gamma frequencies than moving gratings (Swettenham et al., 2009; Muthukumaraswamy and Singh, 2013).

#### **SPATIAL FREQUENCY**

The spatial frequency of gratings is known to modulate gamma power (Adjamian et al., 2004) and firing rates (Lima et al., 2010) following an inverted U relationship. Human neuroimaging data suggest that the peak frequency of gamma response is tuned to a narrow band of spatial frequency (2–4 cycles per degree; cpd), peaking at 3cpd (Adjamian et al., 2004); a finding not currently observed in monkey neurophysiology (Lima et al., 2010).

#### **ORIENTATION AND DIRECTION**

Animal neurophysiology has shown that gamma power and frequency are tuned to stimulus orientation and the direction of stimulus motion (Feng et al., 2010; Jia et al., 2013b). However, these effects have not yet been demonstrated in humans.

#### **NOISE LEVELS**

Progressively adding noise over a high contrast grating reduces the amplitude of gamma-band oscillations and their peak frequency without altering the average firing rates (Jia et al., 2011, 2013b).

#### **LUMINANCE PROFILE**

The luminance profile of a grating also influences gamma responses, with square waves generating more gamma power than sinusoidal gratings (Muthukumaraswamy and Singh, 2008).

#### **STIMULUS TYPE**

Gratings formed by concentric circles have been shown to produce higher gamma power compared to regular gratings formed by straight parallel elements (Muthukumaraswamy and Singh, 2013).

#### **STIMULUS COMPLEXITY**

Increases in stimulus complexity may lead to dramatic reductions in gamma-band power and also to changes in peak frequency (Lima et al., 2010).

#### **COLOR**

Pure color isoluminant gratings have been shown to produce undetectable gamma oscillations in human MEG recordings, that otherwise manifested strong gamma responses to luminance contrast gratings (Adjamian et al., 2008).

and MEG-studies have also examined the role of gamma-band oscillations during perceptual organization (Lutzenberger et al., 1995; Revonsuo et al., 1997; Keil et al., 1999; Spencer et al., 2003; Grützner et al., 2010), demonstrating increased amplitude and synchrony of gamma-band activity during the construction of coherent object representations. More recently, intracranial EEG data have complemented this evidence (Lachaux et al., 2005).

#### **Box 3 | Measures of high-frequency oscillations**

Neural oscillations can be characterized by their frequency, amplitude, and phase. These characteristic features are derived through the time-frequency decomposition of electrophysiological signals by means of a Fourier, Wavelet or Multi-Taper analyses. These techniques estimate the strength (amplitude) and phase (time-variant angle) of the signal at a particular frequency range. A signal's phase consistency across trials, described as the phase-locking index, can be computed from its phase information (Lachaux et al., 1999, 2003). Apart from inter-trial oscillatory phase-locking, the same approach can also be used to assess the inter-areal synchrony of neural oscillations (Lachaux et al., 1999; Rodriguez et al., 1999; Gross et al., 2001; Varela et al., 2001; Siegel et al., 2008; Hipp et al., 2011). Importantly, this phaselocking index provides an estimation of neural synchrony irrespective of the amplitude of the oscillatory signal. This estimate of synchrony between signals is distinct from spectral coherence measures in which amplitude and phase information are intermixed (Gross et al., 2001).

Several different parameters of high-frequency activity can be distinguished which are frequently employed in the analysis of electrophysiological data (Tallon-Baudry et al., 1996; Roach and Mathalon, 2008).


While induced oscillations are strongly enhanced by top-down factors (Vidal et al., 2006; Melloni et al., 2007), several studies have indicated that basic stimulus parameters, such as orientation (Edden et al., 2009), spatial frequency (Hadjipapas et al., 2007; Perry et al., 2013), luminance (Adjamian et al., 2008), and motion (Swettenham et al., 2009) also influence the occurrence of high-frequency activity. These findings, thus, challenge a simple dichotomy between evoked and induced activity. Moreover, recent findings demonstrated that sub-bands of low (∼30–60 Hz) and high (∼70–120 Hz) gamma-band oscillations are flexibly recruited by both feed-forward and feedback processes. For example, 30–60 Hz activity showed increases with initial unconscious associative learning of target-specific context in a search-task while 70–120 Hz oscillations occurred regardless of stimulus contexts (Chaumon et al., 2009). Likewise, amplitude of low gamma-band activity increases with conscious visual awareness in contrast to attention-related gamma band activity at higher frequencies (Wyart and Tallon-Baudry, 2008). The close association between the modulation of both low and high gamma band activity and cognitive processes further suggests that different gamma band frequencies could support the dynamic formation of distinct assemblies that underlie specific behavioral or cognitive function through "multiplexing" neural signal transmission (Vidal et al., 2006; Wyart and Tallon-Baudry, 2008).

#### **ALTERATIONS IN HIGH-FREQUENCY NEURAL OSCILLATIONS DURING VISUAL PROCESSING IN SCHIZOPHRENIA**

The wealth of research highlights that high-frequency neural oscillations are involved in perceptual processing during normal brain functioning (Herrmann et al., 2004a; Tallon-Baudry, 2009; Martinovic and Busch, 2011). It is therefore conceivable that disturbances in the amplitude and synchrony of beta/gammaband oscillations may have an important role in visual dysfunctions in ScZ. Indeed, a growing number of studies exploring this relationship have employed a range of visual tasks and assessed the integrity of the evoked and induced neural responses using different oscillatory parameters (see **Table 1** for an overview and also **Figure 1**).

#### **SSVEPs**

Research investigating steady-state visually evoked potentials (SSVEPs) have observed reduced amplitude-modulation to repetitive stimulation at high but also at lower-frequencies in patients with schizophrenia relative to healthy controls. Krishnan et al. (2005) investigated SSVEPs to photic stimulation at frequencies from 4 to 40 Hz in EEG-recordings and reported decreased occipital amplitude modulation at 17, 23, and 30 Hz stimulation (**Figure 1**). In addition, higher "background noise," which was defined as averaged power of neural activity 1 Hz above and below the photic stimulation frequency, was observed at frequencies 4–20 Hz in ScZ-patients. The data from SSVEPs parallel findings from auditory entrainment experiments suggesting a basic impairment of cortical circuits to support high-frequency activity in ScZ. In contrast to visual SSVEPs, however, auditory entrainment impairments have been predominantly demonstrated at 40 Hz frequency stimulation (Kwon et al., 1999). Although more recent data have also demonstrated entrainments deficits at 80 Hz as well as at theta-frequency ranges (Hamm et al., 2011).

#### **EVOKED ACTIVITY**

Several studies have examined the integrity of evoked oscillations in ScZ using a variety of tasks. Backward masking paradigms are often used to assess early visual processing in ScZ. Given that basic features of any visual stimulus need to be integrated into a percept along the visual processing pathways, the effects of target percept masking could occur through the process of "integrating" the mask percept with the target percept, or through the process of "interrupting" the identification of target perception at a later

#### **Paradigm Imaging Oscillatory Parameter Main References modality measure assessed findings** Steady state stimulation EEG Evoked Amplitude 17–30 Hz range amplitude decrease over occipital electrodes Krishnan et al., 2005 Backward masking EEG Evoked Amplitude 30–40 Hz range amplitude decrease across electrodes Wynn et al., 2005 EEG Evoked Amplitude/Latency 30–35 Hz range amplitude decrease over parieto-occipital electrodes Green et al., 2003 Oddball detection EEG Evoked Inter-trial phase-locking Decreased 30–38 Hz range phase-locking over parieto-occipital electrodes Spencer et al., 2008 Illusory square EEG Evoked Inter-trial phase-locking Decreased 28–35 Hz range phase-locking over parieto-occipital electrodes Spencer et al., 2004 Induced Inter-trial phase-locking i) Decreased 30–45 Hz range phase-locking ii) Decrease in peak phase-locking frequency (at 22–26 Hz cf. controls) over occipital and parietal electrodes in response-locked analysis Evoked Inter-trial phase-locking i) Decreased 24–48 Hz phase-locking ii) Decrease in peak phase-locking frequency in response to 'No-Square' stimuli over occipital and central electrodes Spencer et al., 2003 Induced Inter-sensor phase-coherence Long-range 20–26 Hz range decrease in phase-locking Inter-hemispheric decrease in peak phase-coherence frequency (37–44 Hz cf. 48–57 Hz) particularly over posterior electrodes Mooney faces MEG Evoked Amplitude i) 25–140 Hz range amplitude decrease, especially pronounced in the 60–140 Hz range Grützner et al., 2013 ii) 25–60 Hz range fronto-central amplitude increase Inter-trial phase-locking Decreased 60–140 Hz range phase-locking Induced Amplitude 60–140 Hz range amplitude decrease over occipital sensors EEG Induced Amplitude Insignificant difference in the 40–70 Hz range across electrodes Uhlhaas et al., 2006a Inter-trial phase-locking Decreased and delayed onset latency of 20–55 Hz range phase-locking Inter-sensor phase-coherence Decreased 20–30 Hz phase-coherence between fronto-temporal and parieto-occipital electrodes

**Table 1 | Summary of EEG/MEG studies investigating high-frequency neural oscillations in patients with chronic schizophrenia and healthy controls during visual perceptual tasks.**

*Main findings of these studies are reported in brief, highlighting the frequency range of significant effects observed in patients cf. healthy controls.*

stage of visual processing, or even via the process in which the target percept is "substituted" by that of the mask through a fastacting process (Green et al., 2011). ScZ patients and unaffected siblings require longer inter-stimuli-intervals (ISI) between the target-mask stimuli for accurate feature identification of briefly presented targets (Green et al., 1994, 1997; Kéri et al., 2001). Depending on the type of masking (e.g., integration, interruption, or substitution; Green et al., 2011) the prolonged ISI interval has been linked to deficits in the magnocellular (M) and parvocellular (P) pathways (Schechter et al., 2003, 2005; Green et al., 2011).

The relationship between backward masking for location and object identification and gamma-band activity was examined in a series of studies by Green et al. (Green et al., 1999, 2003; Wynn et al., 2005). Systematic variation of inter-stimuli-intervals (ISIs) revealed that the response functions of ScZ-patients were best fitted with a continuous sine while in controls sensitivity to ISIs was consistent with a damped sinusoid (Green et al., 2003). Conversion of the wavelength parameter indicated that 30–35 Hz frequencies reflected best detection performance in controls. For ScZ-patients, the fitted sinusoids yielded a 32 Hz frequency conversion for the backward-masked location identification but a lower 15 Hz frequency for the backward-masked object identification.

To further link dysfunctions between backward masking and gamma-band activity, Green et al. (2003) also assessed EEG signals in response to backward masking of object identification in the 30–35 Hz frequency range. Peak latency in the 30–35 Hz spectral activity differed between groups with ScZ patients manifesting an earlier occipital-parietal peak around 100 ms while in controls gamma-bad activity was delayed (∼200 ms), suggesting intact sensory registration in ScZ-patients. A follow-up study by Wynn et al. (2005) reported, however, reduced 30–40 Hz spectral power in ScZ patients between 50 and 200 ms during backward masking. In addition, while controls expressed stronger spectral activity to incorrect (vs. correct) trials, the opposite was observed for ScZ patients. It is presently unclear whether backward masking deficits involve impaired evoked oscillations or whether later processing stages might be compromised.

Deficits in high-frequency oscillations are also observed in response to basic sensory stimuli. Spencer et al. (2008) examined evoked EEG responses to auditory and visual stimuli in chronic ScZ patients (**Figure 1**). Interestingly, ScZ-patients' spectral amplitude and measure of inter-trial phase-locking to auditory stimuli were comparable to healthy controls. In contrast, the 25–45 Hz visually-evoked gamma oscillatory response was absent in the phase-locking frequency maps of ScZ patients, whose 30–38 Hz phase-locking over occipital regions was significantly reduced.

Similarly, ScZ-patients are characterized by reduced gammaband responses to illusory square stimuli that presumably engage visual binding processes (Spencer et al., 2003, 2004). Spencer et al. (2003) showed that ScZ patients expressed a weaker P1 component which was accompanied by a reduced phase-locking of occipital evoked (24–48 Hz) activity to illusory square stimuli relative to controls. Moreover, ScZ patients' phase-locking over frontal-central EEG sensors were delayed in response to illusory squares, and occurred at lower frequencies. In a followup study, Spencer et al. (2004) examined response-time (RT) locked (20–45 Hz) evoked beta/gamma-band activity during the same paradigm and found reduced phase-locked activity in the 30–45 Hz frequency range in ScZ-patients which was accompanied by a shift to lower (22–26 Hz) activity relative to controls.

#### **INDUCED ACTIVITY**

Given that non-stimulus-locked (induced) oscillations, have been reported during perceptual organization processes during normal brain functioning (Rodriguez et al., 1999; Tallon-Baudry and Bertrand, 1999), it is likely that a focus on evoked activity only partially addresses the contribution of high-frequency activity toward visuo-perceptual dysfunctions in ScZ. To this end, two studies by Uhlhaas and colleagues (Uhlhaas et al., 2006a; Grützner et al., 2013) investigated induced beta/gamma spectral power during the viewing of Mooney faces, which involve the grouping of the fragmentary parts into coherent images based on the Gestalt principle of closure (Mooney and Ferguson, 1951). EEG-response to Mooney faces revealed largely intact gammaband activity in ScZ-patients relative to controls (Uhlhaas et al., 2006a). However, a subsequent study with MEG (Grützner et al., 2013) reported prominent reduction in evoked and induced 60–120 Hz spectral activity in ScZ-patients (effect size: *d* = 1*.*26; **Figure 1**). Differences between the findings from EEG and MEGdata may be due to the fact MEG has improved sensitivity in detecting low-amplitude high-frequency oscillations than EEG (Muthukumaraswamy, 2013).

The findings of impaired induced gamma-band activity during perceptual organization are complemented by data showing reduced high-frequency activity during working memory and executive processes Haenschel et al. (2009) investigated gammaband activity in EEG-data during a visual working memory paradigm demonstrating significant reductions in gamma-band power at higher working memory load conditions in early-onset ScZ-patients. Similarly, Cho et al. (2006) reported a decrease in induced gamma-band power in chronic ScZ-patients during a cognitive control task which involved the inhibition of a prepotent response.

#### **LONG-RANGE SYNCHRONY**

In addition to the reduction in amplitude and consistency of evoked and induced spectral activity in ScZ patients, several studies have also assessed long-range neural synchrony through analyzing phase-synchronization between electrode pairs. This is of particular relevance because substantial evidence suggests that the functional networks underlying perception, attention, and executive processes rely on dynamic coordination through the inter-areal phase locking rhythmic activity (Lachaux et al., 1999; Varela et al., 2001). Spencer et al. (2003) observed a delayed onset of the 37–44 Hz phase synchrony as well as pronounced decreases in inter-hemispheric coherence during illusory-square perception over parietal electrodes in patients with ScZ. Moreover, Uhlhaas et al. (2006a) reported decreased phase-synchrony over fronto-temporal, and parieto-occipital sensors in the 200–300 ms period post stimulus onset, predominantly at beta (20–30 Hz) but also in the gamma-frequency range (31–38 Hz) during the perception of Mooney faces. The significant reductions in phasesynchrony observed in ScZ patients could indicate a global deficit in generating and sustaining synchrony both within local and also between distributed neural networks relevant for sensory processing.

#### **RELATIONSHIPS WITH CLINICAL VARIABLES**

Preliminary evidence suggests that alterations in high-frequency oscillations during visual processing in ScZ-patients may reflect psychopathological variables. Spencer et al. (Spencer et al., 2003, 2004) reported that evoked phase-synchrony during illusory square perception was correlated with conceptual disorganization and visual hallucinations as well as a relationship between the lowered oscillation frequency and the expression of positive symptoms (delusions) and conceptual disorganization. Finally, Uhlhaas et al. (2006a) reported a positive relationship between 40 and 70 Hz phase synchrony and positive symptoms while a reduction of phasesynchronization correlated with elevated negative symptoms. Significant correlations have also been reported with spectral power. Reduced 60–120 Hz spectral power was found to correlate with elevated levels of disorganization by Grützner et al. (2013). However, an important issue is whether these observed alterations in high-frequency activity are independent of medication status. To date, the only published finding by Minzenberg et al. (2010) indicated that gamma-band activity during cognitive control was reduced in medication-naïve FE-ScZ-patients.

#### **PATHOPHYSIOLOGY OF VISUAL PROCESSING DEFICITS AND NEURAL OSCILLATIONS**

Visually elicited high-frequency oscillations might be ideally suited for translations research (Spencer, 2009; Uhlhaas and Singer, 2012). In the following section, we review the potential involvement of changes in excitatory-inhibition balance, anatomical parameters, and genetic factors that could provide plausible explanations for the breakdown of high-frequency neural oscillations and the associated visual dysfunction observed in ScZ.

#### **EXCITATORY-INHIBITION (E/I) BALANCE**

One important parameter for the generation of high-frequency oscillations in visual circuits but also in the cortex in general is the balance between excitation and inhibition (E/I-balance). Convergence of theoretical (Spencer, 2009; Kopell et al., 2010) and empirical studies (Whittington et al., 1995; Wang and Buzsáki, 1996; Traub et al., 2004) indicate that the generation of high-frequency oscillations crucially involve networks of inhibitory interneurons (Whittington et al., 1995; Bartos et al., 2007; Mann and Paulsen, 2007; Buzsáki and Wang, 2012) and glutamatergically mediated excitatory drive (Lukatch et al., 2005; Chamberlain et al., 2012). Specifically, basket cells which express calcium-binding parvalbumim (PV; Cardin et al., 2009; Sohal et al., 2009; Volman et al., 2011) are of particular relevance for the generation of high-frequency oscillations, specifically at gamma-band frequencies, because of their fastspiking properties (e.g., Buzsaki et al., 1983; Kawaguchi and Kubota, 1997).

More recently, optogenetic tools have enabled more precise links between changes in E/I-balance parameters and network oscillations to be established. For example, Sohal et al. (2009) showed that inhibition of PV interneurons led to an immediate suppression of 30–80 Hz oscillations while 10–30 Hz oscillations increased in power. In contrast, increasing PV-interneuron mediated feedback inhibition by boosting principal cell activity enhanced gamma-band power (Cardin et al., 2009).

Evidence suggests that E/I-balance parameters are disturbed in ScZ (Lewis et al., 2005, 2012). Specifically, the mRNA of GAD67 which synthesizes GABA is reduced in several cortical areas, including visual regions, in ScZ-patients (Akbarian et al., 1995; Mirnics et al., 2000; Hashimoto et al., 2003; Lewis et al., 2011, 2012). Moreover, this decrease is accompanied by reduced expression of the GABA membrane transporter 1 (GAT1; Volk et al., 2001; Lewis et al., 2005; Akbarian and Huang, 2006). GAT1 membrane transporters are expressed on chandelier neurons whose axon terminals synapse exclusively with the axonal initial segment of pyramidal neurons and thus uniquely regulate the excitatory pyramidal output (Lewis, 2000). Further evidence for a dysfunction in GABAergic transmission comes from magnetic resonance spectroscopy (1H-MRS) studies which have shown abnormal GABA-levels (Kegeles et al., 2012). Furthermore, MRS-measured reduction in GABA-levels was found to correlate with psychophysical impairment in orientation-specific surround suppression in ScZ patients (Yoon et al., 2010), suggesting a potential role in visual dysfunctions.

Additional parameters crucial for the generation of highfrequency oscillations include the AMPA- and NMDA-receptormediated activation of PV interneuron (Belforte et al., 2010; Carlén et al., 2012; Gonzalez-Burgos and Lewis, 2012). NMDAreceptor dysfunction has been implicated in the pathophysiology of ScZ through evidence from genetics (Carlén et al., 2012; Kirov et al., 2012) as well as from studies which have tested the impact of NMDA-receptor blockade on cortical processes. In healthy controls, Ketamine, an antagonist of the NMDAreceptor, elicits the full range of psychotic symptoms and impairments in cognitive processes, including visual perception (Hong et al., 2010). Furthermore, it has been shown in animal models that the blockade of NMDA-receptors induced aberrant high-frequency oscillations in extended cortical and subcortical networks (e.g., Hunt et al., 2011; Kittelberger et al., 2012; Phillips et al., 2012). For example, Anver et al. (2011) showed that NMDA-antagonists reduced the frequency of gammaband oscillations as well as induced phase coupling of the normally independent generating networks in cortical layers III and V. These findings suggest that E/I-balance is crucial in assuring coordinated occurrence of high-frequency activity during normal brain functioning in networks involved in visual processing. Consequently, abnormalities in these parameters could lead to changes in both amplitude and synchrony of beta/gamma-band oscillations and, in turn, lead to visual deficits.

#### **ANATOMICAL PARAMETERS**

In addition to the crucial contribution of GABAergic and glutamatergic neurotransmission towards high-frequency oscillations, anatomical parameters such as the layout of excitatory long-range connections have been implicated in long-range synchronization and the integrity of visual processing (Engel et al., 1991). Synchronization of oscillatory activity in the beta and gamma frequency range is dependent on cortico-cortical connections that reciprocally link cells situated in the same cortical area, in different areas, or even in different hemispheres (Engel et al., 1991). Interestingly, a recent study demonstrated that callosal connections contribute to the subjective experience of a visual motion stimulus that requires inter-hemispheric integration (Genc et al., 2011). As such, disruptions in the volume and organization of anatomical connectivity could impair longrange synchronization and impact on visual processes that require large-scale integration. However, a recent study that investigated inter-hemispheric transfer times with ERPs did not support this hypothesis in ScZ (Whitford et al., 2011a).

Further research is required to examine, in greater detail, the relationship between anatomical abnormalities and highfrequency oscillations. This is particularly relevant given the evidence from *in vivo* and post-mortem studies in patients with ScZ highlighting that both the volume and organization of white matter is abnormal, including both early and higher visual areas (Akbarian and Huang, 2006; Hashimoto et al., 2008). Additional evidence supporting the abnormal anatomy of visual regions was reported by Selemon et al. (1995) who observed increased neuronal density in area 17 (occipital cortex) as well as in area 9 (frontal cortex) in ScZ-patients. In contrast, Dorph-Petersen et al. (2007) found no difference in neuronal density in area 17 in ScZ and cortical thickness was in the normal range. However, the authors reported significantly reduced number of neurons as well as volumetric decreases in area 17 (Dorph-Petersen et al., 2007).

Abnormalities in gray matter could also potentially impact on the amplitude of neural oscillations as EEG/MEG signals are dependent on the ionic currents flowing in the dendrites of clusters of synchronously activated neurons during synaptic transmission that could be compromised by either reductions in the number of neurons and/or decreases in neuropil. Supporting a potential link between electrophysiological parameters and anatomical abnormalities in ScZ, Onitsuka et al. (2006) demonstrated an association between the degree of gray matter reduction and a decrease in the N170 ERP component.

#### **GENES**

With the genetic heritability of ScZ being estimated to be around 50–80% (e.g., Sullivan et al., 2003; Harrison and Weinberger, 2004), one plausible hypothesis is that ScZ risk genes may influence the strength and frequency of neural oscillations in the disorder. Indeed, recent animal models which have examined the effects of risk genes on changes in high-frequency oscillations support this view. Carlson et al. (2011) investigated gammaband responses during auditory stimulation in mice with reduced dysbindin-1 expression which is a major ScZ risk gene (Straub et al., 2002; Ross et al., 2006; O'Tuathaigh et al., 2007). Deficits in early evoked gamma-band activity were found which were associated with a decrease in PV cell immune-reactivity. Similarly, Fisahn et al. (2009) demonstrated that polymorphisms in genes encoding Neuregulin-1 (NRG-1) and one of its receptors (ErbB4) selectively increase the power of *in vitro* gamma-band oscillations in hippocampal slices. Accordingly, these data suggest the possibility that ScZ-risk genes modify the E/I-balance parameters which in turn dysregulate the occurrence of high-frequency oscillations.

Recent evidence from genetic studies which have investigated the heritability of visually-induced gamma-band oscillations supports the utility of using high-frequency activity as an intermediate phenotype. van Pelt et al. (2012) recorded visually induced MEG-activity in monozygotic and dizygotic twin pairs (**Figure 2**). The peak-frequency of gamma-band oscillations were highly correlated in monozygotic but not in dizygotic twins, highlighting a strong genetic determination of gammaband oscillations. Additionally, reduced auditory evoked gammaband activity has been demonstrated in first-degree relatives of patients with schizophrenia as well as in unaffected, monozygotic twins with a high degree of heritability (Hall et al., 2011; **Figure 2**). Moreover, Hong et al. (2008) showed that abnormalities in theta- and alpha-band oscillations during sensory gating in ScZ-patients, their relatives, and healthy controls were characterized by heritability rates that exceeded those of traditional ERP measures.

# **ISSUES FOR FUTURE RESEARCH**

The current review has shown a close relationship between visual processing and high-frequency oscillations during normal brain functioning as well as a potential link between aberrant beta/gamma-band activity and dysfunctional visual perception in ScZ. Given the known neurobiological parameters involved in the generation of high-frequency oscillations, we suggest that visually elicited high-frequency oscillations may constitute a useful window for gaining further insights into the pathophysiology of ScZ. To this end, we would like to raise several issues that we consider critical for future research.

The overall conclusion that can be drawn from the studies reviewed is that ScZ is associated with reductions in the amplitude, frequency, and/or synchronization of beta/gammaband oscillations during visual processing. Such deficits have been demonstrated during a wide range of task-conditions, such as in basic responses to visual entrainment (e.g., Krishnan et al., 2005), impaired stimulus-locking of oscillatory activity during perceptual binding (Spencer et al., 2003, 2004, 2008) and visual masking (Green et al., 1999, 2003; Wynn et al., 2005) as well as deficits in generating high-frequency oscillations (Grützner et al., 2013) and their large-scale integration during perceptual organization of complex stimuli (Spencer et al., 2003; Uhlhaas et al., 2006a,b). Abnormalities in visually elicited high-frequency oscillations are consistent with reduced beta/gamma-band activity during auditory (Kwon et al., 1999) and somatosensory perception (Arnfred et al., 2011). Together these findings suggest that cortical circuits in ScZ may be characterized by a comprehensive impairment in the mechanisms responsible for the generation and coordination of adequate high-frequency activity that is present in multiple regions and networks.

#### **HIGH-FREQUENCY OSCILLATIONS, THE VISUAL SYSTEM AND ScZ**

Psychophysical evidence has shown that ScZ-patients are characterized by several deficits in visual processing which include a deficit in stimuli involving the magnocellular pathway (Butler et al., 2008; Javitt, 2009), reduced contextual integration (Yoon et al., 2010; Dias et al., 2011; Yang et al., 2013) and dysfunctions in perceptual organization (Uhlhaas and Silverstein, 2005). Given that the amplitude and the frequency of beta/gammaband oscillations are closely related to stimulus properties during normal brain functioning (see **Box 2**), a combination of precise manipulation of stimulus parameters and electrophysiological approaches may yield novel insights into the relationship between visuo-perceptual dysfunctions and high-frequency oscillations in ScZ.

A reported core deficit underlying visual dysfunction in ScZ is the gain control of visual neuronal responses in ScZ (Butler et al., 2008). Gain control refers to the ability of neurons to modulate their response amplitude and constitutes a general feature of cortical computations (Salinas and Thier, 2000). Impairments in gain control in ScZ are supported by reduced contrast sensitivity (Yang et al., 2013), impaired motion perception (Kim et al., 2006;

parieto-occipital MEG sensors. Time 0s denotes stimulus onset. Right: correlation between gamma-peak frequencies in MZ twins [A] and DZ twins [B]. Each data point represents the peak frequency of one twin vs. that of his or her co-twin (random axis assignment). Slope values are estimated by random permutations of *x* and *y* values. The data suggest a heritability of the gamma-band frequency of 91%. Adapted from van Pelt et al. (2012).

Chen, 2011) and contextual effects (Tadin et al., 2006; Yang et al., 2013; see also Butler et al., 2008 for a review). Moreover, neurophysiological studies have provided psychophysical evidence that these stimulus parameters, which are differentially processed in ScZ, modulate high-frequency activity. For example, increasing the contrast of visual stimuli enhances the frequency of the gamma-band rhythm in V1 (Ray and Maunsell, 2010) and V2 (Roberts et al., 2013). Similar findings have been observed for motion whereby static gratings are associated with lower peak frequencies than moving gratings (Gray et al., 1990; Swettenham et al., 2009; Muthukumaraswamy and Singh, 2013; see **Box 2**). Given these robust relationships, one option for future studies is to parametrically manipulate stimulus contrast and velocity and assess changes in high-frequency oscillations which could yield insights into the integrity of visual circuits in ScZ to support the occurrence and amplitude tuning of beta/gamma-band oscillations.

While such experiments are potentially important for probing dysfunctions in early visual regions, oscillatory dynamics are also crucially involved in mediating the influences of neuronal activity generated in anterior brain regions over the early well-compared with controls, highlighting the genetic contribution toward impairments in high-frequency oscillations in the disorder. Adapted from Hall et al. (2011) by permission of Oxford University Press. stages of visual processing, such as during attention (Womelsdorf

evoked gamma-band power was significantly associated with schizophrenia and unaffected co-twins exhibited significantly reduced 30–60 Hz power as

and Fries, 2007). Of relevance, evidence supporting the facilitatory effects of attention processes, particularly spatial attention, on visually induced high-frequency oscillations constitutes an additional example of gain control whereby neuronal responses to stimuli at attended locations are increased relatively to nonattended locations (Hillyard et al., 1998). This gain in neural response could be mediated through changes in the synchrony of inhibitory networks (Tiesinga et al., 2004a,b).

Previous psychophysical research in ScZ has implicated deficits in the utilization of top-down mediated cues (Silverstein et al., 1996) as well as dysfunctions in bottom-up driven processing in early visual pathways (Uhlhaas and Silverstein, 2005; Butler et al., 2007), suggesting that effects of attention on neural oscillations could be relevant in order to disentangle the contribution of feed-forward mediated vs. top-down processes toward visual deficits in ScZ. Recent studies from invasive electrophysiology (Fries et al., 2001; Ray et al., 2013) as well as from EEG/MEG (Wyart and Tallon-Baudry, 2008) have demonstrated that attention can lead to an increase in both the amplitude as well as the frequency shift of gamma-band activity (Kahlbrock et al., 2012; Koelewijn et al., 2013). These effects occur in early visual areas (Koelewijn et al., 2013) as well as at higher brain areas (Tallon-Baudry et al., 2005) and are accompanied by changes in the coherence of oscillations between early and higher visual regions (Siegel et al., 2008; Bosman et al., 2012). Therefore, detailed probing of attention effects on high-frequency oscillations could potentially offer insights into the differential (e.g., anatomical and frequency-specific) contribution of both bottom-up and top-down processes toward visual processing abnormalities in ScZ.

#### **VISUAL PERCEPTION, HIGH-FREQUENCY OSCILLATIONS IN ScZ AND TRANSLATIONAL RESEARCH**

One important issue concerns the possibility of distinct roles of beta and gamma-band oscillations during visual processing. Recent research have highlighted that beta-band oscillations mediate mainly top down activity and hence are critically involved in the prediction of upcoming sensory events while gammaband oscillations, at least in sensory cortices, are involved in feed-forward signalling (Buschman and Miller, 2007; Arnal and Giraud, 2012). This distinction is supported by the differential laminar expression of beta and gamma-band oscillations. *In vitro* and *in vivo* recordings show that gamma-band activity is prominently generated in superficial layers 2/3 of the cortex (Buffalo et al., 2011), the main origin of feed- forward connections, and dependent upon fast, transient excitation of fast-spiking interneurons via metabotropic glutamate receptors (Whittington et al., 1995). In contrast, beta oscillations are mainly found in infragranular layers, from which feed-back projections originate preferentially. Interestingly, the generation of beta-band oscillations can be independent from excitatory or inhibitory synaptic transmission (Roopun et al., 2008). These observations provide potential hypotheses for future studies to investigate the differential contribution of beta/gamma-band oscillations during visual processing in ScZ. In particular, these investigations could be combined with the investigation of attention effects to address the potentially distinct roles of feed-forward vs. top-down mediated neuronal activity in perceptual dysfunctions in the disorder.

In addition to the modulation of beta/gamma-band power and synchrony, changes in the oscillatory peak-frequency may also be useful in establishing links between non-invasive EEG/MEGmeasures and E/I-balance parameters (see Spencer et al., 2004; Ferrarelli et al., 2012). It is conceivable that the frequency at which a network oscillates may more closely mirror biophysical parameters of the underlying network. For example, the deactivation kinetics of different GABAergic receptors strongly impact on the generation of fast vs. slow GABAergic currents which in turn are an important parameters for the frequency of oscillations (Wang and Buzsáki, 1996). Additionally, the peak-frequency of visually induced gamma-band activity in MEG-data has been shown to be under close genetic control (van Pelt et al., 2012), indicating that the frequency of gamma-band oscillations could be linked to genetically determined differences in channel-subunits.

Furthermore, mechanistic links between disturbed oscillations and visual perception in ScZ may also be established in combination with rhythmic stimulation through transcranial magnetic stimulation (TMS) and transcranial alternating current stimulation (tACS). Available evidence suggests that oscillatory brain processes can be entrained, enhanced or perturbed by means of external stimulation (Romei et al., 2011; Thut et al., 2011a,b; Antal and Paulus, 2013), which raises the possibility of targeting specific oscillations frequencies in conjunction with visuoperceptual processes in patients ScZ. The feasibility of using TMS, for example, to probe neural circuits in ScZ has been demonstrated in several recent studies (Ferrarelli et al., 2012; Frantseva et al., 2012).

Finally, future research should also consider the overlap in visually elicited high-frequency dysfunctions with related disorders, such as bipolar (BP) and autism spectrum disorders (ASDs). There is substantial evidence that ASDs are characterized by impairments in visual processing as well as deficits in high-frequency oscillations (Dakin and Frith, 2005; Sun et al., 2012). Similarly, there is evidence for impairments in bipolar disorder because auditory-steady state responses (O'Donnell et al., 2004) as well as long-range coherence (Özerdem et al., 2010) at gamma-band frequencies are significantly impaired.

Given the substantial overlap in genes, cognitive deficits and clinical symptoms between different diagnostic categories, it also conceivable that neural oscillations can be used to assign patients into novel categories based on neural oscillations. Fingerprints of neuronal dynamics, such as alterations in the frequency, temporal precision, phase locking, and topology of neuronal oscillations, during visual processing provide a rich coding-space for the definition of discrete entities or taxon (Meehl, 1992) within and also between diagnostic categories. As such, the close links between genes, neurobiology, and parameters (**Figure 2**) are perhaps wellsuited to identify pathways mediated by risk genes.

#### **BETA/GAMMA-BAND OSCILLATIONS AND LOW-FREQUENCY ACTIVITY**

While the current review focused on activity at beta/gamma-band frequencies, activity in lower frequencies ranges (e.g., delta, theta, alpha bands) may also be potential targets for understanding visual dysfunctions in ScZ. Existing evidence from EEG-studies suggests impaired amplitude and phase-locking during visual stimulation is not confined to beta/gamma-band frequencies (Haenschel et al., 2010; Hamm et al., 2012).

The alpha-band rhythm (8–12 Hz) is particularly relevant for the understanding of visual perception as the alpha cycle modulates perceptual detection rates (Valera et al., 1981; Dugué et al., 2011). Moreover, there is consistent evidence that oscillations in the alpha-band interact with the amplitude of gamma activity through cross-frequency coupling (Osipova et al., 2008), raising the possibility that impairments in high-frequency activity could also result from an impaired hierarchical organization of oscillations.

In addition to cross-frequency interactions, there is growing consensus that lower-frequency rhythms also play an important role in coordinating sensory predictions within and between modalities (see Schroeder and Lakatos, 2009 for a review). Recent work by Lakatos et al. (2013) demonstrated that impaired sensory discrimination of auditory stimuli in ScZ-patients was correlated with a deficit in effectively entraining inter-trial delta phaselocking to anticipate relevant sensory processing, and a failure to suppress task-irrelevant activity. These findings highlight the potential relevance of sensory predictions for auditory processing impairments in ScZ. It remains to be investigated whether predictive mechanisms in the visual domain are similarly affected in ScZ.

#### **METHODOLOGICAL IMPLICATIONS**

While it is possible that alterations in high-frequency oscillations during visual processing may reflect dysfunctions in specific variables involved in the generation of high-frequency activity, we cannot exclude the possibility that several non-specific factors, such as the impact of antipsychotic medication, chronic stress and the non-neuronal origin of certain EEG/MEG-signal components, contribute toward findings of impaired beta/gamma-band oscillations in ScZ patients. Accordingly, advances in analytic techniques and experimental designs are essential in order to allow clearer links between changes in high-frequencies oscillations and visuo-perceptual deficits in ScZ.

An approach to further identify such relationships is to employ single-trial analysis of EEG/MEG-data in combination with variation of stimulus parameters. At present, EEG/MEG-studies investigating high-frequency oscillations in ScZ have predominantly concentrated on differences in amplitude and peak-frequency values calculated across conditions or groups of participants. Given the substantial variability in behavioral and electrophysiological parameters both within and between groups, analysis of single-trial EEG/MEG-data analyses could potentially yield additional information as it allows a systematic mapping between brain activity and stimulus information as well as with indexes of behavioral variability (Pernet et al., 2011).

Furthermore, high-frequency oscillations during visual stimulation are accompanied by several important sources of artifacts which can resemble neuronally generated gamma band oscillations, and thus make the interpretation of EEG/MEG-signals difficult. Specifically, induced gamma-band activity coincides with the maximal frequency of micro-saccades which elicit a saccadic spike potential (SSP). Seminal work by (Yuval-Greenberg et al., 2008) highlighted that the SSP can mimic gamma oscillations in bandpass-filtered EEG signals if artifact-correction

#### **REFERENCES**


*Clin. Neurophysiol.* 2, 377–388. doi: 10.1016/0013-4694(50) 90075-7


procedures are not adequately employed (Melloni et al., 2009). The presence of SSP-related gamma-band activity has also been recently demonstrated in MEG-recordings (Carl et al., 2012). Additionally, muscle artifacts can constitute another nonneuronal source of high-frequency activity that, if not carefully removed, can simulate power modulations in the gamma band range over visual regions (Whitham et al., 2007; see also Hipp and Siegel, 2013; Muthukumaraswamy, 2013 for recent reviews of these issues).

In addition to the contribution of eye-movement related artifacts toward high-frequency signals, a potentially important issue is also the relationship between neural oscillations, eyemovements, and visual dysfunctions. Abnormalities in the several eye-movement parameters such as smooth pursuit, fixation stability, scan-path, and fixation dispersal during free viewing are one of the more robust domains of impairments in ScZ-patients (e.g., Benson et al., 2012). Moreover, recent findings suggest that patterns of saccades strongly modulate the occurrence of high-frequency oscillations in V1 (Bosman et al., 2009) and that different oscillatory frequencies are involved in the organization of eye-movement patterns (Ito et al., 2011). Indeed, disentangling the relationship between high-frequency oscillations, eye-movements, and visual processing dysfunctions will be an important albeit challenging area of research.

#### **SUMMARY**

The findings reviewed suggest a potential link between the occurrence of beta/gamma oscillations and the pronounced deficits in visual perception in ScZ. Evidence supporting such a relationship comes from EEG/MEG studies indicating reductions in synchrony and amplitude of beta/gamma-band oscillations during basic and complex visual stimuli as well through anatomical findings that highlight impaired structure and composition of visual circuits in the disorder. Importantly, given the known mechanisms involved in the genesis of high-frequency oscillations, the evidence and clinical importance of visual dysfunctions in ScZ, as well as the opportunity to measure high-frequency oscillations non-invasively, visually elicited high-frequency oscillations in ScZ are potentially suited for translational research.


Parnas, J. (2011). Attenuation of beta and gamma oscillations in schizophrenia spectrum patients following hand posture perturbation. *Psychiatry Res.* 185, 215–224. doi: 10.1016/j.psychres.2009.10.005


cortices. *Science* 315, 1860–1862. doi: 10.1126/science.1138071


relation between brain excitation and visual perception. *J. Neurosci.* 31, 11889–11893. doi: 10.1523/JNEUROSCI.1161-11.2011


mapping of the 'C1' component. *Brain Topogr.* 21, 11–21. doi: 10.1007/s10548-008-0063-4


doi: 10.1111/j.1460-9568.2008. 06358.x


studying neural interactions in the human brain. *Proc. Natl. Acad. Sci. U.S.A.* 98, 694–699. doi: 10.1073/pnas.98.2.694


and Clementz, B. A. (2011). Abnormalities of neuronal oscillations and temporal integration to low- and high-frequency auditory stimulation in schizophrenia. *Biol. Psychiatry* 69, 989–996. doi: 10.1016/j.biopsych.2010.11.021


353, 1257–1270. doi: 10.1098/rstb. 1998.0281


(2012). Comparison of the effects of acute and chronic administration of ketamine on hippocampal oscillations: relevance for the NMDA receptor hypofunction model of schizophrenia. *Brain Struct. Funct.* 217, 395–409. doi: 10.1007/s00429-011-0351-8


Tan et al. Oscillations and vision in ScZ

in humans: an EEG-study. *Neurosci. Lett.* 183, 39–42. doi: 10.1016/0304-3940(94)11109-V


muscle artifacts in MEG/EEG: a review and recommendations. *Front. Hum. Neurosci.* 7:138. doi: 10.3389/fnhum.2013.00138


M. D., Lucia, L. C. B. S., et al. (2006). Functional and structural deficits in brain regions subserving face perception in schizophrenia. *Am. J. Psychiatry* 163, 455–462. doi: 10.1176/appi.ajp.163.3.455


their possible use in computation. *Neuron* 67, 885–896. doi: 10.1016/j.neuron.2010.08.004


reflects the focus of spatial attention. *Neuron* 60, 709–719. doi: 10.1016/j.neuron.2008.09.010


schizophrenia. *Am. J. Hum. Genet.* 71, 337–348. doi: 10.1086/341750


Tallon-Baudry, C., Bertrand, O., Delpuech, C., and Pernier, J. (1996). Stimulus specificity of phase-locked and non-phase-locked 40 Hz visual responses in human. *J. Neurosci.* 16, 4240–4249.


and why does it go awry? *Schizophr. Res.* 94, 253–263. doi: 10.1016/j.schres.2007.05.001


and schizophrenia patients: a combined ERP and DTI study. *Neuroimage* 54, 2318–2329. doi: 10.1016/j.neuroimage.2010.10.048


*Clin. Psychol. Sci.* 1, 5–15. doi: 10.1177/2167702612464618


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 May 2013; accepted: 23 August 2013; published online: 09 October 2013. Citation: Tan H-RM, Lana L and*

*Uhlhaas PJ (2013) High-frequency neural oscillations and visual processing deficits in schizophrenia. Front. Psychol. 4:621. doi: 10.3389/fpsyg.2013.00621*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Tan, Lana and Uhlhaas. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Are patients with schizophrenia impaired in processing non-emotional features of human faces?

*Hayley Darke1, Joel S. Peterman2, Sohee Park2, Suresh Sundram3,4 and Olivia Carter <sup>1</sup> \**

*<sup>1</sup> School of Psychological Sciences, University of Melbourne, Parkville, VIC, Australia*

*<sup>2</sup> Department of Psychology, Vanderbilt University, Nashville TN, USA*

*<sup>3</sup> Department of Molecular Psychopharmacology, Florey Institute of Neuroscience and Mental Health, Parkville, VIC, Australia*

*<sup>4</sup> Northern Psychiatry Research Centre, Epping, VIC, Australia*

#### *Edited by:*

*Randolph Blake, Vanderbilt University, USA*

#### *Reviewed by:*

*Mandy Rossignol, University of Louvain, Belgium Laura Germine, Massachusetts General Hospital/Harvard Medical School, USA*

#### *\*Correspondence:*

*Olivia Carter, School of Psychological Sciences, Faculty of Medicine, Dentistry and Health Sciences, University of Melbourne, Redmond Barry Building, VIC 3010, Australia e-mail: ocarter@unimelb.edu.au*

It is known that individuals with schizophrenia exhibit signs of impaired face processing, however, the exact perceptual and cognitive mechanisms underlying these deficits are yet to be elucidated. One possible source of confusion in the current literature is the methodological and conceptual inconsistencies that can arise from the varied treatment of different aspects of face processing relating to emotional and non-emotional aspects of face perception. This review aims to disentangle the literature by focusing on the performance of patients with schizophrenia in a range of tasks that required processing of non-emotional features of face stimuli (e.g., identity or gender). We also consider the performance of patients on non-face stimuli that share common elements such as familiarity (e.g., cars) and social relevance (e.g., gait). We conclude by exploring whether observed deficits are best considered as "face-specific" and note that further investigation is required to properly assess the potential contribution of more generalized attentional or perceptual impairments.

**Keywords: schizophrenia, vision, face, identity, detection, recognition, gait, perception**

# **INTRODUCTION**

Face processing deficits have been repeatedly demonstrated in patients with schizophrenia, however, debate continues regarding the precise nature of these impairments and the mechanisms that underlie them. Some authors posit that face processing deficits are specific to emotion-related content (Edwards et al., 2002; Schneider et al., 2006); others argue that they represent an impairment in processing biological and socially relevant stimuli (Kim et al., 2010); a more general impairment in visual attention (Caharel et al., 2007); or a generalized difficulty in processing complex visual stimuli—of which faces are just one example (Doop and Park, 2009). In contrast to the vast literature using tasks involving identification or recognition of facial expression and the relationship to impaired emotion processing (see Kohler et al., 2010), relatively little has been done to systematically assess face processing outside an emotional context in individuals with schizophrenia (c.f. a review in the current issue comparing face recognition deficits in schizophrenia and autism Watson, 2013). Before discussing the existing patient literature we will briefly consider the current theories of face processing more broadly.

# **THE SEPARATION OF IDENTITY AND EXPRESSION PROCESSING**

The dominant theory of face perception as first proposed by Bruce and Young (1986) makes a division between identity and emotion recognition representing two largely independent processes. A neuroanatomical framework for this dual route model has since been provided by Haxby et al. (2000) and distinguishes between two types of information: invariant and changeable. Invariant information refers to properties that are consistent across different views and facial expressions, and is necessary for recognizing the identity of a face. Changeable information includes eye gaze, expression, and movements of the eyes and mouth, and is necessary for the recognition of facial affect. The initial processing of facial features is proposed to be mediated by neurons in the inferior occipital gyri (Haxby and Gobbini, 2011). From here, the analysis of "variant" or changeable visual features is processed largely via a route involving the posterior superior temporal sulcus (pSTS). In contrast, invariant (unchangeable) information is proposed to be processed via a ventral temporal route including the inferior occipital and fusiform gyri. These two routes then have differing degrees of connectivity with either limbic or cortical regions outside these face selective areas of visual extrastriate cortex.

Support for this separation of identity and expression recognition comes from a broad range of sources (for review see Calder and Young, 2005) including behavioral research (Bruce, 1986; Campbell, 1996; Calder et al., 2000), functional imaging studies (George et al., 1993; Sergent et al., 1994; Winston et al., 2004), and in dissociations exhibited by individuals with brain injury (Tranel et al., 1988; Young et al., 1993; Hornak et al., 1996) and prosopagnosia (Baudouin and Humphreys, 2006; Riddoch et al., 2008). Compelling neurophysiological evidence for this dissociation has also come from studies of non-human primates (Tsao et al., 2003; Pinsk et al., 2005; Rajimehr et al., 2009) where multiple "patches" of face-selective cortex have been identified that show selectivity to identity or expression processing, respectively (Tsao et al., 2008a,b).

While this evidence demonstrates that identity and expression recognition involve separate processes, the level at which this bifurcation occurs, and the degree to which these parallel processes interact, are yet to be resolved. Importantly, it is also unclear the extent to which identity recognition and emotional processing are inextricably linked (Fitousi and Wenger, 2013; and for review see Calder, 2011). To properly understand the nature of face processing deficits in schizophrenia it is important to consider deficits in emotionally neutral judgments based on either changeable or invariant information.

# **IDENTITY PROCESSING DEFICITS**

Deficits in non-emotional face processing have been demonstrated using a variety of behavioral tasks. However, heterogeneity across tasks types and participants has produced mixed results (see **Table 1**). Tasks that assess true identity processing are primarily matching tasks, where the participant views static photographs of faces (with non-face identifying features removed, such as hair and spectacles), either serially or concurrently. The participant must match the identity of the first face to one of several options (Addington and Addington, 1998; Penn et al., 2000; Kucharska-Pietura et al., 2005; Chen et al., 2012). The most commonly used face-matching task is the Benton Test of Facial Recognition (Benton, 1983; see **Figure 1A**). Individuals with schizophrenia have shown impaired performance on the Benton test in many (e.g., Addington and Addington, 1998; Evangeli and Broks, 2000; Whittaker et al., 2001; Hooker and Park, 2002; Kucharska-Pietura et al., 2005; Soria Bauser et al., 2012), but not all studies (Hall

**Table 1 | Identity recognition tasks used in schizophrenia research from the last 20 years—All results relate to performance of Schizophrenia patients relative to controls.**


*1Short form, 2Long form, 3Not specified.*

*\*Did not include a facial affect comparison task.*

**FIGURE 1 | (A)** Plate from the Benton Facial Recognition Test (Benton, 1983). Participants indicate which of the six images match the target. (Published in Busigny and Rossion, 2010, p. 969). **(B)** Identity matching task used by Norton et al. (2009). **(C)** Example of morphed images ranging from "no sex" (50% male, 50% female) to 100% male face (from Bediou et al., 2005, , p. 528). **(D)** Examples of upright and inverted stimuli used in Soria Bauser et al. (2012). **(E)** Example of semi-successive frames of the point-light displays (walking) used in Kim et al. (2005).

et al., 2004; Scholten et al., 2005; Van 't Wout et al., 2007; Pomarol-Clotet et al., 2010). One study using a booklet-based face-matching task with similar properties to the Benton also found no impairment (Hooker and Park, 2002). Three studies using a morphed identity-matching task—in which participants choose which of a pair of faces of varying similarity matches a briefly presented target—have also produced inconsistent results, with two studies reporting no impairment (Norton et al., 2009; Chen et al., 2012) and only one study approaching significance (Chen et al., 2009) (see **Figure 1B**). The reason for these inconsistent findings is unclear. Given the relatively heterogeneous patient samples, however, the influence of factors such as age, illness duration, and gender may be worth exploring through a formal meta-analysis in the future.

An alternative measure of identity recognition is the twoalternative forced-choice identity discrimination paradigm, in which the participant judges whether two serially presented faces are the same or different. Again, some studies reported significant impairment in schizophrenia (Martin et al., 2005; Butler et al., 2008; Shin et al., 2008; Soria Bauser et al., 2012) while others found no impairment relative to controls (Edwards et al., 2001; Johnston et al., 2010; Soria Bauser et al., 2012). Another study requiring participants to distinguish between photographs of two learned identities, person A and person B, also reported significant impairment (Baudouin et al., 2002). Again, the reason for these inconsistent findings is unclear, but given that mean age and illness duration of the patients (Kucharska-Pietura et al, 2005), and the stimuli used in these tasks (e.g., Soria Bauser et al., 2012) are known to influence performance, it is possible these factors contributed here. While it is outside the scope of the current review it would be interesting to consider the relative severity of face perception deficits a non-clinical population of individuals high in psychosis-proneness and whether these deficits predict transition to psychosis. While impairments have been seen in some non-emotional aspects of face processing in these populations before (Poreh et al., 1994), they are not reported as frequently as deficits in face emotion perception (Waldeck and Miller, 2000; Williams et al., 2007; Germine and Hooker, 2011).

#### **SINGLE FEATURE IDENTIFICATION**

Tasks that do not assess true identity discrimination, but just one aspect of facial identity have produced more consistent results. For instance, (Bediou et al., 2005, 2007, 2012) found that patients with schizophrenia have an intact ability to discriminate the sex of faces that have been digitally morphed to increase ambiguity (see **Figure 1C**). Two studies (Schneider et al., 1995; Kohler et al., 2000) found that patients are significantly impaired in judging the age of a face in decades (i.e., teens, twenties, thirties, etc.), while one reported poorer accuracy, but faster reaction times than controls (Schneider et al., 1998). In contrast, one study found that patients were impaired in judging if a face is older or younger than 30 (Schneider et al., 2006), while two studies found no difference (Gur et al., 2002a,b). Taken together, these findings suggest that patients with schizophrenia are not impaired in their ability to make broad judgments about the sex or age of a face, but have greater difficulty on tasks that require more fine grained judgments. This result highlights the importance of selecting the right measure when assessing face recognition in schizophrenia. True identity perception likely reflects a judgment based on complex interactions between multiple facial features, so a task assessing just one aspect of face perception (such as sex) may not be a valid indicator of a true deficit in identity discrimination.

#### **IDENTITY RECOLLECTION AND FAMILIARITY**

Finally, tasks that assess memory of—as opposed to discrimination between—faces have largely revealed significant impairment in schizophrenia (Whittaker et al., 2001; Sachs et al., 2004; Calkins et al., 2005; Silver et al., 2009; Soria Bauser et al., 2012). In contrast, patients have been shown to have an intact ability to recognize famous faces (Evangeli and Broks, 2000; Whittaker et al., 2001; Joshua and Rossell, 2009; although one study reported significant impairment: Pomarol-Clotet et al., 2010), but are less accurate in familiarity judgment of photographs of strangers and known people (i.e., their doctor's face) (Caharel et al., 2007). However, these tasks are not necessarily an indicator of pure face processing in schizophrenia because this disorder is associated with general impairments in memory and new learning (Boyer et al., 2007). Again, care should be taken when employing memory-based face recognition tasks to ensure that general impairments in memory are taken into account.

# **IMPAIRED FACE PROCESSING**

Face information is extracted from the environment using both face-specific and more general perceptual processes (McKone and Robbins, 2011). Some argue that face processing deficits in schizophrenia indeed represent dysfunction in face-specific perceptual processes—generally referred to as "holistic face processing." This represents a rapid, involuntary face-specific perceptual process that integrates information across the face as a whole. It includes such information as the shapes of individual features, the relative distances between them, and the contour of the cheeks and jaw (Maurer et al., 2002; McKone and Yovel, 2009). This process is specific to invariant face information, and is therefore critical for perceiving identity (McKone and Robbins, 2011). For instance, it has been demonstrated in healthy controls that holistic processing predicts an individual's ability to remember and distinguish between faces (Richler et al., 2011). Similarly, it has been shown that individuals with congenital prosopagnosia (inability to recognize faces) perform poorly on holistic processing tasks (Palermo et al., 2011).

One common means of evaluating holistic processing is with the Face Inversion Effect—observed as a reduction in face discrimination performance for inverted faces compared to upright faces (see **Figure 1D**). The magnitude of this effect is thought to represent a loss of holistic information crucial to face discrimination, and is disproportionately larger for faces compared to other non-face stimuli (Yin, 1969). Studies of holistic processing in schizophrenia produced varied results, with some studies reporting normal inversion effects for faces (Schwartz et al., 2002; Chambon et al., 2006; Butler et al., 2008), while others report reduced inversion effects compared to controls (Shin et al., 2008; Kim et al., 2010; Soria Bauser et al., 2012). In particular, Shin et al. (2008) reported that patients with schizophrenia were more impaired when discriminating faces that differed in configural information, rather than featural information. An electrophysiological indicator of the face inversion effect is the N170 (Eimer, 2000), a negative potential seen using electroencephalography (EEG). The N170 is reduced in patients with schizophrenia while viewing inverted faces (Onitsuka et al., 2006; Ibáñez et al., 2012), and is associated with lower scores on measures of social functioning (Obayashi et al., 2009; Tsunoda et al., 2012). These findings suggest there may be an underlying face processing abnormality that may go undetected by commonly used behavioral measures.

In a related behavioral study, Schwartz et al. (2002) employed the composite face task, which is considered to provide a more rigorous measure of holistic processing than other inversion tasks (McKone, 2008). In this task, participants are required to make decisions about the upper halves of faces while ignoring the lower halves. These face halves are either aligned to form a complete face (producing an interference effect) or misaligned (removing the interference). When the stimuli are inverted, however, the aligned faces no longer produce strong interference effects. It was found that patients with schizophrenia showed typical patterns of interference for upright faces and not inverted faces. While this study has not been repeated, it provides support for the argument that holistic processing is largely preserved in schizophrenia and appears to contradict some of the results using the face inversion effect.

## **EVIDENCE FOR IDENTITY PROCESSING DEFICITS USING NON-FACE STIMULI**

As outlined above, a number of studies have shown impaired performance on tasks aimed to assess face-specific processing. However, a number of similar deficits seem to be apparent on tasks using non-face stimuli. For example, Soria Bauser et al. (2012) reported reduced inversion effects for cars and bodies (see **Figure 1D**) that mirrored their findings using face stimuli, suggesting an impairment that encompasses more than just facespecific holistic processing. An interesting comparison is also provided by research looking at gait perception. Previous research has indicated that the identity of an individual can also be extracted from an individual's gait pattern (Cutting and Kozlowski, 1977). Through the use of point-light displays (PLD; Johansson, 1973), visually impoverished stimuli provide body form and structure solely through motion cues of coordinated dots (see **Figure 1E**). Similar to the ERP findings regarding face processing, the N170 component has also been found in healthy individuals during visual processing of inverted PLD and static images of bodies (Stekelenburg and de Gelder, 2004; Jokisch et al., 2005). Loula et al. (2005) demonstrated that healthy subjects exhibited superior performance in identifying self and friend's movement when compared to a stranger's movement. Furthermore, inverting the PLD resulted in chance performance across all three conditions. Unfortunately the use of dynamic gait stimuli in the investigation of true identity recognition deficits in schizophrenia has yet to be conducted. Individuals with schizophrenia are impaired in discriminating PLD-presented body movements (biological motion) from scrambled PLD body movements (Kim et al., 2005). It is therefore, conceivable that the ability to use the information provided in the point-light displays to extract identify information would also be impaired.

# **DOES IMPAIRED IDENTITY PROCESSING REFLECT A GENERALIZED ATTENTIONAL DEFICIT?**

One possible account for face processing deficits in schizophrenia is that they are the result of a more general impairment in allocating visuospatial attention (Baudouin et al., 2002). One suggestion is an impairment in global vs. local visual processing. "Global processing" refers to the ability to attend to *any* visual stimulus as a "whole," as opposed to its component features (Tan et al., 2009). Studies of schizophrenia have revealed impairments in global processing, but largely preserved local processing both for static (Goodarzi et al., 2000; Silverstein et al., 2000; Johnson et al., 2005; Poirel et al., 2010) and dynamic stimuli (Chen et al., 2003). In addition, patients with schizophrenia demonstrate a bias toward attending to the local level of a stimulus, even when task demands favor a global strategy (Landgraf et al., 2011).

It is possible that a global processing deficit could contribute to impairments in identity recognition because the important global-level information is not being processed efficiently. For instance, it has been shown that identity recognition performance is improved when healthy participants are primed to adopt a global processing strategy, and impaired when primed with a local processing strategy (Macrae and Lewis, 2002; Perfect, 2003). Patients with schizophrenia similarly showed less of a reduction in identity recognition performance compared to controls when configural cues were removed from a face (Joshua and Rossell, 2009), indicating that these individuals relied more strongly on local features when identifying famous faces. Global processing deficits could also explain the expected deficits in identity recognition from gait in individuals with schizophrenia. Kim et al. (2005) argued that deficits in biological motion perception in individuals with schizophrenia may arise due to their well-documented difficulties in global motion perception (for review see Chen, 2011).

# **DOES IMPAIRED FACE IDENTITY PROCESSING REFLECT A GENERAL VISUAL PERCEPTUAL DIFFICULTY?**

Individuals with schizophrenia show a gamut of visual perceptual impairments (see Butler et al., 2008 and the editorial of this research topic!). These difficulties include form processing such as object recognition, grouping, perceptual closure, and visual context (Place and Gilmore, 1980; Saccuzzo and Braff, 1986; Rief, 1991; Kerr and Neale, 1993; Rabinowicz et al., 1996; Kohler et al., 2000; Silverstein et al., 2000; Doniger et al., 2002; Brenner et al., 2003; Uhlhaas et al., 2006; Kurylo et al., 2007; Yang et al., 2013). Moreover, neuroanatomical data indicate that the visual cortex in schizophrenia is abnormal with respect to the density of neurons (Selemon et al., 1995), total number of neurons (Dorph-Petersen et al., 2007) and GABA concentration in the visual cortex that is associated with orientation-specific center-surround suppression (Yoon et al., 2010). Interestingly, the face fusiform area (FFA) seems relatively intact, at least functionally (Yoon et al., 2006). Given the exhaustive list of basic visual perceptual deficits in schizophrenia, it seems likely that processing of complex visual stimuli such as faces would also be compromised. Thus, it is likely

# **REFERENCES**


that at least some aspects of face processing deficits observed in schizophrenia arise from visual cortical abnormalities.

# **CONCLUSIONS**

Deficits in face processing have frequently been observed in patients with schizophrenia. In order to fully understand the mechanisms underlying these impairments it is important to consider the relative contribution of the multiple factors that may be involved. The fact that deficits have been seen in face identity tasks without an emotional/expression recognition component suggests that these deficits are unlikely to be limited to emotion processing. Moreover, the observation of more generalized impairments in visual and attentional function in these patients also raises questions about whether there is indeed anything special about faces at all. Lastly, the potential role of medication in these impairments has yet to be clearly determined. It is only through future controlled studies that balance difficulty across memory, attentional and perceptual demands—or directly assess the capacities—that we will begin to understand how face processing deficits emerge in these patients.

#### **ACKNOWLEDGMENTS**

This work was supported by an Australian National Health and Medical Research council fellowship # 628590 to Olivia Carter, National Institute on Mental Health T32 MH018921-21A1 to Joel S. Peterman, Vanderbilt International Office Grant to Sohee Park.

matched visual-processing tasks in schizophrenia spectrum disorders. *J. Abnorm. Psychol.* 112, 28–37. doi: 10.1037/0021-843X.112.1.28


recognition involve separate visual routes?" in *The Oxford Handbook of Face Perception,* eds A. J. Calder, G. Rhodes, M. H. Johnson, and J. V. Haxby (Oxford: Oxford University Press), 427–448. doi: 10.1093/ oxfordhb/9780199559053.013.0022


schizophrenia. *Neuropsychologia* 44, 2437–2444. doi: 10.1016/j. neuropsychologia.2006.04.008


in the structural encoding of faces. *Cogn. Neurosci.* 11, 2319–2324.


schizophrenia patients. *Psychiatry Res.* 112, 41–50. doi: 10.1016/S0165- 178100177-4


visual recognition of biological motion in schizophrenia. *Schizophr. Res.* 77, 299–307. doi: 10.1016/j.schres.2005.04.006


*Handbook of Face Perception,* eds A. J., Calder, G., Rhodes, M. H. Johnson, and J. V. Haxby (Oxford: Oxford University Press), 149–176. doi: 10.1093/oxfordhb/ 9780199559053.013.0009


*Acad. Sci. U.S.A.* 102, 6996–7001. doi: 10.1073/pnas.0502605102


cognition in schizophrenia. *Schizophr. Res.* 68, 27–35. doi: 10.1016/S0920-996400131-2


valence, specific and context dependent. *Psychiatry Res.* 169, 101–106. doi: 10.1016/j.psychres.2008.11.017


*Interv. Psychiatry* 1, 177–182. doi: 10.1111/j.1751-7893.2007.00023.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 12 June 2013; accepted: 26 July 2013; published online: 20 August 2013. Citation: Darke H, Peterman JS, Park S, Sundram S and Carter O (2013) Are patients with schizophrenia impaired in processing non-emotional features of human faces? Front. Psychol. 4:529. doi: 10.3389/fpsyg.2013.00529*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Darke, Peterman, Park, Sundram and Carter. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Noise as a mechanism of anomalous face processing among persons with Schizophrenia

#### *Bruce K. Christensen1 \*, Justine M. Y. Spencer 2, Jelena P. King1, Allison B. Sekuler <sup>2</sup> and Patrick J. Bennett <sup>2</sup>*

*<sup>1</sup> Schizophrenia Research Unit, Department of Psychiatry and Behavioural Neuroscience, McMaster University, Hamilton, ON, Canada*

*<sup>2</sup> Department of Psychology, Neuroscience and Behaviour, McMaster University, Hamilton, ON, Canada*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Giancarlo Dimaggio, Centro di Terapia Metacognitiva Interpersonale, Italy Bruce Turetsky, University of Pennsylvania, USA*

#### *\*Correspondence:*

*Bruce K. Christensen, Schizophrenia Research Unit, Department of Psychiatry and Behavioural Neuroscience, McMaster University, 100 West 5th Street, Hamilton, ON L8N 3K7, Canada e-mail: bruce.christensen@ mcmaster.ca*

There is substantial evidence that people with Schizophrenia (SCZ) have altered visual perception and cognition, including impaired face processing. However, the mechanism(s) underlying this observation are not yet known. Eye movement studies have found that people with SCZ do not direct their gaze to the most informative regions of the face (e.g., the eyes). This suggests that SCZ patients may be less able to extract the most relevant face information and therefore have decreased calculation efficiency. In addition, research with non-face stimuli indicates that SCZ is associated with increased levels of internal noise. Importantly, both calculation efficiency and internal noise have been shown to underpin face perception among healthy observers. Therefore, the current study applies noise masking to upright and inverted faces to determine if face processing deficits among those with SCZ are the result of changes in calculation efficiency, internal noise, or both. Consistent with previous results, SCZ participants exhibited higher contrast thresholds in order to identify masked target faces. However, higher thresholds were associated with increases in internal noise but unrelated to changes in calculation efficiency. These results suggest that SCZ-related face processing deficits are the result of a decreased noise-to-signal ratio. The source of increased processing noise among these patients is unclear, but may emanate from abnormal neural dynamics.

**Keywords: Schizophrenia, face perception, internal noise, calculation efficiency, face orientation**

# **INTRODUCTION**

People with Schizophrenia (SCZ) are impaired in recognizing and discriminating human faces in both upright and inverted orientations (Archer et al., 1992; Whittaker et al., 2001; Sachs et al., 2004). Moreover, although SCZ participants demonstrate visual processing deficits across a broad assortment of stimuli, face stimuli may be particularly problematic in this regard. For example, in a study by Chen et al. (2008), patients with SCZ were less accurate when locating or matching line drawings of faces compared to similar drawings of trees. This suggests that the ability to detect faces, compared to other visual stimuli, may be disproportionately impaired in this patient population1 . Furthermore, when detecting trees, both patients with SCZ and healthy participants displayed similar stimulus inversion effects. However, stimulus inversion effects for faces were observed to be significantly reduced in people with SCZ<sup>2</sup> . The reduced face inversion effect observed in the patient population, in addition to the lack of an inversion effect when detecting trees, suggests an impairment that is particular to face detection in SCZ. Moreover, SCZ-related deficits in face discrimination increased significantly as the duration of the interval between the initial and target faces increased, suggesting that face deficits may exist in working memory, as well as perception (Chen et al., 2009). These results, and others (Novic et al., 1984; Walker et al., 1984; Phillips and David, 1995; Gur et al., 2002) establish that SCZ participants are impaired when processing faces and suggest that this deficit is not likely to be fully accounted for by a general visual processing impairment across all classes of objects (e.g., trees).

Moreover, it is broadly accepted that faces convey an enormous range of socially relevant information about one's identity, gender, age, ethnicity, mood, attractiveness, level of interest, current focus of attention, and/or intentions (Haxby et al., 2002; Little et al., 2011). In turn, this information influences an array of social phenomenon including social categorization (Quinn and Macrae, 2011), discrimination, stereotyping and prejudice (Quinn et al., 2003), judgment of others' emotional state and empathy (Freitas-Magalhães, 2011; Eisenberger, 2012), and romantic attraction, attachment, and friendship (Gobbini et al., 2004; Wang et al., 2010). These observations are relevant to the investigation of

<sup>1</sup>The study by Chen et al. (2008) is the only experiment known to the authors that directly compares the processing of faces to other objects with similar visual characteristics. It should also be noted, however, that detection accuracy rates from this study show that face processing is more difficult (lower accuracy in both SCZ and HC groups). Consequently, the group by object-type interaction may reflect a difficulty confound (Chapman and Chapman, 1973) and, in turn, could suggest that face processing is inherently more complex. 2It is noted that while many studies have shown a reduced face inversion effect

among people with SCZ, others (e.g., Chambon et al., 2006; Butler et al., 2008)

have shown intact face inversion effects within this population. While the reasons for these discrepancies are unknown, it is possible that methodological differences across studies may account for these findings.

SCZ since patients with this disorder also demonstrate prominent social deficits (Penn et al., 1997; Tulloch et al., 2006) and many researchers have directly linked SCZ-related alterations in face processing with social impairment (e.g., Marwick and Hall, 2008). Consequently, investigations aimed at understanding the mechanisms of altered face processing also hold the promise of elucidating determinants of social dysfunction among in persons with SCZ.

However, despite reliable evidence of impaired face processing among people with SCZ, few studies have assessed the mechanisms potentially responsible for this finding. A common speculation is that impaired face processing is a result of compromised neural functioning in the fusiform face area (FFA), or Brodmanns area 37. This suggestion is based on previous findings demonstrating reduced volume of the fusiform gyrus compared to other brain regions (Highley et al., 1999; Lee et al., 2002; Onitsuka et al., 2003). Additionally, reduced fusiform volume has shown to be associated with facial emotion recognition deficits in the population (Goghari et al., 2011). However, functional activation (as ascertained by fMRI) of the FFA has been indistinguishable between healthy and SCZ participants when performing face discrimination tasks (e.g., Yoon et al., 2006). These findings led Yoon et al. (2006) to suggest that other cortical mechanisms are likely responsible for impaired face processing in SCZ.

#### **CALCULATION EFFICIENCY AND INTERNAL NOISE**

Signal detection theory (SDT) (Green and Swets, 1966) provides an alternative way of conceptualizing how face perception differs between healthy individuals and those with SCZ. According to SDT, observers make responses by comparing an internal response evoked by a stimulus to a decision criterion. In classic formulations of SDT, the internal response is unidimensional. However, it is not the case that SDT applies only to stimuli that vary along a single dimension (e.g., tones that vary in intensity); in fact, the theory has been applied successfully in many contexts using multidimensional stimuli (Swets, 1996). SDT casts an internal response as an abstract decision variable rather than a direct response to a simple stimulus attribute like intensity. According to this idea, the decision variable is an index of the information relevant to a particular decision (e.g., a stimulus is or is not present; the face belongs to person A or person B) that may be calculated from several stimulus attributes. The nature of this calculation—which stimulus attributes are encoded, and how they are distilled into a single decision variable—influences the amount of information conveyed by the decision variable and, therefore, constrains performance in perceptual tasks. In the current study, the degree to which a decision variable calculation captures the available stimulus information will be referred to as calculation efficiency. In the case of face identification, a great deal of information is concentrated near the eyes and eye brows (Sekuler et al., 2004; Gaspar, 2006; Keil, 2008), and therefore an efficient calculation would utilize a decision variable based on the distribution of contours in those regions. An inefficient calculation, on the other hand, would derive a decision variable based on subtle changes in contrast from other less informative face areas (e.g., the forehead). Under this assumption, a proficient way to enhance the signal related to a target face (i.e., increase discrimination) is to use perceptual information from that face that best characterizes its uniqueness. Hence, one possible explanation for face perception deficits in people with SCZ is that they base their decisions on less informative aspects of faces (i.e., they have lower calculation efficiency). Evidence consistent with this idea has been described by Williams et al. (1999), who measured the eye movements of SCZ and healthy participants while they viewed human faces. Compared with healthy subjects, people with SCZ fixated less on the most informative regions of the face.

A second fundamental assumption in SDT is that internal responses are probabilistic: the internal response to an identical stimulus will vary across multiple presentations of the same stimulus. From a visual processing point of view, this internal variation, or noise, may arise from a variety of sources, such as jitter in eye position, fluctuations in attentiveness, or random fluctuations in the responses of sensory neurons. Obviously, internal noise degrades perceptual representations and limits performance in perceptual tasks. Hence, within the framework of SDT, poorer face perception in people with SCZ might be caused by elevated internal noise in face-processing mechanisms (Winterer et al., 2000; Rolls et al., 2008).

#### **NOISE MASKING FUNCTIONS**

How can we estimate internal noise and calculation efficiency? In this section we describe a psychophysical framework for estimating these quantities. Consider a simple face discrimination task: on each trial a subject is shown one of two faces and must decide which face was presented. Across trials, face contrast is varied to estimate a discrimination threshold (i.e., the contrast necessary for an observer to respond correctly on 70% of the trials). Finally, in this hypothetical experiment we will measure discrimination thresholds for faces embedded in *external noise* (i.e., a noise mask). Specifically, a zero-mean random number is added to the brightness, or contrast, at each pixel in the visual display. If the noise is independent at each pixel, then the noise is said to be "white." Moreover, if the random numbers are selected from a Gaussian distribution, then the noise is said to be Gaussian and the strength of the noise is related to the variance of the distribution. Our hypothetical experiment will measure discrimination thresholds with white Gaussian noise that varies in strength. The resulting threshold-vs.-noise curve is referred to as a *noise masking function*.

Pelli (1981) outlined a simple framework for interpreting noise masking functions (**Figure 1**): an observer receives a physical stimulus—in our case, a face embedded in external noise—which is transformed into an internal representation. An internal noise is added to the internal representation, and a calculation is performed that converts the internal signal-plus-noise variable into a decision variable. The variance of the internal noise and the nature of the calculation are assumed to be independent of stimulus contrast. Given this framework, discrimination threshold is related to the strength of the internal noise by the equation:

$$
\sigma\_{\rm rms}^2 = k(\sigma\_e^2 + \sigma\_i^2) \tag{1}
$$

**FIGURE 1 | A model of a human observer in a perceptual discrimination task.** The observer (i) transduces a noisy external stimulus; (ii) adds an internally-generated noise; (iii) applies a calculation that transforms the internal representation into a decision

variable; and (iv) uses the decision variable to make a decision. The variance of the internal noise and the nature of the calculation are assumed to be independent of the contrast of the input. Adapted from Pelli (1981).

where *c*<sup>2</sup> rms is threshold expressed as the squared rms contrast, or contrast variance, σ<sup>2</sup> *<sup>e</sup>* is the variance of the external noise, and *k* and σ<sup>2</sup> *<sup>i</sup>* are free parameters. The parameter <sup>σ</sup><sup>2</sup> *<sup>i</sup>* often is referred to as equivalent input noise because it is equal to the variance of the external noise that must be added to the stimulus to double the threshold relative to a no-noise baseline condition. In this framework, the threshold doubles when σ<sup>2</sup> *i* equals σ<sup>2</sup> *<sup>e</sup>*, and therefore the equivalent input noise can be used as an estimate of the level of internal noise. Parameter *k* indicates the rate at which the threshold increases with increasing external noise, assuming internal noise remains constant, and is related inversely to the efficiency of the observer's internal calculation. The values of *k* and σ<sup>2</sup> *<sup>i</sup>* are thought to reflect the influence of different processes and, therefore, measuring noise masking functions, as opposed to measuring a single threshold, provides more information about the processes that constrain perception.

Equation 1 predicts that threshold, expressed as*c*<sup>2</sup> rms, should be a linear function of the external noise variance. **Figure 2** illustrates the effects of the parameters σ<sup>2</sup> *<sup>i</sup>* and *k* on the masking functions predicted by Equation 1. In **Figure 2A**, the two masking functions differ only in terms of σ<sup>2</sup> *<sup>i</sup>* , and the resulting noise masking curves have the same slope but have different x-axis intercepts. In **Figure 2B**, the masking functions differ only in terms of *k*, and the resulting curves have the same x-axis intercepts but differ in terms of slope. Thresholds from a wide variety of tasks (Legge et al., 1987; Tjan et al., 1995; Dosher and Lu, 1998, 2000; Bennett et al., 1999; Gold et al., 1999, 2004, 2005; Pelli and Farell, 1999; Betts et al., 2007), including the discrimination of upright and inverted faces (Gaspar et al., 2008) are consistent with the predictions of Equation 1, including their linear relationship with masking noise.

## **OBJECTIVES OF THE CURRENT STUDY**

factors).

Previous studies have shown that face perception is impaired in persons with SCZ. It is unclear, however, if these group differences in performance reflect differences in internal noise, calculation efficiency, or both factors. If the two groups differed primarily in terms of internal noise, then we would expect the noise masking functions in the two groups to differ like

change in internal noise or calculation efficiency (or a combination of both

those shown in **Figure 2A** (i.e., discrepant intercept parameters). On the other hand, if the two groups differed primarily in terms of calculation efficiency, then the masking functions for the groups should differ like those shown in **Figure 2B** (i.e., discrepant slope parameters). To investigate these issues, the current experiment measured noise masking functions with upright and inverted faces in both healthy subjects and those with SCZ.

# **MATERIALS AND METHODS**

#### **PARTICIPANTS**

Twenty-three people with SCZ (11 females, 12 males) and 24 healthy participants (12 females, 12 males) participated in this study. All patients met criteria for SCZ, Schizoaffective Disorder, or Schizophreniform Disorder, as confirmed by the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID I) (First et al., 2001), but did not meet criteria for any other Axis I disorders, which was an exclusionary criterion. Healthy controls did not meet diagnostic criteria for any Axis I disorders. Structured clinical interviews were administered by either senior graduate students or research assistants, all of whom had received formal training in diagnostic interviewing. All participants had no self-reported history of neurological illness, brain injury, learning disability, current or past substance dependence, or medical conditions which could affect cognitive performance (e.g., coronary heart disease, type I diabetes). Participants also were excluded if they were taking psychotropic medication with known cognitive affects, including tricyclic antidepressants, anticholinergics, or benzodiazepines. Control participants were excluded if they reported having a first-degree relative with a SCZ-spectrum illness. Group means for healthy and patient participants were equivalent on age, years of education, an estimated Full Scale Intelligence Quotient (FSIQ) from the Wechsler Adult Intelligence Scale (Matrix Reasoning and Information subtests), 3rd Edition (Wechsler, 1997) and Wide-Range Achievement Test-III Reading subtest (WRAT-III; Wilkinson, 1993). In contrast, patients scored significantly lower the Hopkins Verbal Learning Test (HVLT) compared to controls (Brandt and Benedict, 2001). Patients with SCZ were administered the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987). In addition, both patients and controls were administered several scales from the Personality Assessment Inventory (PAI), including the Depression (Dep), Alcohol Problems (Alc), Drug Problems (Drg), Positive Impression Management (PIM), and Negative Impression Management (NIM) scales (Morey, 1991). Patients demonstrated significantly higher scores on the Dep, Alc, Drg, and NIM scales, but a significantly lower score on the PIM scale. Mean scale indices, however, were all within normal limits. Additionally, no single subject scored in a range suggesting deliberate distortion of their responses across both validity scales (i.e., PIM and NIM). All patients were medicated on a single antipsychotic agent [mean chlorpromazine = 416.67 mg (*SD* = 267*.*86)]. Patients' mean number of years since diagnosis was 8.04 (*SD* = 8*.*05) ranging from 1 year to 29 years. **Table 1** provides information characterizing the study participants.

# **STIMULUI AND APPARATUS**

Stimuli were generated by a Macintosh G4 computer and displayed on a CRT monitor using MATLAB and the Psychophysics Toolbox (Brainard, 1997; Pelli, 1997). The display had a frame rate of 85 Hz (non-interlaced) and a spatial resolution of 800 × 600 pixels, which from the viewing distance of 114 cm subtended 14.3◦ horizontally and 10.7◦ vertically. Face stimuli were based on digitized photographs of 10 frontal-view faces (5 males and 5 females), cropped to an oval window (width = 2.5◦;

**Table 1 | Means (SD) for demographic, neuropsychological, and clinical characteristics of the sample.**


*\*Indicates significant (p < 0.05) difference between controls and patients.*

*WRAT-3, Wide Range Achievement Test; HVLT-R, Hopkins Verbal Learning Test Revised; PANSS, Positive and Negative Symptom Scale.*

height = 3.39◦) that excluded areas showing chin, ears, and hair (see Gold et al., 1999 for details).

Thresholds were measured with a match-to-sample task. On each trial, two faces were selected randomly from the set of ten faces: one face was designated as the target, and the other as the distractor. A trial began with the presentation of a fixation cross at the center of the display. After a delay of 1 s, a noise-free version of the target face (rms contrast = 0.08) was presented for 200 ms centered at a location that was 2.29◦ above the fixation cross, and was followed by a high-contrast, static white noise mask that lasted for 200 ms. The offset of the mask was followed immediately by a 200 ms presentation of a pair of test faces—consisting of the target and distractor faces—centered 0.27◦ below and 2.29◦ to the left and right of the fixation cross. The target face could appear on the left or right with equal probability. The test faces were followed by the presentation (200 ms) of two high-contrast, static, white noise masks centered on the test faces (see **Figure 3** for an example of the experimental stimuli). The participant's task was to determine which one of the two test faces was the target face, and auditory feedback was provided after each trial to indicate correct and incorrect responses.

The contrast of the test faces was varied across trials using a 3-down/1-up staircase procedure to estimate face identification threshold, which was defined as the contrast necessary to achieve a correct response rate of 79%. A staircase ended after 12 reversals, and a threshold was calculated by taking the average of the last eight reversals. In separate blocks of trials, thresholds were measured with upright and inverted faces. Within each block of trials, thresholds were measured in a low-noise condition, in which test faces were presented without noise, and a high-noise condition, in which faces were embedded in static white noise that had a contrast variance of 0.04. In the high-noise condition, a different noise field was computed for each test face on every trial. Note that in both conditions the test faces were *followed* by the presentation of high-contrast noise masks. Trials in the low- and high-noise conditions were intermixed randomly. Subjects were given 10 practice trials in each condition prior to starting the experiment.

# **RESULTS**

Statistical analyses were performed with R (version 2.8.1; R Development Core Team, 2007). Effect size was expressed as Cohen's *f* (Cohen, 1988) using formulae described by Kirk (1995). For ANOVA tests in which *F <* 1, the effect size was assumed to be zero (see Kirk, 1995, p. 180).

Face identification thresholds measured with upright and inverted faces, expressed in terms contrast variance, are shown in **Figures 4** and **5**, respectively. Thresholds were submitted to a 2 (group) × 2 (orientation) × 2 (external noise) analysis of variance (ANOVA). The ANOVA revealed a significant main effect of group, *F(*1*,* <sup>45</sup>*)* = 13*.*70, *p <* 0*.*001, *f* = 0*.*26, indicating that thresholds were higher in people with SCZ.

Significant main effects of orientation, *F(*1*,* <sup>45</sup>*)* = 28*.*17, *p <* 0*.*001, *f* = 0*.*34, and external noise, *F(*1*,* <sup>45</sup>*)* = 28*.*27, *p <* 0*.*001,

high (right) external noise conditions. Participants were shown a target face (top) that was presented without noise. Following an inter-stimulus interval of 200 ms in which a static mask was presented, a pair of test faces consisting of the target and a distractor appeared. Participants were asked to discriminate the target face.

*f* = 0*.*38, also were found, indicating higher thresholds were obtained in conditions that used inverted faces and higher levels of external noise. None of the interactions were significant, *F*'s *<* 1, *p >* 0*.*33, all *f*s = 0.

To use Equation 1 to calculate internal noise for an individual participant, the slope of the noise masking function must be greater than zero (i.e., threshold must be higher in the highnoise condition). Some participants did not display this result and, therefore, internal noise could not be estimated for those subjects. Consequently, subjects who did not have higher noise thresholds in the high-noise condition were removed from the data for further analysis. To minimize the number of subjects that were removed, we applied the criterion (i.e., higher threshold in the high-noise condition) separately in the upright and inverted conditions. This procedure yielded slightly different subsets of subjects in the upright and inverted face conditions, and therefore the two face orientation conditions were analyzed separately.

In conditions using upright faces, data from four participants in the control group and three in the SCZ group were removed. Thresholds from the remaining 20 participants in each group were analyzed with a 2 (group) × 2 (external noise) ANOVA, which revealed significant main effects of group, *F(*1*,* <sup>38</sup>*)* = 10*.*73, *p* = 0*.*002, *f* = 0*.*35, and external noise level, *F(*1*,* <sup>38</sup>*)* = 49*.*48, *p <* 0*.*001, *f* = 0*.*55. The group × external noise interaction was not significant, *F(*1*,* <sup>38</sup>*)* = 0*.*03, *p* = 0*.*86, *f* = 0. In conditions using inverted faces, data from five participants in the control group and six participants in the SCZ group were removed. A 2 (group) × 2 (external noise) ANOVA on the remaining subjects found significant main effects of group, *F(*1*,* <sup>34</sup>*)* = 7*.*45, *p* = 0*.*01, *f* = 0*.*30, external noise, *F(*1*,* <sup>34</sup>*)* = 63*.*8, *p <* 0*.*001, *f* = 0*.*66; the interaction was not significant, *F(*1*,* <sup>34</sup>*)* = 0*.*37, *p* = 0*.*54, *f* = 0. These analyses indicate that (a) applying the criterion of having a higher threshold in the high-noise condition caused approximately equal numbers of participants to be removed from each group; and (b) the effects of group and external noise measuring in the subsets of subjects were similar to the effects obtained with the entire sample.

Equation 1 was used to estimate equivalent input noise and *k* for each participant who had a higher threshold in the high-noise condition. **Figure 7** shows that equivalent input noise did not vary systematically with face orientation: in the control group, equivalent input noise was slightly higher with inverted faces, but in the SCZ group it was slightly lower with inverted faces. In both the upright and inverted face conditions, however, average equivalent input noise was lower in the control group than in the SCZ group. One-tailed *t*-tests performed on log-transformed data indicated that the group difference was significant in the upright face condition, *t(*38*)* = 2*.*63, *p* = 0*.*006, *f* = 0*.*38, and at trend level significance in the inverted face condition, *t(*34*)* = 1*.*41, *p* = 0*.*08, *f* = 0*.*16.

Average values of *k* are shown in **Figure 6**. In both groups, *k* was higher (i.e., calculation efficiency was lower) in the inverted face condition. However, the difference between the control and SCZ groups was not significant with either face orientation

[upright: *t(*38*)* = 0*.*17, *p* = 0*.*57, *f* = 0; inverted: *t(*34*)* = 0*.*61, *p* = 0*.*73, *f* = 0].

# **DISCUSSION**

Consistent with previous research, this study found that SCZ participants are deficient in their ability to discriminate both upright and inverted faces compared to healthy observers. However, this study further sought to examine whether this effect was a result of increases in internal noise and/or decreased calculation efficiency. It was found that people with SCZ exhibit higher levels of internal noise when processing both upright and inverted faces; however, it is important to underscore that, while the group differences across the upright and inverted condition were of a similar magnitude, these differences were significant in the upright condition but trended toward significance in the inverted condition. In contrast, no between-group differences were observed on measures of calculation efficiency.

Our results suggest that internal noise is a substantive determinant of face discrimination deficits among people with SCZ. These results are also consistent with increasing evidence that signal-to-noise ratio is decreased among those with SCZ (Winterer et al., 2000; Winterer and Weinberger, 2004). Causes of increased internal noise in people with SCZ are unclear. Higher equivalent input noise could be caused by a variety of processes, such as jitter of the eyes or random fluctuations in the responses of sensory neurons. More recent work has suggested that increased noise in those with SCZ may be attributed to an alteration of brain dopamine levels, secondary to reduced NMDA and GABA receptors (Lang et al., 2007).

Although the source of internal noise is unclear, it has been hypothesized by Rolls et al. (2008) that neural noise underlies many of the cognitive and perceptual impairments associated with SCZ. According to the dynamical systems hypothesis, groups of neurons supporting a given behavior or cognition settle into a stable pattern, or attractor state. These attractors are then strengthened over time, and eventually partial activation of the neural network will be sufficient to activate the entire network. However, neural noise (i.e., random spiking) can destabilize the attractor state, and may even induce a sudden change from one attractor state to another, which may result in behavioral distraction or confusion. Rolls et al. (2008) propose that this noise-induced instability of attractor networks may be related to some of the impairments in SCZ, including impairments in memory and attention. While Rolls et al. suggest that the destabilizing of neural networks occurs in the prefrontal cortex, it is possible that the destabilization of neural networks may be a widespread general impairment in SCZ, and may explain much impairment, including face perception, in this population.

As noted above, results also indicate that there are no differences in calculation efficiency between healthy observers and those with SCZ. This is at odds with previously cited scanpath studies showing that people with SCZ direct their voluntary gaze at less informative regions of faces. A plausible explanation for this discrepancy is that scanpath studies actually do not only represent changes in efficiency, but also may be indicative of increased internal noise. In other words, scanpaths may be mediated by attentional processes, and as Rolls et al. (2008) suggest, attention may be deleteriously affected by the destabilization of neural attractor states. Although it is unknown how calculation efficiency and internal noise are related in this context, future research directed at examining this issue would be beneficial.

In the present study, we postulate that a decreased performance in face discrimination among people with SCZ may be a result of an increase in internal noise. It is plausible, however, that increased internal noise is not isolated to face processing, but may also underlie other visual perceptual deficits in SCZ. People with SCZ are also impaired in motion perception (Chen et al., 1999a,b; Clementz et al., 2007), visual context processing (Uhlhaas et al., 2005), and perceptual organization (Silverstein et al., 2006; Sehatpour et al., 2010). The methods as outlined in this study may also be used to disentangle whether increased internal noise contributes to deficiencies in other visual processes. In this context, many studies have demonstrated SCZ-related visual sensory processing deficits on tasks that preferentially involve the magnocellular visual system (Schechter et al., 2003; Butler et al., 2005, 2009; Keri et al., 2005; Martinez et al., 2008). More recently, Butler et al. (2009) investigated magnocellular and parvocellular contributions to emotion processing deficits using affective faces. Butler et al.'s results showed that deficits in low spatial frequency contrast sensitivity, thought to reflect magnocellular visual processing, correlated significantly with the ability to identify emotions from affective faces. Therefore, it is also plausible that SCZ-related performance deficits in affective face processing arise, at least in part, from magnocellular system dysfunction, which, in turn, may reflect greater internal (e.g., neural) noise.

The results from the current study also have clinical implications. As noted above, face processing is deficient among persons with SCZ, which may, in turn, underpin some of the social deficits associated with this disorder. In particular, important cues to others' emotional states and intentions are conveyed via facial expressions and facial affect recognition deficits are prominent features that appear early in the course of the illness and are stable over time (Penn et al., 2008). Deficits in this realm have been demonstrated using instruments such as the Reading the Mind in the Eyes Test (Köther et al., 2012) and the Mayer-Salovey-Caruso Emotional Intelligence Test (Dawson et al., 2012). Accordingly, several social cognitive remediation programs for patients with psychotic disorders have emerged, many of which explicitly train participants to decipher facial expressions as a means to understanding the emotional states of others (for a review see Fiszdon and Reddy, 2012). However, to the extent that such remediation efforts include explicit means for increasing visual focus on informative aspects of the face (e.g., eyes), these efforts may be somewhat misdirected. Although it may be the case that increased attentiveness to informative facial areas will positively influence discrimination or recognition performance across all participants, the current results would not predict disproportionate benefit for persons with SCZ. Instead, methods that may allow one to decipher signals in an otherwise noisy system may be more efficient. For example, sequential sampling models of decision making show that information leading to perceptual or cognitive decisions are quantitatively dependent on the accumulation of information over time via a random-walk process (Ratcliff and McGoon, 2008). In such models, greater noise necessitates greater time to make correct decisions—i.e., the time needed to accumulate adequate information (i.e., drift rate) is longer. Therefore, SCZ-related deficits in face processing may benefit from the simple intervention of encouraging patients to process the information for longer time periods. Similarly, pharmacologic manipulations known to

#### **REFERENCES**


increase neural signal relative to noise may also be a potential avenue for augmenting face processing in this population (Rolls and Deco, 2010).

This study is limited by a modest sample size, following the exclusion of participants necessary for the calculation of internal noise. In this context, replication of the study, in addition to a larger sample size, is essential to confirm these results. Regarding sample characteristics, although participants in the current study were not intentionally matched for education and estimated FSIQ, both healthy controls and people with SCZ demonstrated equivalent results in these domains. This may raise generalizability concerns, as equivalence across these dimensions is not typically observed, and so this patient sample may arguably represent higher functioning participants in the SCZ group. Furthermore, it is unknown whether these findings are specific to SCZ or whether similar results would be observed in other types of psychopathology. It would be useful, therefore, for future studies to explore such potential deficits among persons with a range of psychiatric diagnoses.

Additionally, the current study included the use of only two levels of external noise to determine calculation efficiency and internal noise. The noise masking function, however, could be better represented using multiple levels of external noise in order to obtain more accurate estimations of these measures. Finally, all participants with SCZ who took part in the experiment were medicated and we are unable to comment whether the results seen in the study were confounded by medication status. In this regard, future research would be well advised to include samples of unmedicated SCZ patients.

# **AUTHOR NOTE**

This research was funded by an operating grant to Bruce K. Christensen from The Canadian Psychiatric Research Foundation and by the Canada Research Chair Program (Allison B. Sekuler and Patrick J. Bennett).


Chen, Y., Norton, D., McBain, R., Ongur, D., and Heckers, S. (2009).Visual and cognitive processing of face information in schizophrenia: detection, discrimination, and working memory. *Schizophr. Res.* 107, 92–98. doi: 10.1016/j.schres.2008.09.010


with perceptual learning. *Nature* 402, 176–178. doi: 10.1038/46027


*Psychol. Bull.* 121, 114–132. doi: 10.1037/0033-2909.121.1.114


brain network in schizophrenia: an integrated neuroimaging study. *Arch. Gen. Psychiatry* 67, 772–782. doi: 10.1001/archgenpsychiatry. 2010.85


schizophrenia: from description to action. *Curr. Opin. Psychiatry* 19, 140–144. doi: 10.1097/01.yco. 0000214338.29378.29


processing in schizophrenia. *Psychol. Med.* 31, 499–507. doi: 10.1017/S0033291701003701


*Psychiatry Res.* 148, 205–216. doi: 10.1016/j.pscychresns.2006.06.002

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; accepted: 14 June 2013; published online: 16 July 2013.*

*Citation: Christensen BK, Spencer JMY, King JP, Sekuler AB and Bennett PJ (2013) Noise as a mechanism of anomalous face processing among persons with Schizophrenia. Front. Psychol. 4:401. doi: 10.3389/fpsyg.2013.00401*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Christensen, Spencer, King, Sekuler and Bennett. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any thirdparty graphics etc.*

# Deficient biological motion perception in schizophrenia: results from a motion noise paradigm

#### *Jejoong Kim1, Daniel Norton2, Ryan McBain3, Dost Ongur <sup>4</sup> and Yue Chen4 \**

*<sup>1</sup> Department of Psychology, Duksung Women's University, Seoul, South Korea*


#### *Edited by:*

*Brian P. Keane, Rutgers University Center for Cognitive Science, USA*

#### *Reviewed by:*

*Arseny A. Sokolov, Centre Hospitalier Universitaire Vaudois, Switzerland Masahiro Hirai, Jichi Medical University, Japan*

#### *\*Correspondence:*

*Yue Chen, Department of Psychiatry, Harvard Medical School, Mclean Hospital, Centre Building, 115 Mill Street, Belmont, MA 02478, USA e-mail: ychen@mclean.harvard.edu*

**Background**: Schizophrenia patients exhibit deficient processing of perceptual and cognitive information. However, it is not well-understood how basic perceptual deficits contribute to higher level cognitive problems in this mental disorder. Perception of biological motion, a motion-based cognitive recognition task, relies on both basic visual motion processing and social cognitive processing, thus providing a useful paradigm to evaluate the potentially hierarchical relationship between these two levels of information processing.

**Methods:** In this study, we designed a biological motion paradigm in which basic visual motion signals were manipulated systematically by incorporating different levels of motion noise. We measured the performances of schizophrenia patients (*n* = 21) and healthy controls (*n* = 22) in this biological motion perception task, as well as in coherent motion detection, theory of mind, and a widely used biological motion recognition task.

**Results:** Schizophrenia patients performed the biological motion perception task with significantly lower accuracy than healthy controls when perceptual signals were moderately degraded by noise. A more substantial degradation of perceptual signals, through using additional noise, impaired biological motion perception in both groups. Performance levels on biological motion recognition, coherent motion detection and theory of mind tasks were also reduced in patients.

**Conclusion:** The results from the motion-noise biological motion paradigm indicate that in the presence of visual motion noise, the processing of biological motion information in schizophrenia is deficient. Combined with the results of poor basic visual motion perception (coherent motion task) and biological motion recognition, the association between basic motion signals and biological motion perception suggests a need to incorporate the improvement of visual motion perception in social cognitive remediation.

**Keywords: biological motion perception, visual motion perception, bottom-up process, social cognition, schizophrenia**

#### **INTRODUCTION**

Schizophrenia is characterized by deficits at multiple levels of information processing, including perception and cognition. Within the domain of visual perception, a large body of research indicates that patients with schizophrenia experience global dysfunction, particularly within the dorsal visual pathway (Butler and Javitt, 2005; Butler et al., 2008, 2012; Chen, 2011; Silverstein and Keane, 2011). For instance, they exhibit prolonged visual backward masking effects (Green et al., 1994), poor velocity discrimination (Chen et al., 1999; Kim et al., 2006; Clementz et al., 2007), and deficient global motion perception (Stuve et al., 1997; Chen et al., 2003; Green et al., 2009).

Difficulty in social interaction is another area of dysfunction in schizophrenia (Frith and Frith, 1999). While perception presumably impacts social functioning, whether and how visual perception deficits contribute to social dysfunction in schizophrenia has yet to be thoroughly investigated. To address this issue, it is important to identify and examine functional domains that intimately involve both visual and social cognitive processes. Perception of biological motion (BM) refers to visual recognition of other people's actions that are portrayed solely by motion signals [e.g., point-light animations, (Johansson, 1973)]. BM perception requires bottom-up integration of signals from basic visual motion perception along with top-down social cognition. For example, Neri and colleagues (1998) showed that when visual motion signals were degraded experimentally the perception of BM in healthy people collapsed (Neri et al., 1998), highlighting the importance of bottom-up processing in this task. Interestingly, other previous studies have also found that successful performance on BM perception tasks requires knowledgebased representations of biological organisms' typical actions. This type of evidence serves to reinforce the importance of

*<sup>2</sup> Department of Psychology, Boston University, Boston, MA, USA*

top-down processing in BM perception (Dittrich, 1993; Thornton et al., 2002). The involvement of both perceptual and cognitive systems was also shown in a recent electrophysiological study which reported earlier and later peaks of EEG in response to BM from scalp sites corresponding to area MT, superior temporal sulcus (STS) and the parietal "mirror neuron system"(Krakowski et al., 2011). Similarly, brain imaging studies have revealed that the posterior superior temporal sulcus (pSTS) and some parietal areas are selectively activated during BM perception (Grossman et al., 2000; Grezes et al., 2001; Vaina et al., 2001b; Pavlova et al., 2004; Peuskens et al., 2005; Krakowski et al., 2011). Some frontal areas including inferior frontal sulcus (IFS) and premotor area are also activated during BM perception, which suggests an involvement of higher-order processing and attention (Saygin et al., 2004; Saygin, 2007).

In clinical domains, patients with neurodevelopment disorders such as autism showed abnormal performances in perception of BM (Blake et al., 2003; Annaz et al., 2010; Koldewyn et al., 2010). Not only in autism, impaired recognition and detection of BM have been found in schizophrenia (Kim et al., 2005, 2011; Singh et al., 2011). Schizophrenia patients also exhibited deficits associated with both basic visual motion processing such as detection of coherent motion (Chen et al., 2005) and with higher order social cognitive processing, such as theory of mind (ToM) tasks (Baron-Cohen, 1995; Frith and Frith, 1999). It is still unclear, however, whether impairment in recognition of BM in schizophrenia represents an extension of deficient basic motion perception (i.e., a bottom-up problem), impaired social cognitive processing (i.e., a top-down problem) or both.

In BM tasks, point-light animations are used to depict various types of actions—including walking, jumping, kicking, and running. To successfully distinguish biological motion from non-biological motion observers may rely on knowledge-based top-down cognitive processes as well as bottom-up perceptual processes. Thus, conventional biological motion tasks have not been designed to effectively determine whether perceptual factors or social cognitive factors contribute to impaired performance in schizophrenia during BM perception.

To examine the role of visual motion perception in determining BM perception ability, we designed a motion noise BM paradigm. In this paradigm human action (walking) that was presented as point-light dots was used to make up a part of the stimulus. Visual motion noise was then added as the other part of the stimulus. This paradigm is similar to many other paradigms which have been used previously for studying BM perception (Cutting et al., 1988; Neri et al., 1998; Thornton et al., 2002; Thompson et al., 2007; Garcia and Grossman, 2008). Motion noise in this paradigm served to modulate the perceptual signal strength, which is required for the performance of a BM task. By systematically changing the levels of noise, we were able to directly evaluate the effect of the perceptual signal strength on perception of BM. We call this motion noise task paradigm "perceptual discrimination of BM" (p-BM), due to its focus on the perceptual dimension of the task.

The relationship between basic visual motion perception and the perception of BM can also be evaluated by a comparison of respective performances on several associated tasks (e.g., detection of coherent motion vs. recognition of biological motion). In this study we measured performance on both a basic visual motion task—detection of coherent motion (Newsome and Wurtz, 1988)—and on another BM task—recognition of BM (Blake et al., 2003; Kim et al., 2005). We also measured performance in a widely used theory-of-mind task—the Eyes Test (Baron-Cohen et al., 2001)—in order to evaluate the relationship between perception of BM and higher-order social cognition. Comparison of performance among these tasks provided supplemental information on how basic perceptual and higher level social cognitive processes are related to the perception of BM.

# **MATERIALS AND METHODS SUBJECTS**

Twenty-one patients with schizophrenia or schizoaffective disorder and twenty-two healthy controls participated in this study (**Table 1**). General inclusion criteria for both groups of participants were (1) between the ages of 18 and 55 years old, (2) no history of drug or alcohol abuse in the 6 months prior to participation, (3) no neurological problems such as seizure, stroke, or major head injury, and (4) Verbal IQ *>* 70.

Patients were recruited when they responded to advertisements posted on the campus of McLean Hospital as well as in the Greater Boston area. Patient diagnosis was established using the Structured Clinical Interview for DSM-IV [SCID, (First et al., 2002)] which were administered by independent clinicians who were blind to the purposes of the study and by evaluating available medical records. Psychotic symptoms of patients were assessed with the Positive and Negative Symptom Scale (PANSS) (Kay et al., 1987). Their mean positive, negative, and general scores of the PANSS were 15.2 [standard deviation *(SD)* = 6*.*5], 13.4 (*SD* = 5*.*5) and 29.2 (*SD* = 8*.*6), respectively. Their average prescribed antipsychotic dose, calculated using the chlorpromazine equivalent (Woods, 2003), was 429 (*SD* = 361) mg.

Healthy controls (HC) were recruited when they responded to advertisements posted on the campus of McLean Hospital as well as in the Greater Boston area. They were screened for exclusion of psychiatric illness using the non-patient version of the SCID-IV (First et al., 2002).

The verbal component of the Wechsler Adult Intelligence Scale-Revised (WAIS-R) (Wechsler, 1981) was administered to all participants. The two participant groups were similar in age and education, but the patient group scored lower in verbal IQ when compared to the control group. All participants had normal or corrected-to-normal vision.

The study protocols were approved by the Institutional Review Board of McLean Hospital.

#### **Table 1 | Demographic information of participants.**


# **PROCEDURES**

Participants performed all four tasks within the same research laboratory. All task procedures were implemented on a Macintosh G4 computer (Apple Inc. Cupertino, CA) which was placed in an otherwise dark room. Stimuli (see below for each task) were displayed on a ViewSonic CRT monitor GS 790 (ViewSonic Corp. Walnut, CA). A chinrest was used to stabilize the head of subjects and to maintain the viewing distance (57 cm). The entire procedure took approximately 40 min to complete. When needed, resting breaks were available between tasks.

# *Task 1: perceptual discrimination of biological motion (p-BM)*

This task was specifically used to assess the perceptual capacity for discerning BM. The target was a point-light animation [12 dots on the head and major joints of the body; see details in (Blake et al., 2003)] of one type of BM—walking (leftward and rightward). The size of each dot was 5-arc min with the average speed within a sequence of 4◦/s, and each sequence consisted of 20 frames. This target was embedded in a number of noise dots, and they together constituted a stimulus for perceptual discrimination of BM (**Figure 1**). The proportion of target dots in the BM stimulus is considered the perceptual signal; a large percentage of target dots provides a strong perceptual signal and therefore makes the task easier.

Noise dots consisted of duplicated dots from the original biological walkers (Cutting et al., 1988; Saygin et al., 2010), but with the following manipulations applied: half of the motion paths of the noise dots were generated from a walker moving rightward, whereas the other half were generated from a walker moving leftward. The makeup of these noise patterns was different from many previous paradigms in which the noise was derived from just one particular type or direction of action such as the movement of a leftward or a rightward walker alone. This "half and half " methodological modification allowed our paradigm to provide a more balanced noise profile in terms of randomness of motion direction and also allowed us to adjust signal strength at basic visual motion processing levels. The starting points of the noise dots were repositioned to random spatial locations.

**FIGURE 1 | An illustration of stimuli used in Task 1 (perceptual discrimination of BM).** Here, for the purpose of illustration, signals dots are signified in a more salient color (dark). Noise dots are signified in a less salient color. The arrows indicate two possible walking directions (left and right). In an actual display, the two types of dot were in the same color and the arrows were not shown.

This perceptual discrimination paradigm used only one type of BM (walking) and a simple task of direction discrimination to minimize the requirement for knowledge-based top-down information about a variety of prototypes of the BM. That is, participants did not have to reallocate extra attentional resources to determine whether one of many BMs or a scrambled motion would appear. Rather, since a walker was always present, participants could simply focus on discriminating between two walking directions by interpreting the kinematical information in the spatiotemporal pattern of the stimulus. As a result, performance on this BM task relies primarily on the bottom-up process based motion perception.

The stimulus was generated within MatLab's (Mathworks Inc., Natick, MA) Psychophysics Toolbox (Brainard, 1997; Pelli, 1997) programming environment. The entire array of dots, including noise dots, had a size of 7 × 7◦ in visual angle, and was displayed for 1 s in each trial. The size of the target (walker) was approximately 4◦ (height)×3◦(width). Participants indicated the direction of walking (leftward or rightward) by pressing one of the two pre-assigned buttons, and were instructed to guess when necessary. The total number of noise dots varied between 12, 24, 48, 96, or 192. These five noise levels corresponded to five signal-tonoise ratios for the stimulus—100.0%, 50.0%, 25.0%, 12.5%, and 6.3%. The different signal-to-noise ratios were presented across trials according to the method of constant stimuli. Each combination of noise condition and walking direction was repeated 10 times. The total number of trials was 100. Compared to other methods (e.g., QUEST, a staircase threshold estimation procedure), the method of constant stimuli is more thorough but less efficient, but allowed us to show the stimulus conditions (or noise level) under which the performance of patients and the performance of controls do and do not differ. The data generated from this method provide an illustration of the pattern of performance change with stimulus condition.

One performance measure we utilized was the accuracy with which participants identified the moving direction of the walker presented at each of the five signal-to-noise ratios. Another performance measure was the participant's perceptual threshold, which was defined as the maximum noise level at which participant's performance reached 80% accuracy. Unlike many perceptual threshold metrics, for the purposes of this study a higher threshold value corresponds with a better performance.

# *Task 2: detection of coherent motion (CM)*

This task has been widely used to assess the capacity of perceptual processing of visual motion information (Newsome and Wurtz, 1988), including in patients with neurological disorders (e.g., Vaina et al., 2001a) and schizophrenia (Stuve et al., 1997; Li, 2002). The stimulus consisted of signal dots moving coherently toward one direction (left or right), and noise dots moving in random directions. Those two portions of dots were intermixed and randomly distributed in space through a circular window. The proportion of signal dots (coherence level) in the random dot pattern (RDP) is considered the motion signal strength or task difficulty; a larger percentage of the signal dots corresponds with a stronger motion signal and therefore makes the task easier.

The stimulus was generated within the C programming environment. The stimulus had a size of 7◦ in diameter. In each trial, an RDP was presented for 400 ms. Participants indicated the direction of coherent motion (left or right) by pressing one of two designated keys, guessing when necessary. Six levels of motion coherence (0, 5, 10, 20, 40, and 100%) were presented in a random order across trials according to the method of constant stimuli. Performance was measured in two ways: (1) the participant's response accuracy at each of the six motion coherence level, (2) his/her perceptual threshold, which was defined as the minimum coherence level at which participant's performance level reached 80% accuracy. A lower threshold value corresponds with a better performance.

#### *Task 3: recognition of BM (r-BM)*

This task was used to assess a participant's ability to discriminate BM from non-biological, scrambled motion. The BM recognition task has been used in various clinical populations including patients with autism (Blake et al., 2003), schizophrenia (Kim et al., 2005) and obsessive-compulsive disorder (Kim et al., 2008), as well as in infants (Hirai and Hiraki, 2005). Stimuli consisted of point-light animations depicting various prototypes of human actions (e.g., walking, kicking, jumping, throwing, and so on), and their spatially scrambled versions. The scrambled versions were generated by randomizing the initial positions of each dot in their corresponding point-light BMs (Grossman et al., 2000; Blake et al., 2003).

The stimulus was generated within the MatLab/Psychophysics Toolbox programming environment. The stimulus had a size of approximately up to 6◦(height) × 4◦(width) in visual viewing angle and was displayed for 1 s in each trial. The biological and scrambled motions were presented in a random order across trials. The task was to indicate whether a given stimulus in each trial was a biological or scrambled motion by pressing one of two preassigned keys. There were 25 prototypes of BM, each of which had two facing directions, comprising 50 different BMs in total. The total number of trials was 100, including 50 presentations of BM and 50 presentations of scrambled motion.

Participants could not predict which type of action would be presented for each trial. Performance on this task thus relied not only on perceptual processing of motion signals, but also on knowledge-based cognitive representations of biological action prototypes.

Performance was measured by discrimination sensitivity (*d*'), which is defined as the difference of standardized "hits" (BM responses to BM stimuli) and standardized "false alarms" (BM responses to scrambled motion stimuli). A higher value of discrimination sensitivity corresponds to a better performance.

#### *Task 4: eyes test*

This task was used to assess the ability of participants to recognize emotional expressions of other people based upon an image of the eye region of their face (Baron-Cohen et al., 2001). This ability is closely associated with theory of mind (ToM) of social functioning. The original Eyes Test (the revised version of the "Reading the Mind in the Eyes" task) was in paper version. For this study, the Eyes Test was converted to an electronic version; the images were displayed on a computer screen within the MatLab/Psychophysics Toolbox programming environment. There were 36 images, and each image remained on the screen until a response from the participant was registered. The task was to view each image and then choose a word (out of four options) that best described what the person in the given image was feeling or thinking.

Performance was measured by the proportion of responses for which the word that correctly describes an emotional expression was chosen.

## **RESULTS**

# **PERCEPTUAL DISCRIMINATION OF BIOLOGICAL MOTION (p-BM)**

A Two-Way ANOVA (group × signal strength) on performance accuracy showed a significant main effect for signal strength [*F(*4*,* <sup>160</sup>*)* = 103*.*10, *p <* 0*.*001] and for group *(*1*,* 40*)* = 7*.*76, *p* = 0*.*008). The interaction effect between group signal strength was not significant [*F(*4*,* <sup>160</sup>*)* = 1*.*90, *p* = 0*.*11]. This study had limited statistical power due to the moderate sample size, therefore we have chosen to compare the group differences for each signal strength (noise) level. *Post-hoc* tests showed that the performance accuracies of schizophrenia patients were significantly lower than those of healthy controls when perceptual signals were moderately degraded (*p* = 0*.*006, Cohen's *d* = 0*.*88 for 100.0%; *p* = 0*.*005, Cohen's *d* = 0*.*91 for 50.0%; and *p* = 0*.*012, Cohen's *d* = 0*.*81 for 25.0%)<sup>1</sup> . When perceptual signals were more substantially degraded, the performances of the two groups did not differ significantly (*p* = 0*.*13, Cohen's *d* = 0*.*48 for 12.5%; and *p* = 0*.*19 Cohen's *d* = 0*.*44 for 6.3%) (**Figure 2**) and (**Table 2**).

1With Bonferroni correction for multiple comparisons under the five stimulus conditions (*p <* 0*.*01), the group differences remained significant for the conditions of 100.0% (*p* = 0*.*006) and 50.0% (*p* = 0*.*005), and became statically non-significant for the 25.0% condition (*p* = 0*.*012).

function (Weibull equation) for each subject group.

Perceptual discrimination thresholds, defined as a maximum noise level at which participant's performance maintained at 80% accuracy, were 73.14% (*SD* = 42*.*36%) for controls and 42.51% (*SD* = 40*.*96%) for patients, which yielded a significant group difference (*p* = 0*.*02, Cohen's *d* = 0*.*74) (**Figure 3B**). This result indicates that schizophrenia patients tolerated a smaller number of noise dots than healthy controls in order to adequately perform the task.

When verbal IQ score, the only behavioral variable differing between patients and controls, was used as covariate, an ANOVA on performance accuracy yielded similar results to the original analysis [*F(*4*,* <sup>156</sup>*)* = 2*.*78, *p* = 0*.*03 for signal strength, *F(*1*,* <sup>39</sup>*)* = 5*.*31, *p* = 0*.*03 for group, and *F(*4*,* <sup>156</sup>*)* = 1*.*44, *p* = 0*.*22 for interaction].

#### **TASK 2: DETECTION OF COHERENT MOTION (CM)**

A Two-Way ANOVA (group × signal strength) on performance accuracy showed a significant main effect for signal strength [*F(*5*,* <sup>165</sup>*)* = 25*.*92, *p <* 0*.*001] and for group *(*1*,* 33*)* = 4*.*42, *p* = 0*.*04). The interaction effect between group signal strength was not significant [*F(*4*,* <sup>165</sup>*)* = 1*.*33, *p* = 0*.*25]. Performance accuracies of schizophrenia patients were lower than those of healthy controls, yet *post-hoc* tests showed that the group differences did


**perception and BM perception tasks. (A)** Perceptual threshold for coherent motion detection (CM), **(B)** Perceptual discrimination threshold (r-BM). **(D)** Accuracy for the Eyes Test. Error-bars indicate one standard error (SE).

not reach a conventional statistical criterion level at each motion coherence condition (except for the 100% condition).

Perceptual thresholds for detecting coherent motion (the minimum coherence level guaranteeing 80% accuracy) were higher (lower performance level) in patients than in controls [controls: 25.05 (*SD* = 16*.*05), patients: 43.25(*SD* = 36*.*06)]; the group difference was marginally significant (*p* = 0*.*049, Cohen's *d* = 0*.*65). That is, compared to controls, patients required a higher level of motion coherence or stronger motion signals in order to reach the designated performance accuracy level (80%) (**Figure 3A**) and (**Table 3**).

When the verbal IQ score was used as a covariate, an ANOVA yielded non-significant effects for signal strength [*F(*5*,* <sup>160</sup>*)* = 2*.*18, *p* = 0*.*059], and group [*F(*1*,* <sup>32</sup>*)* = 2*.*24, *p* = 0*.*14], both of which differed from effects for the original analysis and in terms of the interaction effect [*F(*5*,* <sup>160</sup>*)* = 1*.*08, *p* = 0*.*37].

#### **TASK 3: RECOGNITION OF BIOLOGICAL MOTION (r-BM)**

The discrimination sensitivity of patients was significantly lower than that of controls (patients: *d* = 1*.*94 (*SD* = 0*.*69), controls: *d* = 2*.*59 (*SD* = 1*.*16), *p* = 0*.*03, Cohen's *d* = 0*.*68) (**Figure 3C**). Separate analyses of the hit rates and false alarm rates, the two components of discrimination sensitivity, revealed that the poor performance of patients was mainly due to their high false alarm rates [hits: *t(*41*)* = 0*.*52, *p* = 0*.*61; false alarm: *t(*41*)* = −2*.*68, *p* = 0*.*01]. The resulting false alarm rate indicated that patients were more likely to attribute non-biological, scrambled motion as BM, compared to controls. This result replicated one of our previous findings (Kim et al., 2011).

When the verbal IQ score was used as a covariate, an ANOVA yielded non-significant group effects in *d* [*F(*1*,* <sup>40</sup>*)* = 3*.*38, *p* = 0*.*072] and hit rate [*F(*1*,* <sup>40</sup>*)* = 0*.*29, *p* = 0*.*59], and a significant group effect in false alarm [*F(*1*,* <sup>40</sup>*)* = 6*.*16, *p* = 0*.*017].

## **TASK 4: THE EYES TEST**

The performance accuracy of the patient group was significantly lower than that of the control group [*t(*39*)* = 2*.*55, *p* = 0*.*015, Cohen's *d* = 0*.*79] (**Figure 3D**). For the total 36 images, the group mean (*SD*) of correct responses of the controls and of the patients were 27.15 (5.22) and 23.33 (4.34), respectively. This result confirms deficient social functioning in the patient group.

When the verbal IQ score was used as a covariate, an ANOVA yielded a non-significant effect for group [*F(*1*,* <sup>38</sup>*)* = 3*.*38, *p* = 0*.*074].

#### **RELATIONSHIP OF THE PERFORMANCES IN THESE TASKS**

We used Pearson's correlation to evaluate the relationships among performances of the four tasks. The analysis results are summarized in **Table 4**.

In the patient group, CM was significantly correlated with r-BM (*r* = −0*.*62) and the Eyes Test (*r* = −0*.*66). There was a moderate, yet non-significant correlation between CM and p-BM (*r* = −0*.*31).


*\*p < 0.05.*

**Table 4 | Correlations between the performances on the visual and cognitive tasks used in this study.**


*CM, Detection of coherent motion; p-BM, perceptual discrimination of biological motion; r-BM, recognition of biological motion. \*p <* 0*.*05*, \*\*p <* 0*.*01*.*

#### **RELATIONSHIP BETWEEN THE TASK PERFORMANCES AND OTHER CLINICAL VARIABLES**

In the patient group, neither positive nor negative PANSS scores were correlated with performances on perceptual tasks. Likewise, CPZ was not correlated with performance on any tasks.

Across all participants verbal IQ was significantly correlated with Eyes Test scores (*r* = 0*.*41, *p* = 0*.*008). Since the patients had lower scores in both the Eyes Test and verbal IQ compared with healthy controls, the significant correlation across all participants may be due to diagonal alignment of two clusters rather than a truly linear relationship. In the patient group alone, the correlation between verbal IQ and the Eyes Test was 0.42 (*p* = 0*.*062).

# **ADDITIONAL TESTING OF THE p-BM IN A SUBGROUP OF PARTICIPANTS**

As a follow-up, we assessed performance on the main task (p-BM) under a no-noise condition in a subgroup of participants (*n* = 8 for patients and *n* = 7 for controls). This no-noise condition, while not included during the initial testing, provides a baseline for comparisons with those conditions containing various levels of noise. For this condition, the performance accuracy was 100% (*SD* = 0%) for healthy controls and 99.38% (*SD* = 1*.*77%) for the patients. The group performances did not differ significantly (*p* = 0*.*37).

# **DISCUSSION**

The results of this study show that patients performed significantly worse at BM perception in the presence of visual motion noise. Compared to controls, perceptual discrimination of BM in the patients was impaired when perceptual signals were moderately degraded by motion noise. When perceptual signals were more substantially degraded, perceptual discrimination of BM was similarly impaired in both groups. This study also replicated previous results of deficient basic visual motion perception (Li, 2002; Chen et al., 2005; Slaghuis et al., 2007), deficient recognition of BM (Kim et al., 2005, 2011) and poor theory of mind (Frith and Corcoran, 1996; Abu-Akel, 1999) in this patient population. In the following sections, we consider the mechanisms that may contribute to deficient BM perception in schizophrenia.

## **BOTTOM UP PROCESSES FOR BIOLOGICAL MOTION PERCEPTION**

Performance in a BM task requires extraction and recognition of a prototype of human action from dozens of possibilities. Such a requirement involves information processing from both bottom-up and top-down systems. For example, a knowledge base about possible BMs must exist and be accessed, which is a top-down component important for recognition of these human actions. Thus, both bottom-up and top-down processes may be implicated in patients' impairments on the r-BM task.

Compared to the r-BM task, perceptual discrimination of BM (p-BM) restricts itself to a single prototype of human action (walking) and evaluates a simple visual feature of the action (walking direction). Such a design minimizes the influence of top-down cognitive processes like the retrieval of information from one's knowledge base about various other types of BM. Importantly, the manipulation of the independent variable motion noise—in this design is stimulus-based and does not involve top-down processes. For this reason changes in subjects' performance as a result of changes in this variable cannot be attributed to top-down abilities. Patients' poor performance on this task primarily implicates the bottom-up processes that support BM perception.

This vulnerability of BM perception to motion noise is not necessarily associated with a specific disease process. Children (6 years old and younger) under normal perceptual and cognitive development showed immature performances when motion noise was present (Freire et al., 2006). Adverse effects of such noise masking also appeared when adolescents with autism performed on perception of BM (Koldewyn et al., 2010) [but also see (Saygin et al., 2010)]. Along with the degraded performance on the CM task in the present study and in previous studies (Stuve et al., 1997; Chen et al., 2003), patients' poor performance on the p-BM task may reflect a problem of bottom-up processing in this disorder.

# **EFFECTS OF PERCEPTUAL MODULATION ON BIOLOGICAL MOTION PERCEPTION**

BM perception of patients and controls was substantially and comparably impaired in the presence of a high level of motion noise (96–192 dots, **Figure 2**). For a moderate level of motion noise (12–48 dots), however, patients' performance was significantly more degraded than that of controls. For the no-noise condition, the data from a subgroup of participants shows perfect or nearly perfect performance in both controls and patients. The results of degraded BM perception in the presence of visual motion noise may be interpreted in two ways. First, patients' lower performance level for the moderate noise conditions may be due to their deficient processing of BM information which is evident across all stimulus conditions and may be irrelevant to the presence of noise. This interpretation would be consistent with the existence of a generalized deficit in BM perception for patients even when no noise is present. Second, impairments in patients' performance at moderate noise levels could be particular to the presence of noise. This interpretation would suggest that patients and controls would have shown similar performance when no noise is present. The additional testing of patients (*n* = 8) and controls (*n* = 7) in this no-noise condition did in fact show this type of similar performance between the two groups, suggesting that the latter interpretation is the more likely scenario. The result from the r-BM task casts further light on the subject. In the r-BM task, one aspect of performance—hit rate—is analogous to the performance accuracy under the no noise condition of p-BM task. Both of these performance indexes measured the detection of a BM while no irrelevant stimuli (such as noise) were present. The result of a similar hit rate in patients and controls in the r-BM task (Task 3 in the Result section) is consistent with a data collected in a small group of additional patients and controls which showed similar performance while no noise was present. The sensitivity of the patient group to visual motion noise in these tasks highlights the role of visual motion signals in the processing of BM.

#### **RELATIONSHIPS WITH BASIC VISUAL PERCEPTION**

Given the putative relationship between basic visual processing and high level cognitive processing, one may consider that patients' deficit in BM perception is closely related to the deficit in basic visual motion perception. Both a non-significant correlation (p-BM vs. CM: *r* = −0*.*31) and a significant correlation (r-BM vs. CM: *r* = −0*.*62) were found in this study. This latter result is similar to that from a previous study in which a significant correlation between patients' performances on a CM task and on a BM recognition task was found (Brittain et al., 2011). The mixed correlation result is generally consistent with the notion that basic motion perception and BM perception may engage different cortical mechanisms (Vaina et al., 1990; Poom and Olsson, 2002).

An alternative interpretation would consider the role of form information. Random dot patterns, used in CM, do not contain any explicit or implicit form (except for the superimposed circular aperture which simply served as boundary). Although point-light animations, used in p-BM, do not have explicit form, spatiotemporal kinematics does draw form information of body shape. A recent study suggested that there are two critical features for precise perception of point-light walkers: upper body structure (form) and limb movements crossing each other (motion) (Thurman et al., 2010). Another recent study reported that observers were able to perceive global motion but not able to discriminate walking direction of BM when structural information was eliminated and motion information was intact (Lu, 2010). In this context, performance in CM task would only require integration of local motion signals into global motion while performance in p-BM task would require both a local motion integration and processing of implied form information as suggested by Giese and Poggio's model (Giese and Poggio, 2003). Therefore, schizophrenia patients' poor performance on p-BM task may implicate compromised processing of form information in addition to deficient motion processing (Takahashi et al., 2010). If this were the case, patients' performances on the two tasks should be partially correlated. Note that the interpretation of additional form processing for BM should be applicable to the relationship between the performances in CM and r-BM. The stronger correlation between patients' performances in the latter two tasks (*r* = −0*.*62) discounts this form processing interpretation.

# **ALTERED COGNITIVE PROCESSES**

BM perception requires attention (Cavanagh et al., 2001; Wang et al., 2011). A general attention problem may affect patients' performance on BM tasks. However, such an attention problem cannot be a primary factor here, as patient performance under different task conditions would be similarly degraded if driven by a gross attention deficit. This was not the case in this study, where patients were differentially impaired at moderate noise levels. Patients also seemed differentially impaired among different tasks. For example, after IQ was used as covariate, the large group difference in the performance of p-BM task remained whereas the moderate group difference in the performance of CM task as well as Eyes Test became non-significant. This non-significance also does not favor a general attention deficit interpretation.

One may wonder about the role of general intelligence in determining performance in these groups, especially in light of results from autism research which has shown that IQ can predict perceptual ability on BM tasks (Rutherford and Troje, 2012) and account for perceptual deficits on CM tasks (Koldewyn et al., 2010). In the present study, the verbal IQ of schizophrenia patients was not significantly correlated with BM but explained about 20% of the variability in the BM task, suggesting a relationship between the two tasks whose nature (e.g., spurious vs. causal, direction of causality if present) is undetermined. Degraded IQ is generally considered *an inherent part of schizophrenia*, but nevertheless our result showed that when this IQ variable was adjusted, patients' deficient performance remained on the BM tasks but not on the CM or the Eyes Test. This result suggests that a selective deficit in processing biological motion information exists in this mental disorder.

Despite remarkable phenomenological differences between autism and schizophrenia, an overlap of aberrant biological processes is suggested by recent research (e.g., McCarthy et al., 2009; Crespi et al., 2010). This overlap may be reflected in the presence of similar impairments beginning at more basic behavioral levels such as basic visual motion perception in both of the disorders (Stuve et al., 1997; Spencer et al., 2000; Chen et al., 2003; Koldewyn et al., 2010) [but also see (White et al., 2006; Chen et al., 2012)]. However, whether or not the problems of BM perception in autism and schizophrenia are of the same nature is unclear. In autism, it has been suggested that high level cortical processes are to blame for this impairment (Saygin et al., 2010; Koldewyn et al., 2011). In schizophrenia, the results of this study have shown that deficits in BM perception were related to basic visual motion perception. Just how basic visual processing problems are associated with impairments in high level cognitive processes (such as BM perception or IQ) remains a topic for further explorations, as this effect seems to differ between autism and schizophrenia.

#### **IMPLICATIONS ON GENERAL VISUAL AND COGNITIVE PROCESSES**

Recent studies have found an increased effect of visual modulation on other perceptual and cognitive processes in schizophrenia. One of our studies showed that patients' cognitive control of visually bi-stable images was more significantly influenced by the contrast level of visual stimulus than in controls (McBain et al., 2011b). Another study showed that patients' emotion perception (fear and happiness) was more substantially changed by a manipulation of spatial frequency of facial images (McBain et al., 2011a). It has also been shown that surrounding visual context influenced the action of finger-reaching toward a central target to a greater extent in patients than in controls (Chen et al., 2011). Along the same lines, this study found greater degradation of BM perception by visual motion noise in patients. The implications of increased interaction between basic visual motion perception and BM perception in schizophrenia are not immediately clear. But given the presence of deficient basic visual motion signal in this mental disorder, a stronger connection from basic motion processes to BM processes would be needed in order to utilize such weakened visual inputs during BM perception. Increased connectivity between the two levels of visual and cognitive processing may serve as one compensatory strategy.

#### **CORTICAL PROCESSING FOR BIOLOGICAL MOTION PERCEPTION**

The processing of BM information is primarily mediated in the superior temporal sulcus [for review, see Blake and Shiffrar (2007), Pavlova (2012)]. It has been shown that patients' cortical responses in this area were not selective to biological or scrambled motion (Kim et al., 2011). Given that schizophrenia is a brain disorder involving many cortical systems, it is important to ask if other cortical systems could potentially contribute to the processing of BM information. Currently, no neuroimaging data are available to directly address this question. The behavioral data of this study support the notion that BM processing in schizophrenia is more sensitive to signal modulation in the basic visual domain. A key test for this notion is to what extent the visual cortical areas such as MT and STS are functionally connected in patients. If the processing in the basic motion system contributes to a greater extent in higher level processing in the BM system as suggested by the behavioral data of this study, then increased functional connectivity between the two anatomically separate systems should exist in this mental disorder. A recent study on functional connections of cortical systems during resting states found that while local functional connectivity was reduced, global distances of functionally connected brain areas, or "connection distance," were increased in schizophrenia (Alexander-Bloch et al., 2013). The greater global functional connectivity suggested by this result is generally consistent with the suggestion that there may be heightened interaction among separate cortical systems (e.g., the visual cortex vs. the STS) in this mental disorder. A direct measurement of functional connectivity between the areas of interests would more definitely describe the relationships between basic visual processing and BM processing.

#### **VISUAL PROCESSING AND SOCIAL FUNCTIONING**

How basic visual processing deficits contribute to poor social functioning in schizophrenia is an important topic in schizophrenia research as this relationship could help inform the strategies of cognitive interventions. Based on a series of correlations among visual motion perception, BM perception and social cognition, Brittain et al. suggested that the processing of BM may act as an intermediate between visual perception and social behaviors (Brittain et al., 2011). Our study showed that patients' visual motion perception and bottom-up process driven BM perception (p-BM) were correlated with BM recognition (r-BM) which is supported by both bottom-up and top-down processes. Visual motion perception and the p-BM were also correlated with theory of mind, another aspect of social cognition. These results highlight the role of basic visual motion processing in socially meaningful tasks. Such a functional relationship suggests that the problems of processing visual and social cognitive information are likely associated in schizophrenia.

It is intuitive that the performances in BM tasks and in other social cognitive tasks like the Eyes Test are correlated. It is also intuitive that performances in the coherent motion and in the Eyes Test are correlated to a lesser extent, as the latter does not seem to be motion-related. The results of this study suggested otherwise: the correlation between perception of BM and the Eyes Test is lower than that between coherent motion detection and the Eyes Test. One way to understand such a result is to consider that the social brain receives common perceptual inputs and engages in two separate social cognitive processes, one dealing with dynamic signals (BM perception) and the other dealing with static signals (face processing, including signals involved in the Eyes Test). The common perceptual inputs, including those from coherent motion, feed into both social cognitive processes. Upon such a scenario, one would expect a robust correlation between the performances on coherent motion and the Eyes Test. Assuming that there are separate operations between dynamic and static social cognitive processes, one would also expect a weaker correlation between performances on BM and the Eyes Test.

Like a previous study (Kelemen et al., 2005), we found that patients' performances in CM and the Eyes Test were significantly correlated. This suggests that deficient visual motion processing in this mental disorder not only impacts motion-based but also non motion-based social cognition. The correlation between the performances in CM and the Eyes Test (*r* = −0*.*66) was as robust as the correlation between the performances in CM and r-BM (*r* = −0*.*62), which suggests that there are similar functional connections from the visual motion system to the social cognitive systems mediating BM perception and theory of mind.

Other visual perception deficits in schizophrenia may relate to social functioning problems differently. For example, Sergi and Green (2003) showed weak correlations between visual masking deficit and social functioning problems in patients. One may hypothesize that higher level visual processing like motion perception should be more strongly linked to social behavioral outcomes because this processing taps into social cognition. This hypothesis, while plausible, remains to be thoroughly tested by more systematic investigation.

# **LIMITATIONS**

One limitation of this study is that the data from the no-noise condition for the p-BM task were available only from a subgroup of participants. A more complete set of data would serve as a firm basis for comparison with those obtained under the noise conditions, and would more definitively indicate whether patients' degraded task performance under the noise conditions is due to the visual motion noise factor.

Another limitation is the use of medicated patients. It is difficult to exclude a medication effect on patients' visual and cognitive performances (e.g., Allen et al., 1997; Purdon et al., 2000). Yet, in this patient group, the CPZ dose equivalent was not correlated with the performance in any of the tasks used in the study. This suggests a minimal, if any, role of antipsychotic medications in the visual and cognitive tasks.

Still another limitation was the presence of a group difference in verbal IQ. It seems that when verbal IQ was taken into account in analyses the group differences remained for the BM task, but not for the CM task or the Eyes Test. Such a pattern of results points to a complex relationship between verbal IQ and perceptual/social cognitive measures. Given that the performance of high level cognitive tasks (e.g., the Eyes Test) relies upon verbal information, patients' low verbal IQ performance could potentially be a confounding factor when inspecting group differences in performances on social cognitive tasks. These limitations call for the results of this study to be verified with independent methods in future studies.

To summarize, this study found that in the presence of visual motion noise, BM perception was more substantially degraded in schizophrenia. Combined with patients' deficient performances in and relationships among basic motion perception, BM perception and theory of mind, the results of this study suggest that basic motion processing in schizophrenia plays an increased role in BM perception. The functional

# **REFERENCES**


Brainard, D. H. (1997). The psychophysics toolbox. *Spat. Vis.* 10, 433–436. doi: 10.1163/156856897X00357


relationships between different levels of information processing highlight the importance of including methods for improvement of reduced visual motion perception capacity for social cognitive remediation in this mental disorder.

#### **ACKNOWLEDGMENTS**

This study was supported in part by grants from the NIH (R01 MH 096793) and Harvard University. We thank Dr. Ken Nakayama for valuable discussions during the early stages of this research. We also thank Mr. Tor Ekstrom for technical assistance.

*Neuroscience* 172, 419–426. doi: 10.1016/j.neuroscience.2010.10.039


*Structure Clinical Interview for DSM -IV-TR Axis I Disorders - Nonpatient Edition (SCID-I/NP, 11/2002 revision)* New York, NY: Biometric Research Department, New York State Psychiatric Institute.


21, 2864–2875. doi: 10.1111/j.1460- 9568.2005.04106.x


182–186. doi: 10.1016/j.schres.2011. 04.004


11656–11661. doi: 10.1073/pnas. 191374198


1083–1089. doi: 10.1177/0956797 610376072


*Psychiatry* 64, 663–667. doi: 10.4088/JCP.v64n0607

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 January 2013; accepted: 11 June 2013; published online: 04 July 2013.*

*Citation: Kim J, Norton D, McBain R, Ongur D and Chen Y (2013) Deficient biological motion perception in schizophrenia: results from a motion* *noise paradigm. Front. Psychol. 4:391. doi: 10.3389/fpsyg.2013.00391*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Kim, Norton, McBain, Ongur and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# *Caitlín N. M. Hastings1, Philip J. Brittain2 and Dominic H. ffytche1\**

<sup>1</sup> Department of Old Age Psychiatry, Institute of Psychiatry, King's College London, London, UK

<sup>2</sup> Department of Psychological Medicine and Psychiatry, MRC Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, King's College London, London, UK

#### *Edited by:*

Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA

#### *Reviewed by:*

Randolph Blake, Seoul National University, Korea (South) Yue Chen, Harvard Medical School, USA

#### *\*Correspondence:*

Dominic H. ffytche, Department of Old Age Psychiatry, Institute of Psychiatry PO70, King's College London, De Crespigny Park, London SE5 8AF, UK e-mail: dominic.ffytche@kcl.ac.uk

**Background:** Biological motion perception is served by a network of regions in the occipital, posterior temporal, and parietal lobe, overlapping areas of reduced cortical volume in schizophrenia. The atrophy in these regions is assumed to account for deficits in biological motion perception described in schizophrenia but it is unknown whether the asymmetry of atrophy found in previous studies has a perceptual correlate. Here we look for possible differences in sensitivity to leftward and rightward translation of point-light biological motion in data collected for a previous study and explore its underlying neurobiology using functional imaging.

**Methods:** n = 64 patients with schizophrenia and n = 64 controls performed a task requiring the detection of leftward or rightward biological motion using a standard psychophysical staircase procedure. six control subjects took part in the functional imaging experiment.

**Results:** We found a deficit of leftward but not rightward biological motion (leftward biological motion % accuracy patients = 57.9% ± 14.3; controls = 63.6% ± 11.3 p = 0.01; rightward biological motion patients = 62.7% ± 12.4; controls = 64.1% ± 11.7; p > 0.05). The deficit reflected differences in distribution of leftward and rightward accuracy bias in the two populations. Directional bias correlated with functional outcome as measured by the Role Functioning Scale in the patient group when co-varying for negative symptoms (r = −0.272, p = 0.016). Cortical regions with preferential activation for leftward or rightward translation were identified in both hemispheres suggesting the psychophysical findings could not be accounted for by selective atrophy or functional change in one hemisphere alone.

**Conclusion:** The findings point to translational direction as a novel functional probe to help understand the underlying neural mechanisms of wider cognitive dysfunction in schizophrenia.

**Keywords: functional outcome, motion perception, STS, social cognition, translational motion, fMRI**

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 1 — #1

# **INTRODUCTION**

Visual function has long been recognized as altered in schizophrenia (Silverstein and Keane, 2011). Motion perception is one aspect of vision affected, with differences between patients and controls reported for motion coherence, velocity and luminance contrast (for review see Chen, 2011). There is an apparent hierarchy of motion effects within schizophrenia with global motion more affected than local motion (Chen et al., 2003) and biological motion more than global motion (Brittain et al., 2010a). Furthermore, at the top of the motion hierarchy, biological motion appears linked to an important outcome measure – the level of social functioning. Patients with poorer biological motion perception have less favorable social outcome, biological motion sensitivity correlating directly with social outcome (Kim et al., 2005) or indirectly through social perception (Brittain et al., 2010a). The neurophysiology, brain networks and psychophysics of motion perception are well understood, providing a useful model system from which to approach the underlying

neurobiology of wider cognitive dysfunction schizophrenia (Chen, 2011; Silverstein and Keane, 2011). In its link to social functioning, biological motion is of particular interest in this regard.

Motion perception involves a network of regions in the occipital, posterior temporal, and parietal lobes. In the occipital lobe, the primary visual cortex and its immediate surrounds (areas V1 and V2) respond to all classes of motion (Watson et al., 1993) while different sub-regions of the lateral surface of the occipital and posterior temporal lobes respond to different classes of motion. Biological motion is a term used for a class of motion first characterized by Johansson (Johansson, 1973) in which walking or movements such as jumping, running, kicking, throwing, crawling, shoveling, dancing are defined by point-light sources. Such stimuli have attracted considerable research interest due to their inherent combination of motion, form and action that may help reveal how such properties are integrated in the brain. The key brain regions implicated in previous studies are: (i) the posterior superior temporal gyrus (STG) and cortex surrounding the superior temporal sulcus (STS), bilaterally in some studies and predominantly right hemispheric in others (Bonda et al., 1996; Howard et al., 1996;Vaina et al., 2001; Peuskens et al., 2005; Peelen et al., 2006) (ii) regions in the ventral temporal lobe overlapping or in close relation to regions involved in face, object, figure, and kinetic contour processing (Vaina et al.,2001; Grossman and Blake, 2002; Peuskens et al., 2005) (iii) the cerebellum (Vaina et al., 2001; Sokolov et al., 2012) (iv) frontal cortex (Vaina et al., 2001; de Lussanet et al., 2008) and (iv) the parietal lobe (Bonda et al., 1996; Vaina et al., 2001).

It is assumed that functional or structural changes in the networks described above underlie the psychophysical motion deficits found in schizophrenia, a view supported by the finding that differences in functional activation for hits, false alarms and correct rejections in the posterior STS for biological motion stimuli differ in patients with schizophrenia from controls (Kim et al., 2011). Brain lesions in the parietal lobe/parieto-temporal junction (Battelli et al., 2003), superior temporal or inferior frontal regions (Saygin, 2007) and anterior temporal lobe (Vaina and Gross, 2004) are associated with deficits in biological motion. These regions typically have reduced cortical volume in structural imaging studies of schizophrenia. Although imaging findings vary from study to study, Shenton et al. (2001) in a review of the literature found 15/15 studies reporting a decrease in STG gray matter volume. Similarly, 9/15 studies reported volume reductions in the parietal lobe and 30/50 studies in the frontal lobe, particularly prefrontal cortex.

A consistent finding in structural imaging studies of schizophrenia is an asymmetry of atrophy in the left and right hemispheres. Within the network of areas linked to biological motion, the left STG is typically more affected than the right (Shenton et al., 2001). Similarly, the left inferior parietal lobule is typically more affected than right inferior parietal lobule (Niznikiewicz et al., 2000). The question therefore arises as to whether the asymmetry in hemispheric atrophy has a perceptual correlate. One aspect of motion perception that seems to be represented differently in each hemisphere is the direction of translational motion – the movement of an object or dot pattern across the visual field. Unlike primary visual cortex, which responds to stimuli in the contralateral hemifield only, motion specialized cortex (area V5) responds to motion in both contralateral and ipsilateral fields through interactions between the two hemispheres (Tootell et al., 1998; ffytche et al., 2000). Motion specialized areas thus respond to movement across the whole visual field, with evidence to suggest a bias of representation in each hemisphere. Patients with left hemispheric lesions have a predominance of leftward motion perception deficits while patients with right hemispheric lesions have a predominance of rightward motion perception deficits (Barton et al., 1995). This suggests a relative specialization for leftward translational motion in the left hemisphere and rightward translational motion in the right hemisphere. Evidence from an intraoperative study disrupting motion specialized areas in the right hemisphere through stimulation resulted in predominantly rightward motion perception deficits (Blanke et al., 2002), consistent with this view.

To date, most studies of biological motion have used stimuli that remain fixed, without translation across the visual field (i.e., a figure walking in place as if on a treadmill). It is therefore unclear whether leftward translational biological motion is linked to the left hemisphere and rightward translational biological motion to the right hemisphere, as seems to be the case for coherent motion. However, the existence of an asymmetry is hinted at by studies of biological motion figures *facing* leftward or rightward while walking in place. Leftward-facing figures walking in place in the left hemifield are associated with greater activation of right hemispheric frontal and parietal regions than rightwardfacing figures. Similarly, rightward-facing figures walking in place in the right hemifield are associated with greater activation in left hemispheric frontal and parietal regions than leftward-facing figures (de Lussanet et al., 2008). Leftward and rightward facing figures are also represented in spatially distinct sub-regions of the fusiform gyrus (Michels et al., 2009). Such findings lend support to the possibility of a difference in the representation of leftward and rightward translational biological motion in each hemisphere.

If leftward and rightward translation of biological motion are represented differently in each hemisphere, then asymmetrical atrophy within the biological motion network described in previous studies may be reflected as a difference in sensitivity to leftward and rightward biological motion translation. We have sought evidence to support this view using data from our previous study in which we found reduced sensitivity in schizophrenia to biological motion translation direction (Brittain et al., 2010a). Here we reexamine this data to establish whether the reduction in sensitivity identified related to one direction more than the other and report preliminary functional imaging evidence of the neurobiology of translation direction for biological motion stimuli.

# **MATERIALS AND METHODS**

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 2 — #2

Full details of patient recruitment and testing methods for the psychophysical study have been presented elsewhere (Brittain et al., 2010a,b). Patients with a DSM-IV diagnosis of schizophrenia (*n* = 64) were recruited from outpatient and inpatient facilities in South London and controls (*n* = 64) from local advertisement and a volunteer database. The study was approved by the Institute of Psychiatry Ethical committee and all subjects gave informed, written content. The two groups were matched for age, gender, level of education, visual acuity, and handedness but differed in IQ (estimated with the two-subtest version of the Wechsler Abbreviated Scale of Intelligence WASI, Psychological Corporation, 1999; patients = 101.91 ± [SD]15.24; controls = 107.37 ± 13.49). The inclusion criteria for both participant groups were: age between 18 and 65, English as a first language, no current alcohol or drug dependency, predominantly right handed (assessed using a six item version of the Annett Handedness Questionnaire Annett, 1970), no history of electroconvulsive therapy (none in the past 3 years for the patient group), no significant ophthalmological disease, sensory disability, history of epilepsy, or known neurological condition. Other tests performed of relevance to the analyses reported here are: (i) *Role Functioning Scale* (Goodman et al., 1993). This assesses functional outcome in four domains (working productivity, independent living/self-care, immediate social network relationships, extended social network relationships) with scores ranging from one (severely impaired) to seven (optimal). The Global Role Functioning Index (GFI) is the sum of the domain scores ranging from 4 (worst functioning) to 28 (best functioning) (ii) *Positive and Negative Syndrome Scale* (PANSS Kay et al., 1987) assessing positive, negative and general psychopathology symptoms in separate sub-scales.

## **BIOLOGICAL MOTION TEST**

An array of 50 randomly moving white dots appeared in a square area subtending 10◦ of visual angle on a black background. Each trial lasted 3500 ms. At a random time after trial onset, 12 of the dots moved as a biological motion array forming a figure walking at 4.5◦/s either leftward or rightward. The figure could appear at any position in the screen at the onset of the trial. When the figure reached the left or right vertical edge of the square array of dots it re-appeared at the opposite edge. Subjects were not required to maintain fixation. At the end of the trial, participants were asked to respond whether the figure had moved (i) leftward, (ii) rightward or (iii) was not seen. Responses were logged by the experimenter. A correct response increased the number of randomly moving dots by 20 for the next trial (increment = 10 after first incorrect response). An incorrect or "not seen" response resulted in a decrease of 10 dots for the next trial. This resulted in a psychophysical staircase function that reached a plateau after approximately 15 trials in each block. Two blocks of forty-two trials were performed for each subject. Each block contained 21 leftward trials and 21 rightward trails in pseudorandom order (i.e., the number of trials for each direction was fixed but their order of presentation randomised) so that each trial (i.e., each point on the psychophysical staircase) had equal probability of being leftwards or rightwards. All subjects were trained on the task prior to testing and confirmed they were able to see the walking figure.

#### **ANALYSIS**

The trials were sorted into leftward and rightward directions (42 trials for each direction in the two blocks combined) and an accuracy score for each direction derived for each subject. The leftward accuracy score = (number of correctly identified leftward trials / 42) × 100. The rightward accuracy score = (number of correctly identified rightward trials / 42) × 100. Trials with "did not see" responses were deemed incorrect. We also derived an accuracy score for the subset of trials at the plateau of the psychophysical staircase where the level of distractor dots was approximately constant. This threshold accuracy value related to the last 28 trials in each block and, because of the randomization of direction, varied from subject to subject in the total number of leftward and rightward trials. For each direction, group differences in accuracy score between patients and controls were tested using two-sample t tests. ANOVA models were used to examine the effects of gender and degree of right handedness. Within-subject measures of leftward and rightward accuracy were compared in a repeated measures ANOVA model with within-subject factor direction(left, right) and between-subject factor group(patient, control). Correlations between accuracy and functional outcome were explored using non-parametric tests (Spearman's Rho, one-tailed tests) and parametric tests (Pearson's) when co-varying for negative symptoms. Correlations between leftward and rightward accuracy and with IQ were measured using parametric tests (Pearson's).

## **FMRI METHODS**

Six control subjects without history of neurological or psychiatric illness took part in the study (two male, four female; mean age 30 ± 6 years). All had normal corrected visual acuity and gave informed consent. Subjects were presented the same translational biological motion stimulus as used in the psychophysical study, with timings adapted for fMRI (8 s trial length with the stimulus appearing at a random time around 4 s after the appearance of distractor dot noise; inter-trial interval = 8 s) and a fixed number of distractors determined for each subject prior to the scan to standardize performance at ∼70% correct. Subjects were not required to maintain fixation. Biological motion trials were interleaved with trials of coherent motion, optic flow and blank trials. Only data from biological motion trials is presented here (14 trials for each subject, seven leftward, seven rightward). Subjects responded with a right hand button press to indicate whether they had seen rightward, leftward or no motion.

#### **MRI ACQUISITION AND ANALYSIS**

Functional images were acquired on a 1.5 Tesla GE Neurooptimised Signa LX Horizon System (General Electric, Milwaukee, WI, USA), using a gradient echo planar sequence sensitive to blood oxygenation level dependent (BOLD) contrast (TR = 2 s; TE = 40 ms; flip angle 90◦; 64 × 64 matrix; in-plane voxel size 3.75mm × 3.75 mm). 16 axial slices, 7 mm thick with 0.7 mm interslice gap, were acquired every 2 s. For each subject, the functional time series was motion corrected (Friston et al., 1996), transformed into stereotactic space and smoothed with a 7 mm FWHM Gaussian filter using SPM software (http://www.fil.ion.ucl.ac.uk/spm). The activity at each voxel was high-pass filtered and modeled by three covariates (distractor dot onset, leftward biological motion onset for correct trials; rightward biological motion onset for correct trials), convolved with the hemodynamic response function. Group activation maps for leftward > rightward and rightward > leftward translational biological motion were created using a fixed effect model that included all subjects.

# **RESULTS**

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 3 — #3

As described previously for the biological motion task, control subjects had a higher number of distractor dots at threshold than patients (number of distractor dots at threshold controls = 211.95 ± 63; patients = 186.25 ± 61; Brittain et al., 2010a). The level of distractor dots at threshold reflects the number of errors made in the staircase and one would expect, therefore, a significant correlation between accuracy and the number of distractor dots at threshold (*r* = 0.845, *p* < 0.001 for the group as a whole, the correlation is not perfect due to the varying position on the staircase of the first error and its associated change in step size). The issue we explore here is whether accuracy across the staircase differed for one direction and the other or, put another way, whether the reduced sensitivity in schizophrenia overall was driven primarily by reduced sensitivity in a single direction.

The accuracy results from each direction in all participants are illustrated in **Figure 1A**. There was no significant difference in accuracy between patients and controls for rightward motion (patients = 62.7% ± 12.4[SD]; controls = 64.1% ± 11.7;

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 4 — #4

*t*<sup>126</sup> = 0.66; *p* > 0.05). In contrast, a significant difference was found for leftward motion (patients = 57.9% ± 14.3; controls = 63.6% ± 11.3; *t*<sup>126</sup> = 2.49; *p* = 0.01). The reduction in leftward motion accuracy was not influenced by gender (*F*1,124 = 0.75; *p* > 0.05) or right/ambiguous handedness (*F*1,123 = 1.16; *p* > 0.05). It does not reflect a bias in the patient group to respond "right" for trials that they were unsure of rather than using the "did not see" option as the average number of trials reported as "did not see" was similar in the two groups (patients = 28.9%, controls = 28.4%). These accuracy values relate to all trials in the staircase and thus contain a mixture of easy trials presented at the beginning of a block when the number of distractor dots is low and difficult trials presented at the end of a block when the number of distractor dots is high. We found the same pattern of results when examining the subset of difficult trials at threshold, although the level of significance was lower given the smaller number of trials and variability in the number of leftward and rightward trials used to derive the accuracy value (rightward motion at threshold patients = 56.2% ± 13.6; controls = 56.9% ± 14.1; *t*<sup>126</sup> = 0.297; *p* = 0.76; leftward motion at threshold patients = 49.9% ± 16.7; controls = 55.1% ± 14.2; *t*<sup>126</sup> = 1.86; *p* = 0.06). We also examined whether the accuracy values might be related to IQ. Treating the patients and controls as a single group, IQ correlated with leftward accuracy (*r* = 0.186; *p* = 0.01; higher IQ better accuracy score), but not rightward accuracy (*r* = −0.81; *p* = 0.18). The association with leftward accuracy was also found for the control group considered alone (*r* = 0.206; *p* = 0.05) and a negative correlation was found between rightward accuracy and IQ (higher IQ lower accuracy, *r* = −0.244; *p* = 0.02). In the patient group neither leftward accuracy (*r* = 0.113; *p* = 0.18) or rightward accuracy (*r* = 0.033; *p* = 0.39) correlated with IQ.

The reduction in leftward accuracy observed in the patient group is more complex than implied by the average data. There was a significant *negative* correlation between leftward and rightward accuracy for both patients and controls (controls*r* = −0.444, *p* < 0.001; patients *r* = −0.468, *p* < 0.001) such that, for most subjects, accuracy for one direction was better than the other. This relationship is illustrated in **Figure 1B** where a line is drawn for each subject connecting their leftward and rightward accuracy scores. While some subjects have approximately equivalent accuracy for leftward and rightward directions (horizontal lines), the majority have an asymmetrical bias (diagonal line). We derived an index of directional bias for each subject to further explore this issue (rightward accuracy – leftward accuracy; > 0 = rightward bias, < 0 = leftward bias; 0 = no bias). In controls, mean leftward bias was 17% ±11 and mean rightward bias was 18% ±11 with the magnitude of leftward and rightward bias balancing out such that, overall, mean accuracy was equivalent for leftward and rightward directions. In the schizophrenia group the distribution of bias was such that the rightward bias outweighs the leftward bias (mean leftward bias 18% ± 14; mean rightward bias 21% ± 12), with a consequent overall reduction in mean leftward accuracy. This effect is hidden in the within-subject ANOVA (group by direction interaction *F*1,126 = 1.285, *p* = 0.26) due to the high variance of accuracy difference across subjects with leftward and rightward direction bias. For the same reason, the within-subject *t*-test of leftward v rightward accuracy in the patient group is only at trend significance (patients mean difference 4.7% ± 23, *t*<sup>63</sup> = 1.66, *p* = 0.10; controls mean difference 0.48% ± 20, *t*<sup>63</sup> = 0.197, *p* = 0.84).

We next examined whether the directional bias index was linked to functional outcome. For the patient group as a whole there was a trend significant association for the functional outcome total score (rho = −0.175, *p* = 0.083) but significant and trend significant correlations with subscales of working productivity (rho = −0.208, *p* = 0.049), independent living/self-care (rho = −0.194, *p* = 0.063) and immediate social network relationships (rho = −0.173, *p* = 0.086). PANSS negative symptoms were strongly associated with functional outcome (rho = −0.577, *p* < 0.001) and controlling for negative symptoms, the relationship between directional bias index and functional outcome was strengthened (total score *r* = −0.272, *p* = 0.016; working productivity *r* = −0.225, *p* = 0.038; independent living/self-care *r* = −0.212, *p* = 0.047 and immediate social network relationships *r* = −0.235, *p* = 0.032).

#### **FMRI RESULTS**

Of the six subjects taking part in the fMRI study, four had leftward bias on the asymmetry index (32 ± 14%) and two had rightward bias (43 ± 0%). Pooling both sets of subjects we identified regions activated more for leftward than rightward translation of biological motion and vice versa. Given the exploratory nature of the study and small number of subjects and trials, a lenient threshold of *p* < 0.05 uncorrected and 10 contiguous voxels was used. **Figure 2** shows regions activated at this threshold by leftward motion (blue bars) more than rightward motion (red bars; **Figure 2A**) or rightward motion more than leftward motion (**Figure 2B**). Areas preferentially activated by leftward motion included bilateral regions of dorsolateral prefrontal cortex (MNI co-ordinates ± 56 30 28) bilateral regions in the intra-parietal sulcus (MNI co-ordinates ± 32 -68 40) and right cuneus (MNI co-ordinates 14 -80 40). Areas preferentially activated by rightward motion included bilateral regions in the supramarginal gyrus (MNI co-ordinates ± 52 -62 22), left STS/middle temporal gyrus (MNI co-ordinates -54 -32 - 14) and bilateral medial frontal regions (MNI co-ordinates ± 4 60 28). The pattern of preferential leftward and rightward activation was the same when the analysis was restricted to the four subjects with leftward bias alone. The number of subjects with rightward bias was too small to draw any conclusions as to whether regions of preferential activation differed in this subgroup.

# **DISCUSSION**

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 5 — #5

We have sought evidence for an asymmetric sensitivity to direction of biological motion translation in patients with schizophrenia based on an asymmetry of atrophy within regions linked to biological motion found in previous studies. Although we found evidence in support of an asymmetry, the findings suggest a more complex relationship between direction and hemisphere than envisaged. Below we discuss the findings in the light of preliminary functional imaging evidence and explore their wider implications.

#### **LEFTWARD AND RIGHTWARD DIRECTION DISCRIMINATION**

Although motion speed, coherence, local/global features and direction have been studied extensively in schizophrenia (seeChen, 2011 for review), as far as we are aware no studies have reported thresholds for leftward and rightward motion separately. Where direction discrimination has been investigated in previous studies (e.g., Chen et al., 2003; Slaghuis et al., 2005; Brittain et al., 2010a) the methods used measure overall performance on a leftward/rightward discrimination task rather than thresholds for leftward and rightward directions separately. The apparent deficit in leftward motion reported here is therefore an entirely novel finding. We were unable to explore whether it is also apparent in the global coherent motion task reported in our previous studies (Brittain et al., 2010a,b) as the coherent motion task involved upward and downward, not leftward and rightward directions.

The interpretation of the leftward direction deficit found in schizophrenia is more complex than anticipated. Unexpectedly, sensitivity for leftward and rightward directions were negatively correlated in the group as a whole, with the deficit in schizophrenia reflecting a difference in relationship between leftward and rightward accuracy rather than a deficit of sensitivity for leftward translation alone. As far as we are aware, a negative correlation of leftward and rightward direction sensitivity has not been reported before. It is important to note that our analysis is retrospective, based on previously collected data, and uses a non-conventional analysis of a standard psychophysical staircase. The analysis potentially introduces systematic biases as, for the ideal observer at the staircase plateau that defines threshold, successive trials may be seen and not seen in alternation because the number of distractor dots alternately increases and decreases. If the randomization of directions in the plateau allocates alternating leftwards and rightwards trials, one direction would be seen and the other not. However, this chance occurrence would not favour one direction over the other, i.e., could equally be "left seen, right not seen" as "right seen, left not seen". Any small bias in one direction for a block of trials would even out across repeated blocks and through combining data from different subjects. Furthermore, longer sequences of alternating directions lead to higher % accuracy for one direction but have no effect on % accuracy for the other so that such biases do not account for the negative correlation between directions found. It therefore seems unlikely our results can be explained by the non-conventional nature of our

"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 6 — #6

analysis. Indeed, the use of a single staircase to compare accuracy for the two directions would be more likely to introduce a spurious positive correlation between directions than the negative correlation found. The negative correlation of direction accuracy is also not explained by systematic differences in reporting in the two groups. IQ in the patient group was lower than in the control group and thus one might argue the patient group had greater difficulty understanding the task. However, while this might account for an overall reduction in accuracy, it seems unlikely that it could account for a deficit in one direction of translation but not the other. Similarly, one might argue that the patient group had medication and psychopathology not present in the control group that could influence performance in the task but it seems improbable that such effects could impact on one direction only. Sensitivity to biological motion is influenced by a number of factors including size and eccentricity (Gurnsey et al., 2008), body part (Takahashi et al., 2011), facing direction for stimuli presented in a given hemifield (de Lussanet et al., 2008) and executive control (Chandrasekran et al., 2010). However, apart from facing direction, these factors were identical for leftward and rightward translation so it is difficult to account for differences in sensitivity for the two directions in terms of these factors. The leftward translating stimulus was presented as if facing left and the rightward translating stimulus presented as if facing right; however, the trajectory of the walking figure crossed both left and right hemifields so that differential sensitivity to facing direction in one hemifield would be offset by the opposite sensitivity in the other hemifield. In support of this view, we did not find differential activation in our study within sub-regions of the fusiform gyrus sensitive to facing direction (Michels et al., 2009). In summary, although the underlying mechanism of directional bias and leftward accuracy deficit in schizophrenia requires further investigation, it does not seem to be accounted for by non-specific differences between the patient group and controls or by known factors influencing sensitivity to biological motion.

Our fMRI analysis was exploratory and used a lenient threshold in which many of the regions identified would not survive correction for multiple comparisons. However, it provides clues as to the types of functional or structural change in schizophrenia that could underlie the psychophysical findings. Importantly the fMRI findings suggest that our hypothesis of preferential representation for leftward motion in the left hemisphere and rightward motion in the right hemisphere derived from the coherent motion literature is over simplistic for biological motion. The areas identified in this study are predominantly bilateral so it is unlikely that any differences in structure or function in patients with schizophrenia restricted to one hemisphere would cause the shift in bias and consequent decrease in leftward accuracy found in the psychophysical data. What is more likely is that, in schizophrenia, functional changes in bilateral subsets of regions,for example decreased activity in bilateral dorso-lateral pre-frontal cortex or increased activity in bilateral STG, is responsible for the psychophysical changes. The fMRI data also raises the intriguing possibility that the negative correlation of leftward and rightward accuracy described in the psychophysical data might be linked to the reciprocal relationship of leftward and rightward responses within brain areas.

#### **DIRECTIONAL BIAS AND FUNCTIONAL OUTCOME**

Why might functional outcome be linked to a bias in translational direction perception? It seems unlikely the small overall deficit in leftward direction (a decrease in accuracy of 5% in the patient group) would have specific effects on social function. What seems more probable is that directional bias is an indirect measure of wider cognitive functions including (i) theory of mind cognition, linked in previous studies to motion perception (Kelemen et al., 2005) or (ii) the comprehension of action movements (Bonda et al., 1996). The unexpected association of IQ with leftward accuracy in the cohort as a whole and with leftward and (negative) rightward accuracy in the control group lends support to this view. We assume an aspect of social cognition or wider

# **REFERENCES**


processing, social cognition and functional outcome in schizophrenia. *Psychiatry Res.* 178, 270–275. doi: 10.1016/j.psychres.2009.09.013


cognitive function, co-localized or localized in close proximity to regions underlying directional bias, are responsible for these associations. The correlation coefficient linking functional outcome to directional bias is higher than that between functional outcome and biological motion threshold although the difference is not statistically significant (*r* = 0.129 for threshold versus outcome, *r* = (−)0.272 for directional bias versus outcome, co-varying for negative symptoms in both tests, *p* = 0.3 in Z transform test). In contrast, both these associations with functional outcome are lower than that reported by Kim et al. (2005)for biological motion sensitivity (*r* = 0.7; difference *p* = < 0.01 Z transform test). However, the Kim et al. (2005) study used a measure of outcome weighted by age and education which might account for the higher correlation.

# **CONCLUSION**

The deficit in biological motion perception for leftward translation we have identified in patients with schizophrenia and its link to functional outcome remains unexplained. However, it points to direction of translation sensitivity as a potentially important area of investigation in schizophrenia. The results presented here suggest measures of leftward and rightward biological motion translation may help explore cortical function in key frontal, parietal and temporal regions serving social cognitive function and their interaction across hemispheres to better understand the neurobiology of cognitive change in schizophrenia.

# **ACKNOWLEDGMENT**

Dominic H. ffytche acknowledges the support of the Wellcome Trust.

M. (2008). Interaction of visual hemifield and body view in biological motion perception. *Eur. J. Neurosci.* 27, 514–522. doi: 10.1111/j.1460- 9568.2007.06009.x


"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 7 — #7

Stimulus magnification equates identification and discrimination of biological motion across the visual field. *Vision Res.* 48, 2827–2834. doi: 10.1016/j.visres.2008.09.016


cerebellum communicates with the right superior temporal sulcus. *Neuroimage* 59, 2824–2830. doi: 10.1016/j.neuroimage.2011. 08.039


"fpsyg-04-00436" — 2013/7/12 — 20:02 — page 8 — #8

Area V5 of the human brain: evidence from a combined study using positron emission tomography and magnetic resonance imaging. *Cereb. Cortex* 3, 79–94. doi: 10.1093/cercor/ 3.2.79

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 10 May 2013; accepted: 25 June 2013; published online: 16 July 2013. Citation: Hastings CNM, Brittain PJ and ffytche DH (2013) An asymmetry of translational biological motion perception in schizophrenia. Front. Psychol. 4:436. doi: 10.3389/fpsyg.2013.00436 This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Hastings, Brittain and ffytche. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Contribution of coherent motion to the perception of biological motion among persons with schizophrenia

# *Justine M.Y. Spencer 1, Allison B. Sekuler 1, Patrick J. Bennett <sup>1</sup> and Bruce K. Christensen2\**

<sup>1</sup> Department of Psychology, Neuroscience and Behaviour, McMaster University, Hamilton, ON, Canada

<sup>2</sup> Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, ON, Canada

#### *Edited by:*

Randolph Blake, Vanderbilt University, USA

#### *Reviewed by:*

Duje Tadin, University of Rochester, USA Michael Herzog, École Polytechnique Fédérale de Lausanne, Switzerland

#### *\*Correspondence:*

Bruce K. Christensen, Department of Psychiatry and Behavioural Neurosciences, McMaster University, 100 West 5th, Hamilton, ON L8N 3K7, Canada e-mail: bruce.christensen@ mcmaster.ca

People with schizophrenia (SCZ) are impaired in several domains of visual processing, including the discrimination and detection of biological motion. However, the mechanisms underlying SCZ-related biological motion processing deficits are unknown. Moreover, whether these impairments are specific to biological motion or represent a more widespread visual motion processing deficit is unclear. In the current study, three experiments were conducted to investigate the contribution of global coherent motion processing to biological motion perception among patients with SCZ. In Experiments 1 and 2, participants with SCZ (n = 33) and healthy controls (n = 33) were asked to discriminate the direction of motion from upright and inverted point-light walkers in the presence and absence of a noise mask. Additionally, participants discriminated the direction of nonbiological global coherent motion. In Experiment 3, participants discriminated the direction of motion from upright scrambled walkers (which contained only local motion information) and upright random position walkers (which contained only global form information). Consistent with previous research, results from Experiment 1 and 2 showed that people with SCZ exhibited deficits in the direction discrimination of point-light walkers; however, this impairment was accounted for by decreased performance in the coherent motion control task. Furthermore, results from Experiment 3 demonstrated similar performance in the discrimination of scrambled and random position point-light walkers.

**Keywords: biological motion, schizophrenia, paranoid, motion, global mechanisms, local mechanisms, perception**

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 1 — #1

# **INTRODUCTION**

Since Johansson (1973) first introduced point-light walkers as an experimental tool for examining the perception of human motion, many studies have demonstrated the sensitivity of the human visual system with respect to detecting and perceiving biological motion. For example, healthy individuals are able to recognize human actions from point-light stimuli following short presentation durations (e.g., 200 ms; Johansson, 1973), under conditions of additional dynamic noise (Bertenthal and Pinto, 1994), and when the number of illuminated joint markers has been substantially reduced (Johansson, 1973). Furthermore, infants as young as 3 months old are able to discriminate upright point-light walkers from inverted walkers (Bertenthal et al., 1984). Dynamic biological motion has also been shown to convey relevant social information, such as cues regarding a person's affect (Atkinson et al., 2004), sex (Barclay et al., 1978; Pollick et al., 2005), intention to deceive (Runeson and Frykholm, 1983), and identity (Cutting and Kozlowski, 1977; Loula et al., 2005). As a result, biological motion can convey information not only about the perceptual characteristics (e.g., size and shape) of a walker, but also higher-order social information regarding a walker's intentions and emotional states.

Individuals with schizophrenia (SCZ) exhibit deficits in several aspects of visual motion processing including speed discrimination (Chen et al., 1999; Clementz et al., 2007) and the perception of coherent global motion (Stuve et al., 1997; Chen et al., 2003). Furthermore, this population exhibits deficits in their ability to recognize and interpret social stimuli (Bigelow et al., 2006; Baas et al., 2008) or detect emotions from affective facial expressions (Edwards et al., 2001; Kohler et al., 2003; Johnston et al., 2006; Monkul et al., 2007). Given known deficits in both motion processing and social cognition, it has been suggested that people with SCZ also may be impaired on tasks of biological motion processing, and that such deficits may contribute to the abovenoted social deficits. Indeed, several studies that have examined biological motion perception in SCZ suggest that these patients are impaired on biological motion processing tasks. Kim et al. (2005) demonstrated that people with SCZ show a deficit in recognizing biological motion activities compared to biological scrambled motion sequences. Moreover, people with SCZ performed similarly when completing a static global form detection task, suggesting that the observed group difference in the biological motion task was not due to a general performance deficit in SCZ patients. Additionally, Kim et al. (2011) found that people with SCZ were less able to detect and discriminate biological motion on tasks that included a noise mask and the perturbation of kinematic information, respectively.

Despite evidence suggesting that SCZ patients have greater difficulty detecting and discriminating biological motion, it is unclear whether these deficits are specific to biological motion *per se* or represent a more general deficit in perceiving global motion. Chen et al. (2003) measured direction discrimination thresholds using a sine wave grating, a task that requires only local motion processing, and random dot kinematograms, which requires global motion processing. Chen et al. (2003)found that SCZ patients had elevated direction discrimination thresholds only in the task that used random dot stimuli, which suggests that processing of global, but not local, motion is impaired in SCZ patients. Point-light walker stimuli contain both local and global cues: the trajectory of each dot constituting a point-light walker conveys information about the motion of a particular part of the human figure (e.g., the feet, the elbows, etc.), and grouping these local elements creates a holistic perception of the walker's global form (e.g., a whole body). Moreover, previous research suggests that both local (Mather et al., 1992; Troje and Westhoff, 2006) and global processes (Bertenthal and Pinto, 1994; Beintema and Lappe, 2002; Pilz et al., 2010) contribute to the perception of point-light walkers. Hence, the results of Chen et al. (2003) raise the possibility that at least some of the SCZ-related deficit in biological motion tasks reflects a general deficit in global motion processing. According to this hypothesis, the effect of SCZ on the perception of biological motion should be diminished or eliminated once the general effect of SCZ on global motion processing is taken into account. The current experiments examine this hypothesis. To our knowledge, previous studies investigating biological motion processing in SCZ have not used control tasks that could estimate deficits in global, non-biological motion processing. For example, Kim et al. (2005) used a control task that measured the ability of participants to group stationary lines into a larger global form. Although this task considered the grouping of visual elements into a Gestalt, the use of static stimuli means that it does not provide an appropriate control for examining global motion deficits.

To investigate the contribution of global motion to biological motion perception, three experiments were completed in which participants were asked to discriminate the direction of motion from point-light walkers. In Experiment 1, we measured direction discrimination thresholds for upright and inverted point-light walkers embedded in a dynamic noise mask. Importantly, direction discrimination thresholds also were measured for non-biological global motion, consisting of coherently translating dots, embedded in a dynamic noise mask. To determine if the results of obtained in Experiment 1 generalize to supra-threshold conditions, Experiment 2 measured response accuracy in a direction discrimination task that used stimuli similar to those used in Experiment 1 but which did not contain dynamic noise. Finally, Experiment 3 investigated the contribution of local and global mechanisms to biological motion among participants with SCZ by using point-light walkers that contained only local motion information (scrambled point-light walkers) or global motion information (random position point-light walkers).

#### **GENERAL MATERIALS AND METHODS PARTICIPANTS**

Thirty-three people with SCZ (5 female, 28 males) and 33 healthy controls (18 females, 15 males) participated in all experiments. All participants had normal or corrected-to-normal visual acuity, as ascertained using a standard Snellen chart, and reported an absence of lifetime neurological illness, brain injury, learning disability, current or past substance dependence, or medical conditions that could affect cognitive performance (e.g., coronary heart disease, type 1 diabetes).

All patients met criteria for SCZ (21 patients) or schizoaffective disorder (12 patients), as confirmed by the Mini International Neuropsychiatric Interview (M.I.N.I.; Sheehan et al., 1998), but did not meet criteria for any other Axis 1 disorder. Patients with SCZ were outpatients, medication-stable for at least the past 6 months, and were prescribed either typical (5 patients) or atypical antipsychotics (28 patients). Healthy control participants did not meet criteria for any Axis 1 disorder and were also excluded if they reported having a first-degree relative with a SCZspectrum illness. Estimates of Full Scale Intelligence Quotient (FSIQ) were obtained by prorating performance on the Matrix Reasoning and Information subtests from the Wechsler Adult Intelligence Scale, 3rd Edition (Wechsler, 1997). General cognitive functioning was assessed via the Repeatable Battery for the Assessment of Neuropsychological Status (Randolph, 1998). Patients with SCZ were also administered the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987) to assess current symptom status. In addition, both patients and controls were administered select scales from the Personality Assessment Inventory (PAI), including the Depression (Dep), Alcohol Problems (Alc), Drug Problems (Drg), Positive Impression Management (PIM), and Negative Impression Management (NIM) scales (Morey, 1991). Both groups were age-matched, but healthy controls had achieved significantly higher years of education. Regarding neuropsychological measures, significant differences were found across WAIS-III FSIQ and all RBANS indices. Although participants with SCZ also had significantly elevated scores on the PAI-Drg and PAI-Dep subscales compared to healthy controls, no single participant scored in a range suggesting significant clinical problems in these domains (i.e., *T* > 70). Moreover, no participant evidenced deliberate distortion of their responses across both validity scales (i.e., PIM and NIM). **Table 1** provides information characterizing the study participants. Ethics approval for the study was obtained by the St. Joseph's Healthcare Hamilton Research Ethics Board. All participants provided written, voluntary consent to participate and received \$10/hour for their participation. Each participant was tested in all three experiments. The order of the experiments was counterbalanced across participants.

#### **APPARATUS AND STIMULI**

All experimental tasks were programmed and presented using a Macbook Pro laptop computer with MATLAB and the Psychophysics and Video ToolBox extensions (Brainard, 1997; Pelli, 1997). Stimuli were presented on a 19-in monitor with a resolution of 1024 × 864 pixels and a refresh rate of 60 Hz.

# **PROCEDURE**

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 2 — #2

In each experiment, participants were seated in a darkened room and viewed the stimuli at a distance of 60 cm with their heads stabilized by a chin rest. On each trial, the direction of motion was either toward the left or the right, and participants were asked to identify the direction of motion displayed by the stimulus by pressing a key on a standard QWERTY computer keyboard [i.e., "a" key (left) and"l"key (right)]. Stimulus durations were intermixed randomly across trials. Prior to the experiments, participants performed 10



\*Indicates a significant difference between healthy controls and people with SCZ. FSIQ, Full Scale Intelligence Quotient; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; PAI, Personality Assessment Inventory; PANSS, Positive and Negative Syndrome Scale.

practice trials for each stimulus type to familiarize themselves with the stimuli.

#### *Experiment 1*

*Methods.* Experiment 1 used upright and inverted point-light walkers, in addition to the non-biological motion control task. The stimulus used in the control task consisted of 11 dots that moved coherently to the left or right at a speed of 7◦/s. Pointlight walker stimuli were generated using a modified version of Cutting's classic point-light walker algorithm (Cutting, 1978; Thornton et al., 1998). The walkers consisted of 11 dots (2 × 2 pixels) that simulated points on the head, shoulder, elbows, wrists, hip, knees, and ankles. The starting position of the stride cycle was chosen randomly on every trial, which prevented participants from recognizing the walker simply from the starting point or from a specific frame. The walker, which subtended 1.9 × 4.2◦, did not move across the screen, but rather appeared to walk in place, as if on a treadmill. Inverted walkers were rotated by 180◦ so that they appeared to be walking on the ceiling. The walker stimuli consisted of 5, 15, 30, or 45 frames presented at 25 frames per second, resulting in a total presentation times of 0.2, 0.6, 1.2, and 1.8 s, respectively. These specific durations were chosen based on previous research suggesting that durations shorter than 0.2 s are insufficient for direction discrimination whereas increasing duration beyond 1.8 s does not improve performance (Pilz et al., 2010). One complete gait cycle was achieved after 40 frames, or 1.6 s.

All stimuli were occluded with a dynamic noise mask composed of an array of dots whose positions varied randomly on each stimulus frame. The point-light walkers, the dynamic noise mask, and the control stimulus, were all constructed with dots that were identical in size and contrast. All stimuli were presented on a black background and were centered on the middle of the screen. Dot luminance was 67.4 cd/m<sup>2</sup> and the background luminance was less than 1 cd/m2. Trials in the conditions using upright walkers, inverted walkers, and drifting dots were randomly intermixed.

Direction discrimination thresholds were estimated by varying the number of dots presented in the dynamic noise mask using a 3-up/1-down staircase procedure. Note that the staircase increased the number of mask dots (i.e., reduced the signal-to-noise ratio) after three consecutive correct responses, and decreased the number of mask dots (i.e., increased the signal-to-noise ratio) after one incorrect response. The staircase converged on the number of mask dots needed to produce 79% correct responses, which in this task corresponds to a *d'* of 1.14. Thresholds for each stimulus duration were estimated by averaging the last 10 reversals.

*Results.* Direction discrimination thresholds are shown in **Figure 1**. Because thresholds are expressed in terms of the number

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 3 — #3

**FIGURE 1 | Direction discrimination noise threshold for SCZ and healthy controls in the three conditions and across all durations.** Threshold is expressed as the number of dots in a dynamic noise mask; the point-light walker consisted of 11 dots. Both groups showed better performance in

the dots condition compared to upright and inverted point-light walkers. Both groups also showed better performance in the upright condition compared to the inverted condition. No interactions were found.

of mask dots that are required to produce 79% correct responses, higher values correspond to better performance.

An initial 2 (group)×3 (stimulus type)×4 (stimulus duration) Analysis of Variance (ANOVA) was conducted on resulting thresholds. The ANOVA revealed a significant main effect of group (*F*(1, 64) = 4.89, *p* = 0.031), where participants with SCZ required fewer noise dots to reach threshold compared to healthy controls across all walker types and stimulus durations. A significant main effect of stimulus type was also observed (*F*(2, 128) = 34.85, *p* < 0.001) where both groups of participants demonstrated better performance in the coherent motion task compared to the upright (*t*(263) = 9.17, *p* < 0.001) and inverted point-light walker (*t*(263) = 11.42, *p* < 0.001) conditions. Bonferroni adjusted paired *t*-tests also revealed that participants performed significantly more accurately in the upright condition compared to the inverted condition (*t*(263) = 3.41, *p* < 0.001). A significant main effect of stimulus duration was also revealed: participants required fewer noise dots to reach threshold at 0.2 s duration, followed by 0.8, 1.2, and 1.6 s duration, respectively. All Bonferroni adjusted *t*-tests were significant (*p* < 0.05). No significant interactions were observed.

Results from the initial analysis suggest that people with SCZ have deficits in discriminating the direction of both biological motion and coherent non-biological motion. Importantly, the group × stimulus type interaction was not significant (*F*(2, 128) = 0.79, *p* = 0.45), which suggests that the difference between groups did not depend on the type of stimulus. Hence, the SCZ-related deficit in discriminating the direction of biological motion (i.e., point-light walkers) was no bigger than the deficit in discriminating non-biological motion (i.e., drifting dots). We therefore examined whether group differences in the upright and inverted walker conditions could be accounted for by performance in the control condition. First, we confirmed that the group difference in the upright and inverted walker conditions was significant: a 2 (group) × 2 (walker orientation) × 4 (stimulus duration) condition revealed significant main effects of group (*F*(1, 64) = 5.42, *p* = 0.023), walker orientation (*F*(1, 64) = 26.17, *p* < 0.001), and duration (*F*(3, 192) = 40.28, *p* < 0.001). The group × walker orientation interaction (*F*(1, 64) = 1.71, *p* = 0.19) was not significant, nor were any of the other interactions (in each case, *F* < 1.19, *p* > 0.31). To investigate whether group differences in biological motion discrimination can be accounted for by differences in global coherent motion discrimination, thresholds in the upright and inverted walker conditions were submitted to a 2 (group) × 2 (walker orientation) × 2 (stimulus duration) Analysis of Covariance (ANCOVA), where discrimination threshold in the control condition served as the covariate. In this analysis, the covariate was calculated by averaging thresholds in the control condition across stimulus duration, because a preliminary ANOVA on thresholds in the control condition failed to reveal a reliable group × stimulus duration interaction, (*F*(3, 192) = 2.10, *p* = 0.100), suggesting that group differences in the control condition did not vary as a function of stimulus duration. The ANCOVA revealed a significant effect of the covariate (*F*(1,63) = 34.68, *p* < 0.001), and significant main effects of walker orientation (*F*(1, 63) = 25.91, *p* < 0.001) and stimulus duration (*F*(3, 192) = 23.41, *p* < 0.001). Importantly, the main effect of group was not significant (*F*(1, 63)=1.92, *p* = 0.171), nor were any of the interactions. These results suggest thresholds in the upright and inverted walker conditions do not differ between groups, once threshold in the control condition is taken into account.

#### *Experiment 2*

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 4 — #4

Consistent with previous reports (Kim et al., 2005, 2011), Experiment 1 found that people with SCZ exhibit deficits in the direction discrimination of biological motion. However, Experiment 1 also found that the group difference was not significant once threshold in a control task, which used non-biological global motion, was taken into account. This result suggests that deficits observed in the direction discrimination of point-light walkers among SCZ participants were not specific to biological motion but instead may represent more a more widespread processing deficit for global visual motion.

Experiment 1 measured direction discrimination thresholds, which means that our stimuli were presented at low signal-tonoise ratios. However, biological motion stimuli are presented at very high signal-to-noise ratios in many naturalistic contexts. Do SCZ patients exhibit biological motion processing deficits in such situations? Previous studies suggest that such deficits do exist: Kim et al. (2005), for example, used high signal-to-noise stimuli in a biological motion detection task and found that sensitivity in SCZ patients (*d'* = 2.2) was significantly lower than sensitivity in control participants (*d'* = 2.8). Also, Spencer et al. (2013) found that response accuracy in an emotion discrimination task using unmasked, high signal-to-noise ratio point-light walkers was significantly lower in SCZ patients than control subjects. However, neither one of these studies examined whether the SCZ-related deficits could be accounted for by deficits in non-biological, global motion processing. Therefore, Experiment 2 examined this question by having participants perform the same tasks as Experiment 1, but without the dynamic noise mask.

*Methods.* Experiment 2 used identical stimuli to those presented in Experiment 1, but with the dynamic noise mask removed.

*Results.* The dependant variable, proportion of correct responses, was not normally distributed; therefore, statistical analyses were performed on the arcsine-transformed data. First, transformed data were submitted to a 2 (group) × 3 (stimulus type) × 4 (stimulus duration) ANOVA. Results of this analysis revealed a significant main effect of group (*F*(1, 65) = 4.47, *p* = 0.038) where people with SCZ demonstrated reduced accuracy across all conditions and stimulus durations (**Figure 2**). A significant main effect of stimulus type was also observed (*F*(2, 130) = 9.19, *p* < 0.001) where both groups of participants performed more accurately in the coherent motion task compared to the inverted point-light walker condition (*t*(267) = 2.23, *p* =0.026). Bonferroni corrected pairwise *t*-tests did not reveal significant differences between the other conditions. A main effect of stimulus duration was also revealed (*F*(3, 195) = 5.88, *p* < 0.001). Subsequent Bonferroni corrected *t*-tests revealed that significant differences were only found in the stimulus duration of 0.2 s (*p* < 0.05). Comparisons between 0.6, 1.2, and 1.8 s were not significantly different. The group × stimulus type interaction

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 5 — #5

(*F*(2, 130) = 0.8, *p* = 0.45) was not significant, nor were any of the other interactions (*F* < 1.54, *p* > 0.20 in each case).

As in Experiment 1, we next conducted analyses to determine if the group differences in the upright and inverted walker conditions could be accounted for by differences in the control condition. First, we confirmed that the group difference was significant in the two walker conditions: a 2 (group) × 2 (walker orientation) × 4 (stimulus duration) ANOVA revealed significant main effects of walker orientation (*F*(1, 65) = 4.51, *p* = 0.037) and stimulus duration (*F*(3, 195) = 3.93, *p* = 0.01), and a significant walker orientation × stimulus duration interaction (*F*(3, 195) = 2.72, *p* = 0.046). The main effect of group approached significance (*F*(1, 65) = 3.75, *p* = 0.057), although a one-tailed test, which is appropriate for the prediction that accuracy is lower in the SCZ group, was significant (*t*(65) = 1.94, *p* = 0.028). Next, we analyzed the results in the control condition using a 2 (group) × 4 (stimulus duration) ANOVA: the main effect of group was significant (*F*(1, 65) = 5.69, *p* = 0.02) but the group × duration interaction was not (*F*(3, 192) = 1.83, *p* = 0.143), suggesting that the group difference did not vary as a function of stimulus duration. We therefore averaged performance across stimulus durations and used the resulting value as a covariate in a 2 (group) × 2 (walker orientation) × 4 (stimulus duration) ANCOVA. The ANCOVA revealed a significant effect of the covariate (*F*(1, 64) = 495.07, *p* < 0.001), significant main effects of both walker orientation (*F*(1, 64) = 4.47, *p* < 0.038) and stimulus duration (*F*(3, 192) = 3.99, *p* < 0.009), and a significant walker orientation × stimulus duration interaction (*F*(3, 189) = 2.70, *p* = 0.046). However, the main effect of group (*F*(1, 64) = 0.075, *p* = 0.389) was not significant, nor were any of the remaining interactions (*F* < 1.87, *p* > 0.13 in all cases). These results suggest that healthy controls and patients with SCZ perform similarly in the direction discrimination of supra-threshold biological motion stimuli once differences in coherent global motion are taken into account.

#### *Experiment 3*

The results of Experiments 1 and 2 suggest that SCZ-related deficits in the discrimination of the direction of point-light walkers is not specific to biological motion, but instead can be accounted for by more general global coherent motion processing deficits. This result was the same across both experiments, indicating that SCZ-related deficits in discriminating biological motion can be accounted for by global coherent motion processing regardless of signal-to-noise conditions.

To further investigate the contribution of global motion processing to biological motion discrimination, Experiment 3 compared SCZ and control group direction discrimination of scrambled and random position point-light walkers. In the *scrambled* condition, the trajectory of each local dot was maintained, but the initial dot positions were shifted randomly along the *x* and *y*-axes of the display, resulting in a point-light walker with intact local motion information but distorted global form (e.g., Thornton et al., 1998; Troje and Westhoff, 2006; Pilz et al., 2010). In the *random position* condition, each dot was shifted randomly between two adjacent joints across successive frames (e.g., Beintema and Lappe, 2002; Pilz et al., 2010), resulting in disrupted local trajectories of individual dots but preserved global form of a walker. A recent study by Pilz et al. (2010) using similar stimuli showed that among both younger and older adults, the removal of global elements from point-light walkers (i.e., scrambled point-light walkers) resulted in significantly reduced direction discrimination. Conversely, the removal of local motion information (i.e., random position point-light walkers) had little impact on performance. Given the results from Experiments 1 and 2 suggesting that SCZ-related deficits in biological motion discrimination can be accounted for by general deficits in global coherent motion processing, it was hypothesized that SCZ patients would be negatively impacted by the removal of global form information but undeterred by the removal of information regarding local position.

*Methods.* In Experiment 3, healthy controls and patients with SCZ discriminated the direction of upright, scrambled, and random position point-light walkers. As in the previous experiments, stimuli were presented in four durations (0.2, 0.8, 1.2, and 1.6 s), which were randomized on every trial. Participants completed 20 trials for each type of walker at each stimulus duration, resulting

in a total of 240 trials. The dependent variable was the proportion of correct responses.

compared to the scrambled condition. No interactions were found.

*Results.* Response accuracy is plotted as a function of stimulus duration in **Figure 3**. A 2 (group) × 3 (stimulus type) × 4 (duration) ANOVA on arcsine-transformed data revealed a significant main effect of group (*F*(1,64) = 7.17, *p* = 0.009), where people with SCZ were less accurate overall compared to healthy controls. The ANOVA also found a significant main effect of stimulus type (*F*(2, 128) = 1002.03, *p* < 0.001), such that response accuracy in both groups was greater in the upright and random position conditions compared to the scrambled condition. No other significant main effects or interactions were observed.

Because accuracy did not vary with stimulus duration, we averaged accuracy across stimulus duration for each participant. The mean of the averaged accuracy is plotted as a function of group and stimulus type in **Figure 4**, which illustrates that accuracy in both groups of participants was quite high in the upright and random walker conditions and near chance in the scrambled walker condition. Hence, manipulation of local and global information had qualitatively similar effects in SCZ patients and healthy controls.

However, **Figure 4** also highlights the existence of ceiling effects in the upright and random walker conditions and a possible floor effect in the scrambled condition. These ceiling and floor effects would make it difficult for an ANOVA to find group differences in performance, and therefore we analyzed the data in another way. Following an approach previously suggested by Kim et al. (2011), first, we compared the proportion of participants in each group whose accuracy in each of the upright and random conditions was less than 1.0, and whose accuracy in the scrambled condition was significantly greater than chance (i.e., accuracy ≥0.587, *p* < 0.05, one-tailed). In each condition, the proportions of participants in the SCZ and control groups did not differ (see **Table 2**). Next, we used *t*-tests to compare the mean response accuracy for participants in each group who had an accuracy less than 1.0 in the upright and random walker conditions and above chance in the scrambled condition: for these subsets of participants, accuracies in the SCZ

group and the control group differed in the upright and random walker conditions, but not the scrambled walker condition (see **Table 3**).

Taken together, our analyses suggest that there was a small group difference in response accuracy in the upright and random conditions, but not in the scrambled condition. However, the primary finding was that manipulations of local motion information and global form had similar effects on direction discrimination in both groups.

#### **GENERAL DISCUSSION**

across all stimulus durations.

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 6 — #6

The present study examined the effects of SCZ on the perception of biological motion. Consistent with previous reports (Kim et al., 2005, 2011), Experiments 1 and 2 found that SCZ patients

**Table 2 | Proportion of participants in Experiment 3 with response accuracies that were less than 1.0 in the upright and random walker conditions and greater than chance in the scrambled walker condition.**


**Table 3 | Mean response accuracy of participants who had an accuracy less than 1.0 in the upright and random conditions and greater than chance in the scrambled condition.**


were worse than healthy control participants at discriminating the direction of upright and inverted point-light walkers. However, Experiments 1 and 2 also found that SCZ patients were worse in a control task that required participants to discriminate the direction of non-biological global motion (i.e., coherently drifting dots). Furthermore, we found that group differences in conditions that used point-light walkers were eliminated once performance in the control task was taken into account. Taken together, the results of Experiments 1 and 2 suggest that although people with SCZ do exhibit deficits in the direction discrimination of pointlight walkers, this impairment is not specific to biological motion *per se* but likely represents more general deficits in global motion processing.

Experiment 3 also found evidence that SCZ patients were slightly less accurate at discriminating the direction of standard upright point-light walkers as well as random walkers that contained the global form, but not the local motion cues. These results suggest that patients with SCZ exhibit deficits utilizing global form compared to healthy controls. Although no significant group differences were found in the scrambled condition, it is more difficult to speculate regarding local motion mechanisms, as most participants in both groups were found to perform at chance level. Additionally, results of Experiment 3 demonstrate that manipulations of the global form and local motion information in point-light walkers had similar effects on direction discrimination in both patients with SCZ and control participants: in both groups, removing local motion cues (i.e., the random position condition) had small effects on performance, but removing global form cues (i.e., the scrambled condition) made discrimination much more difficult. This result suggests that the relative influence of global form and local motion cues on the perceived direction of point-light walkers is similar in people with SCZ and healthy controls. Despite these results, it is difficult to speak directly to the nature of local and global mechanisms in this experiment given the observed floor and ceiling effects (see **Figure 4**). The use of a noise mask to remove performance from the floor and ceiling would be helpful to examine specific contributions of local and global mechanisms to biological motion processing and represents an avenue of future study. Nevertheless, despite floor and ceiling effects, results from Experiment 3 demonstrate that people with SCZ performed similarly to that of healthy controls, in that form information is important for the direction discrimination of point-light walkers. Furthermore, these results are consistent with Pilz et al. (2010), in which removing global form from point-light walkers was shown to reduce performance in both younger and older healthy adults. As a result, using the identical mechanism to alter the point-light walker stimuli resulted in similar performances among healthy controls and people with SCZ, suggesting that the mechanisms used during this experiment were also likely similar.

It is also important to note that the stimuli used in the current study were generated using a modified version of Cutting's classic point-light walker algorithm. Although these stimuli have been used repeatedly in previous studies, more recent research by Saunders et al. (2009) has suggested that the Cutting point-light walker algorithm lacks important visual information associated with the local motion of dots representing the feet compared to more naturalistic point-light walkers displays. Specifically, Saunders et al. (2009) show that while detection performance of Cutting and naturalistic point-light walkers was unchanged, participants were better able to discriminate the direction of motion from scrambled naturalistic point-light walkers compared to Cutting point-light walkers. Given these results,future research should utilize naturalistic point-light walkers to further examine deficits in SCZ regarding biological motion perception.

Regarding additional limitations, all patients with SCZ who took part in the current study were medicated, and we are unable to comment as to whether the results observed in the study were confounded by medication status. Additionally, analysis of sample characteristics revealed that the patients in the study had a significantly lower education level, as well as estimated intelligence levels and general neuropsychological scores. The issue of how to approach confounding group differences, however, is a complicated one and a topic of active debate within SCZ research. On the one hand, several characteristics that reliably distinguish SCZ from healthy participants are indeed correlated with outcomes of interest. In this situation, it is conventional to attempt to equate between-group differences on the confounding variable via linear covariate analyses. The popularity of these methods notwithstanding, they have been criticized on both statistical and conceptual grounds. In the case of the latter, Meehl (1970) has argued that if the confounding variable is a valid reflection of a pathological state (e.g., psychological symptoms), linear removal of the shared variance will necessarily attenuate the between-group variance of interest. Statistically, as Miller and Chapman (2001) and others discuss, the use of ANCOVA to correct for factors such as IQ is statistically dubious, as this analysis assumes that the covariate and independent variable, such as diagnostic group, are independent (Silverstein, 2008). As such, in psychopathological research, these variables are often not independent, and using ANCOVA to control for a covariate in psychopathology research

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 7 — #7

removes meaningful variance from the independent variable of interest. Consequently, in the absence of random assignment, group membership is generally acknowledged to represent a broad collection of symptoms and problems that denote the entirety of a psychopathological category.

The perception of global coherent motion in random dot patterns requires the visual system to represent the speed and direction of individual dots and to integrate such information across space and time. Chen et al. (2003) presented evidence that processing of global, but not local, motion is impaired in SCZ, which suggests that spatiotemporal integration of local motion cues is deficient in SCZ patients. Given this apparent spatiotemporal motion integration deficit in patients with SCZ, and the fact that biological motion processing involves the integration of both local and global cues (Mather et al., 1992), it is not surprising that SCZ-related deficits in biological motion processing have been observed in previous studies (Kim et al., 2005, 2011). Deficits in global motion processing also are consistent with an fMRI (functional magnetic resonance imaging) study by Chen et al. (2008) which found that SCZ patients had reduced activation in middle temporal area (MT), a cortical area that has been implicated in global motion processing (Maunsell and Newsome, 1987), during tasks of coherent motion and speed discrimination, but not during a task of contrast discrimination. Interestingly, Chen et al. (2008) also found greater activation in prefrontal cortex during motion tasks in SCZ patients than control participants, suggesting that higher-order cognitive processes may be used by SCZ patients as a compensatory mechanism for motion processing deficits.

However, not all studies have found differential activation of MT in SCZ patients. Recently, Kim et al. (2011) reported that the overall pattern of brain activity associated with the processing of biological motion, but not activation in area MT, differs between healthy controls and patients with SCZ. Kim et al. (2011) interpreted their results as showing that deficits observed in biological motion among patients with SCZ were not solely attributable to motion processing more generally. One explanation for the lack of differential MT activity observed by Kim et al. (2011) is that behavioral differences in biological motion may instead involve mechanisms underlying the integration of spatial and temporal motion cues. People with SCZ exhibit deficits in spatial (Doniger et al., 2001, 2002) and temporal (Schwartz et al., 1983; Izawa and Yamamoto, 2002) integration. For example, compared to healthy controls, patients with SCZ are less able to spatially integrate

#### **REFERENCES**


Barclay, C., Cutting, J., and Kozlowski, L. (1978). Temporal and spatial factors in gait perception that influence gender recognition. *Percept. Psychophys.* 23, 145–152. doi: 10.3758/BF03208295

Barraclough, N. E., Xiao, D., Baker, C. I., Oram, M. W., and Perrett, D. I. (2005). Integration of visual and auditory information by superior temporal sulcus neurons responsive to the sight of actions. *J. Cogn. Neurosci.* 17, 377–391. doi: 10.1162/0898929 053279586

fragmented images into coherent objects (Doniger et al., 2001). Furthermore, using event-related potential recordings, the inability to integrate these fragments has been correlated positively with dorsal stream processing in people with SCZ (Doniger et al., 2002), which is consistent with a wealth of literature suggesting impaired dorsal stream function in this population (Selemon et al., 1995; Gur et al., 2000; King et al., 2008).

Many studies have also implicated cortical area STSp (posterior superior temporal sulcus), a component of the dorsal stream network, in the perception of biological motion (e.g., Bondra et al., 1996; Grossman and Blake, 2001, 2002; Puce and Perrett, 2003). However, there is some evidence that the activity of the STSp differs in SCZ patients. For example, Kim et al. (2011) used fMRI to show that activation in STSp was higher when viewing biological motion than non-biological motion in healthy controls but not SCZ patients. Given that the STSp is a component of the dorsal stream network and also involved in the integration of other sensory information (Barraclough et al., 2005) the lack of STSp activity in response to biological motion among patients with SCZ may reflect a general integration deficit regarding visual elements. Furthermore, Giese and Poggio (2003) have argued that STS plays an important role in integrating information from the ventral and dorsal pathways into a single, coherent percept of biological motion. Given proposed integration deficits in people with SCZ, the differential activation of STSp reported by Kim et al. (2011) may reflect a more general impairment in the integration of visual elements.

In summary, the current experiments suggest that differences between SCZ and healthy controls in the ability to discriminate the direction of point-light walkers can be accounted for by SCZ-related deficits in the ability to perceive the direction of non-biological motion, which may be caused by deficits in spatial and/or temporal integration. However, in addition to having a perceived direction of motion, point-light walkers also can convey complex social information such as affect, intention, and identity. It is entirely possible that SCZ patients have deficits in perceiving this social information that cannot be accounted for by differences in spatial and temporal integration of non-biological motion. Given well-documented impaired social cognition among persons with SCZ (Hooker and Park, 2002; Bigelow et al., 2006; Monkul et al., 2007; Baas et al., 2008), it is important examine the perception of biological motion in its relation to social cognition.


"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 8 — #8

*Psychol.* 37, 213–230. doi: 10.1016/ 0022-0965(84)90001-8


recognition of biological movements. *Nat. Rev. Neurosci.* 4, 179–192. doi: 10.1038/nrn1057


*Psychol.* 117, 799–811. doi: 10.1037/ a0013500


"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 9 — #9

of biological motion. *Philos. Trans. R. Soc. Lond. B Biol. Sci.* 358, 435–445. doi: 10.1098/rstb.2002. 1221


patients with schizophrenia and normal controls. *Psychol. Med.* 23, 143–152. doi: 10.1017/S0033291796 004230


821–824. doi: 10.1016/j.cub.2006. 03.022

Wechsler, D. (1997). *Wechsler Adult Intelligence Scale*, 3rd Edn. San Antonio: The Psychological Corporation.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; accepted: 17 July 2013; published online: 13 August 2013.*

*Citation: Spencer JMY, Sekuler AB, Bennett PJ and Christensen BK (2013) Contribution of coherent motion to the perception of biological motion among persons with schizophrenia. Front. Psychol. 4:507. doi: 10.3389/fpsyg.2013. 00507*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

"fpsyg-04-00507" — 2013/8/11 — 13:07 — page 10 — #10

*Copyright: © 2013 Spencer, Sekuler, Bennett and Christensen. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Imagined motor action and eye movements in schizophrenia

# *Céline Delerue\* and Muriel Boucart*

*Laboratoire de Neurosciences Fonctionnelles et Pathologies, Centre National de la Recherche Scientifique, Hôpital Roger Salengro, Université Lille – Nord de France, Lille, France*

#### *Edited by:*

*Anne Giersch, Institut National de la Santé et de la Recherche Médicale, France*

#### *Reviewed by:*

*Wolfgang Tschacher, Universität Bern, Switzerland Rebekka Lencer, University of Muenster, Germany*

#### *\*Correspondence:*

*Céline Delerue, Laboratoire de Neurosciences Fonctionnelles et Pathologies, Centre National de la Recherche Scientifique, Hôpital Roger Salengro, Université Lille – Nord de France, CHRU de Lille, F-59037 Lille, France e-mail: celine.b.delerue@wanadoo.fr* Visual exploration and planning of actions are reported to be abnormal in schizophrenia. Most of the studies monitoring eye movements in patients with schizophrenia have been performed under free-viewing condition. The present study was designed to assess whether mentally performing an action modulates the visuomotor behavior in patients with schizophrenia and in healthy controls. Visual scan paths were monitored in eighteen patients with schizophrenia and in eighteen healthy controls. Participants performed two tasks in which they were asked either to (1) look at a scene on a computer screen (free viewing), or (2) picture themselves making a sandwich in front of a computer screen (active viewing). The scenes contained both task-relevant and task-irrelevant objects. Temporal and spatial characteristics of scan paths were compared for each group and each task. The results indicate that patients with schizophrenia exhibited longer fixation durations, and fewer fixations, than healthy controls in the free viewing condition. The patients' visual exploration improved in the active viewing condition. However, patients looked less at task-relevant objects and looked more at distractors than controls in the active viewing condition in which they were asked to picture themselves making a sandwich in moving their eyes to task-relevant objects on an image. These results are consistent with the literature on deficits in motor imagery in patients with schizophrenia and it extends the impairment to visual exploration in an action imagery task.

**Keywords: schizophrenia, eye movements, visual exploration, imagined motor action, attention**

# **INTRODUCTION**

Both behavioral (Ellis and Tucker, 2000; Tucker and Ellis, 2001, 2004) and brain imaging (Chao and Martin, 2000; Gerlach et al., 2002; Grèzes and Decety, 2002; Grèzes et al., 2003) studies have shown that objects, or photographs of objects, potentiate the appropriate action and that this process involves activation of motor representations and depends on the intention of the observer. Even in the absence of explicit intentions to act, attending to an object, or searching for an object, may activate motor representations appropriate to reaching, grasping, and manipulating the object (Tucker and Ellis, 1998; Bekkering and Neggers, 2002; Grèzes et al., 2003).

The aim of the present study was to investigate whether the mental imagery of an action sequence modulates the visuomotor behavior in patients with schizophrenia and in healthy control participants. Orientation of attention and eye movements have been shown to depend both on the characteristics of the stimulus (stimulus driven) and on the goals and intentions of the observer [goal driven; see Henderson (2003) for a review]. We assessed if the pattern of eye movements varies as a function of whether the observer mentally performs an action or not. Visual exploration of photographs of scenes was compared in two conditions: one in which participants were required to picture themselves interacting with several objects (active viewing) and one in which they were asked to just look at a scene (free viewing).

Motor imagery is a dynamic state in which an individual simulates mentally the performance of a specific motor action (Decety, 1996; Jeannerod, 1997). Behavioral studies in healthy participants show that real and imagined motor actions are subject to the same environmental and physiological constraints (Decety, 1996; Jeannerod, 1997; Maruff et al., 1999). Neuroimaging studies have reported common activation patterns in the mental simulation and the real motor actions (Decety et al., 1994, 1998). Moreover, motor imagery requires participants to generate an internal representation of intended but unexecuted motor actions and to anticipate the consequences of that action as if it had really been carried out (Jeannerod, 1997). Tasks of motor imagery may therefore provide one means by which the self-monitoring of goal directed actions can be examined in patients with schizophrenia.

Several studies have reported eye movement disturbances in schizophrenia, such as deficits in anti-saccade tasks (Harris et al., 2009), in smooth pursuit (Clementz and Sweeney, 1990; Sweeney et al., 1998), and in visual exploration. Visual scan path impairments have even been proposed to serve as a trait marker for schizophrenia (Loughland et al., 2002a, 2004). The majority of these studies have been conducted with faces as stimuli. Patients with schizophrenia usually make fewer fixations than healthy controls on salient facial features (eyes, nose, and mouth) and their exploration duration is reduced. Restricted visual scan path have also been reported with other, less socially relevant stimuli, like geometrical shapes (Kojima et al., 1992; Obayashi et al., 2003); Rorschach stimuli (Minassian et al., 2005), and photographs of landscapes, meaningless textures, and fractals (Bestelmeyer et al., 2006). In most of these studies, visual scanning has been examined under free viewing conditions (Green et al., 2003; Bestelmeyer et al., 2006) and for some of them (Streit et al., 1997; Loughland et al., 2002a,b), participants were asked to determine the facial expression.

Based on these findings, especially the reduced visual exploration, the fact that patients with schizophrenia exhibit significant difficulties with planning and organization of action (Pantelis et al., 1997; Jogems-Kosterman et al., 2001; Delevoye-Turrell et al., 2006, 2007), and that these patients show an exaggerated susceptibility to distraction (Ducato et al., 2008a,b), we expected patients to explore less than controls in the free viewing condition and to exhibit a larger amount of fixations to the task-irrelevant (distractor) objects in the active viewing condition. In this study, we also used motor imagery in patients with schizophrenia to assess whether the generation of an internal image of intended actions would be impaired in schizophrenia.

#### **MATERIALS AND METHODS**

#### **PARTICIPANTS**

Eighteen medicated in- and out-patients fulfilling the DSM-IV diagnostic criteria for schizophrenia (American Psychiatric Association, 1994) and 18 healthy control participants without a psychiatric diagnosis (axis I and II) and without a family history of mental illness took part in the experiment. Control participants did not take any medication at the time of test. Patients were recruited in the Department of General Psychiatry in Lille University Hospital, the Department of General Psychiatry at Arras General Hospital and the Psychology Center in Béthune (all located in northern France). Control participants were age and sex-matched students and members of the medical staffs of the different hospitals. Inclusion criteria for all groups were normal or corrected-to-normal vision (assessed by the Snellen chart). Exclusion criteria were recent history of substance abuse, ocular disease, epilepsy and other neurological disorders, and failure to understand the instructions. After interview, schizophrenic symptoms were rated using the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1986). The study was approved by the local Ethics Committee. Informed consent was obtained from all participants. Group characteristics are summarized in **Table 1**.

#### **STIMULI**

Different scene layouts were built containing task-relevant objects (*n* = 7) required to make a butter and jelly sandwich and to pour a glass of water, and task-irrelevant objects (*n* = 7). All objects were laid out on a table. When the participant was seated at the table, with all objects within reach, the plate close to the observer subtended about 20◦ of visual angle, and the butter and jelly subtended about 7◦. All objects were located within a region covering 90◦. These scenes were then photographed. The scene pictures were 768 pixels in height and 1024 pixels in width, and subtended a vertical visual angle of 21◦ and a horizontal visual angle of 29◦ at a viewing distance of 60 cm (**Figure 1**).

#### **Table 1 | Means and standard deviations (***SD***) of participant demographics.**


*\*12 out of 18 patients were treated with benzodiazepine medication.*

**FIGURE 1 | Example of scene layout used in this experiment.** The scene contained both 7 task-relevant *(bread, butter, jelly, knife, plate, glass, water bottle)* and 7 irrelevant objects.

#### **EQUIPMENT**

Visual scan paths were collected monocularly. Monocular (right) eye position was monitored using the iViewX™ HED from SensoMotoric Instruments (Teltow, Germany) eye tracker with a scene camera. The video based eye tracker was head-mounted, using infrared reflection to provide an eye-in-head signal at a sampling rate of 50 Hz and accuracy of 1◦. The scene camera mounted on the head was positioned so that its field of view was centered on the subject's field of view. Calibration was performed using a five-point grid. Following calibration, the eye tracker creates a cursor, indicating eye-in-head position, which was superimpose on the scene video. This cursor was just for purposes of analyses but it was not visible for the participant. The scene camera moved with the head, so the eye-in-head signal indicated the gaze point. The eye tracker thus provided a video recording of eye position from the participant's perspective on the scene and the data analysis was based on the video, as there was no separate numerical data stream. The video recordings were analyzed on a frame-by-frame basis, recording the time of initiation and termination of each eye movement, and the spatial locations of the fixations. Saccades appeared in the large displacements of the cursor between video frames. The beginning and end of each saccade was identified and recorded using a video analysis tool. Fixations were defined visually when the cursor stayed within a given location (less than a degree) defined by the noise level of the tracker. Thus, fixations were defined jointly by position and velocity. Blinks were detected by occlusion of the pupil and the cursor was occluded during the blink.

#### **PROCEDURE**

Participants wore an eye tracker mounted on the head, and were seated in front of a computer screen.

Participants started with a "free viewing" task in which they were asked to look at a scene on a computer screen for 10 s. Then, participants performed an "active viewing" task: picturing themselves making a butter and jelly sandwich and pouring a glass of water in front of a computer screen. In the imagined movement task (active viewing), timing began when the experimenter said "go" (when the scene was displayed), and stopped when the participant said "stop" upon completing imagined action.

Before the experiments, the layout was occluded by a black screen showing the five calibration points, enabling the participants to be calibrated on the plane of the working surface. The participant had to fixate the targets (white dots) while his/her eye positions were recorded by the system. Using these reference points, the system creates a mapping function that relates all eye positions to points in the calibration area. Once the calibration was completed, this was removed, and participant immediately started the task. A re-calibration procedure was performed after each task. The entire session lasted approximately 30 min; the calibration time depending on participants.

#### **DATA ANALYSIS**

Temporal and spatial characteristics of gaze patterns were extracted for each group of participants and each task. We measured the total gaze duration on specific objects, in instances where several successive fixations were made on the same object. Gaze duration on both relevant and irrelevant objects was determined. Eye movement variables were submitted to analyses of variance (ANOVA) using the STATISTICA software from StatSoft (version 7.1, F-94700 Maisons-Alfort, France). Possible confounding effects of medication, illness duration, positive and negative symptom categories (indexed by the PANSS), patient categories (in- and out-patients), gender were examined. For patients with schizophrenia, there were no significant correlations between medication, illness duration or symptom category on one hand and any of the scan path variables on the other. Relationships between gender and scan path variables were computed for both groups but no statistically significant correlations emerged.

# **GAZE DURATIONS ON RELEVANT OBJECTS (RO) AND IRRELEVANT OBJECTS (IO)**

A 2 (group: Patients/Controls) × 2 (task: Free/Active viewing) × 2 (objects: Relevant/Irrelevant) repeated measures multivariate ANOVA showed a significant main effect of group [*F(*1*,* <sup>34</sup>*)* = 37*.*0, *p <* 0*.*0001], task [*F(*1*,* <sup>34</sup>*)* = 38*.*9, *p <* 0*.*0001], and objects [*F(*1*,* <sup>34</sup>*)* = 79*.*0, *p <* 0*.*0001] on the gaze durations. There was also a significant interaction between group, task and objects [*F(*1*,* <sup>34</sup>*)* = 34*.*6, *p <* 0*.*0001] (**Figure 2**).

#### *Free viewing condition*

Patients with schizophrenia exhibited longer gaze durations than controls on objects in the free viewing condition [RO: *F(*1*,* <sup>34</sup>*)* = 63*.*3, *p <* 0*.*0001 and IO: *F(*1*,* <sup>34</sup>*)* = 10*.*9, *p <* 0*.*003].

Patients and controls looked equally at relevant and irrelevant objects in the free viewing condition [Patients, RO vs. IO, *F(*1*,* <sup>34</sup>*)* = 1*.*4, *p* = 0*.*2, ns – Controls, RO vs. IO, *F(*1*,* <sup>34</sup>*)* = 0*.*9, *p* = 0*.*4, ns].

#### *Active viewing condition*

Patients with schizophrenia exhibited longer gaze durations than controls on irrelevant objects in the active viewing condition [IO: *F(*1*,* <sup>34</sup>*)* = 40*.*5, *p <* 0*.*0001], whereas controls made longer gaze durations on relevant objects in the same condition [RO: *F(*1*,* <sup>34</sup>*)* = 9*.*6, *p <* 0*.*004].

Patients and controls looked more at relevant than at irrelevant objects in the active viewing condition [Patients, RO vs. IO, *F(*1*,* <sup>34</sup>*)* = 27*.*3, *p <* 0*.*0001 – Controls, RO vs. IO, *F(*1*,* <sup>34</sup>*)* = 174*.*5, *p <* 0*.*0001].

#### *Free vs. active viewing conditions*

As controls, patients looked more at relevant objects in the active viewing condition [Patients, RO: free vs. active viewing, *F(*1*,* <sup>34</sup>*)* = 12*.*8, *p <* 0*.*002 – Controls, RO: free vs. active viewing, *F(*1*,* <sup>34</sup>*)* = 196*.*1, *p <* 0*.*0001], but unlike controls, patients with schizophrenia did not show any significant difference on irrelevant objects between the free viewing and the active viewing condition [IO:

free vs. active viewing, *F(*1*,* <sup>34</sup>*)* = 1*.*9, *p* = 0*.*2, ns], whilst controls made fewer gaze durations on irrelevant objects in the active viewing condition [IO: free vs. active viewing, *F(*1*,* <sup>34</sup>*)* = 7*.*5, *p <* 0*.*01].

An example of one participant's fixations for each group is given in **Figure 3**. Each participant began to make a series of fixations on the task-relevant objects as the bread, the butter, the knife or the plate. The order in which the relevant objects to accomplish the task were fixated was the same for patients and controls.

#### **PERFORMANCE TASK DURATION**

A 2 (group: Patients/Controls) × 1 (task: Active viewing) ANOVA showed no significant difference between the two groups on the total task duration [*F(*1*,* <sup>34</sup>*)* = 0*.*1, *p* = 0*.*8]. For accomplishing the imagined task, patients needed on average 27 s and controls needed 28 s.

# **DISCUSSION**

Eye movements of patients with schizophrenia and healthy control participants were recorded both under a free viewing condition and under an active viewing condition with scene image as stimuli. In the active viewing condition, participants were instructed to imagine making accurate movements between different task-relevant objects.

Consistent with previous eye movement studies using images as stimuli (e.g., Bestelmeyer et al., 2006), abnormalities in patients with schizophrenia were found in the free viewing condition. Patients showed longer fixation durations, and fewer eye fixations, than healthy control participants in the free viewing condition, but they did not differ from control participants in the active viewing condition for temporal scan path variables. This result can be related to studies on attentional control and cognitive flexibility in schizophrenia (Granholm et al., 1999; Ducato et al., 2008a,b). Indeed, patients with schizophrenia are able to normalize their pattern of visual exploration when they are actively involved in more demanding tasks (Kurachi et al., 1994; Tonoya et al., 2002; Delerue et al., 2010; Delerue and Boucart, 2012, 2013).

Our results show that the patients' visual pattern improved in the active viewing condition. However, for spatial scan path variables, patients looked more at task-irrelevant objects (distractors) than controls in the active viewing condition in which participants were asked to picture themselves making a sandwich in moving eyes to task-relevant objects on a scene image. This difference in the spatial distribution of fixations between the two groups in the imagined task raises the possibility that patients with schizophrenia did not perform the imagined motor task correctly. However, an aspect of the results suggests otherwise. Indeed, the order in which the relevant objects to accomplish the task were fixated was the same for patients and controls. In addition to recording eye movements, the time to execute the imagined task was measured as a means of ensuring that the task was performed in an equivalent manner between the two groups of participants. The results showed no significant difference between the two groups on the imagined task duration. This suggests that patients were complying with instructions in the imagined task.

Visual search has been found to be modulated by action intentions in healthy participants (Bekkering and Neggers, 2002; Hannus et al., 2005). Abnormalities in action have been observed in various forms in patients with schizophrenia. Clinically, they appear as poverty of action, disorganized behavior, stereotyped, and incoherent actions. Behavioral experimental studies have reported disturbances of action production, whether an action has to be explicitly performed, like figure copying tasks (Jogems-Kosterman et al., 2001; Grootens et al., 2009) or planning of motor sequences (Delevoye-Turrell et al., 2007) or not, like intention of action, observation of action, or manipulation of action knowledge. Zalla et al. (2001, 2006) investigated the ability of patients with schizophrenia to organize action knowledge and elaborate a plan of action. The authors found that patients with schizophrenia performed significantly worse than controls, and frontal lobe patients, in an action verbal generation task. They suggested that patients with schizophrenia are impaired in the sequential organization of action events.

Other studies (Danckert et al., 2002b; Maruff et al., 2003) have assessed the ability of patients with schizophrenia to perform a motor imagery task. These studies have reported deficits in motor imagery in these patients. The degree of impairment in imagined movements was not correlated with symptom profile (Danckert et al., 2002b). However, in a separate study, patients with and without passivity delusions were tested using a similar motor imagery task, and only the patients with passivity delusions showed a specific impairment in the execution of imagined motor sequences (Maruff et al., 2003). These authors have found a deficit in the ability to generate or make use of internal model of intended actions. This impairment is likely to reflect dysfunction in parietal association cortices that have been shown to be crucial for making use of internal models of goal-directed movements (Sirigu et al., 1995; Danckert et al., 2002a).

Compared with healthy participants, the visual exploration pattern of patients could be explained by their susceptibility to

distraction. Indeed, it is known that patients with schizophrenia exhibit a higher sensitivity to distraction that healthy observers (Ducato et al., 2008a,b). In our study, it might be that a salient feature in the object (e.g., color, shape, or contrast edges) automatically captured the patients' attention and that they found it difficult to disengage their attention from that salient feature. Impaired spatial attention, appearing in longer duration of attentional disengagement, has been reported in patients with schizophrenia with the Posner's paradigm (Posner et al., 1988; Wigal et al., 1997). Moreover, studies using the antisaccade paradigm have demonstrated that patients show difficulty inhibiting a reflexive saccade (Hutton and Ettinger, 2006).

The patients' visual exploration may also be viewed in the framework of Gestalt dysfunctions in schizophrenia. Indeed, a majority of studies showed impairments of perceptual organization in schizophrenia (Uhlhaas and Silverstein, 2003, 2005; Silverstein and Uhlhaas, 2004; Brand et al., 2005), indicating that many patients with schizophrenia appear to have deficient gestalt perception.

One possible limitation of this study is that only one trial per condition was used. Thus, measures of reliability come from average performance over fixations within a task, and similarity between subjects, as indicated by the standard error measures. Moreover, only a small sample of participants (2 × 18) has been

#### **REFERENCES**


Goodale, M. A. (2002a). Selective, non-lateralized impairment of motor imagery following right parietal damage. *Neurocase* 8, 194–204. doi: 10.1093/neucas/8.3.194


tested. The magnitude of the reported effects indicate that this is not a major concern, however. This is an exploratory pilot study, and future studies on different imagining tasks are needed to confirm the presented results. Another limitation is that although in our study, antipsychotic dosage equivalents did not appear to be correlated to various eye movement variables, we cannot exclude an affect of medication. Kojima et al. (1990) have reported that neither the number of fixations nor scan path length were correlated with the chlorpromazine-equivalent dosage in 50 chronic schizophrenic patients, and this was confirmed by Matsushima et al. (1992). Streit et al. (1997) and Loughland et al. (2002a,b) found no relation between dysfunctional scan paths in schizophrenia and medication, whilst Williams et al. (2003) reported that patients treated with risperidone showed greater attention to salient features. Reilly et al. (2008) showed that pharmacological treatment might have an effect on eye movement control in saccadic tasks, and Giersch et al. (2010) reported effects of benzodiazepines on cognitive functions.

The present study shows that patients with schizophrenia improve their visual exploration under task-driven attentional conditions, in an imagined task that is an example of embodied cognition. However, our results also show that patients with schizophrenia have some difficulties to generate accurate internal representation of goal-directed actions.

processing in schizophrenia. *Cogn. Neuropsychiatry* 17, 334–350. doi: 10.1080/13546805.2011.646886


*Br. J. Psychol.* 91, 451–471. doi: 10.1348/000712600161934


of implicit processing. *Europ. J. Neurosci.* 17, 2735–2740. doi: 10.1046/j.1460-9568.2003.02695.x


in patients with schizoplrenia. *Schizophr. Res.* 12, 75–80. doi: 10.1016/0920-9964(94)90086-8


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; accepted: 21 June 2013; published online: 12 July 2013.*

*Citation: Delerue C and Boucart M (2013) Imagined motor action and eye movements in schizophrenia. Front. Psychol. 4:426. doi: 10.3389/fpsyg. 2013.00426*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Delerue and Boucart. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Advanced analysis of free visual exploration patterns in schizophrenia

# *Andreas Sprenger 1†, Monique Friedrich2†, Matthias Nagel 2,3, Christiane S. Schmidt 4, Steffen Moritz <sup>4</sup> and Rebekka Lencer 2,5\**

*<sup>1</sup> Department of Neurology, University of Luebeck, Luebeck, Germany*

*<sup>2</sup> Department of Psychiatry and Psychotherapy, University of Luebeck, Luebeck, Germany*

*<sup>3</sup> Asklepios Klinik Nord – Wandsbek, Clinic of Psychiatry and Psychotherapy, Hamburg, Germany*

*<sup>4</sup> Department of Psychiatry and Psychotherapy, University Medical Centre Hamburg-Eppendorf, Hamburg, Germany*

*<sup>5</sup> Department of Psychiatry and Psychotherapy, University of Muenster, Muenster, Germany*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA Rick Adams, University College London, UK*

#### *\*Correspondence:*

*Rebekka Lencer, Department of Psychiatry and Psychotherapy, University of Muenster, Albert Schweitzer Campus 1, Geb. A9, 48149 Muenster, Germany e-mail: rebekka.lencer@ ukmuenster.de †These authors have contributed equally to this work.*

**Background:** Visual scanpath analyses provide important information about attention allocation and attention shifting during visual exploration of social situations. This study investigated whether patients with schizophrenia simply show restricted free visual exploration behavior reflected by reduced saccade frequency and increased fixation duration or whether patients use qualitatively different exploration strategies than healthy controls.

**Methods:** Scanpaths of 32 patients with schizophrenia and age-matched 33 healthy controls were assessed while participants freely explored six photos of daily life situations (20 s/photo) evaluated for cognitive complexity and emotional strain. Using fixation and saccade parameters, we compared temporal changes in exploration behavior, cluster analyses, attentional landscapes, and analyses of scanpath similarities between both groups.

**Results:** We found fewer fixation clusters, longer fixation durations within a cluster, fewer changes between clusters, and a greater increase of fixation duration over time in patients compared to controls. Scanpath patterns and attentional landscapes in patients also differed significantly from those of controls. Generally, cognitive complexity and emotional strain had significant effects on visual exploration behavior. This effect was similar in both groups as were physical properties of fixation locations.

**Conclusions:** Longer attention allocation to a given feature in a scene and less attention shifts in patients suggest a more focal processing mode compared to a more ambient exploration strategy in controls. These visual exploration alterations were present in patients independently of cognitive complexity, emotional strain or physical properties of visual cues implying that they represent a rather general deficit. Despite this impairment, patients were able to adapt their scanning behavior to changes in cognitive complexity and emotional strain similar to controls.

**Keywords: schizophrenia, visual scanpath, visual exploration, focal processing, exploration strategy, attentional landscape, scanpath similarity**

#### **INTRODUCTION**

Deficits in the perception of social situations are suggested to underlie impaired social interaction in patients with schizophrenia (Addington et al., 2006). Reduced integration of visual context information in real-world situations has been suggested as one factor contributing to altered visual perception in patients (Green et al., 2005, 2008; Butler et al., 2008). Analyses of visual scanpaths provide important knowledge about how and when visual information is processed during visual exploration. A visual scanpath constitutes a sequence of voluntary saccades each shifting the focus of attention from one location of interest to the next thereby tracing the direction and extent of gaze when a subject extracts information from complex visual scenes (Noton and Stark, 1971). Scanpaths are affected both by sensory information such as physical stimulus properties, e.g., luminance or chromaticity contrasts, as well as the semantic relevance of a stimulus, e.g., cognitive complexity or emotional content (Bradley et al., 2011).

Previous studies using visual scanpath analyses in schizophrenia have mainly investigated face recognition (Frith et al., 1983; Gordon et al., 1992; Phillips and David, 1997; Streit et al., 1997; Williams et al., 1999) and the exploration of more complex scenes (Gaebel et al., 1987; Phillips et al., 2000). Some analyses focused on comparisons of quantitative eye-movement parameters, e.g., saccade frequency and amplitude (Phillips and David, 1997; Streit et al., 1997; Benson et al., 2007). Such studies have revealed restricted visual scanpaths in patients compared to healthy controls in terms of fewer and smaller saccades (Phillips and David, 1997, 1998; Williams et al., 1999; Loughland et al., 2002a). However, these kind of analyses do not provide information about whether patients and healthy individuals explore similar features in a scene with respect to their semantic and emotional contents or their physical properties. More recent studies suggested that patients tend to make fewer fixations on salient features in faces (eyes, nose, and mouth) than healthy controls (Williams et al., 1999; Loughland et al., 2002a,b). This pattern of visual avoidance was associated with deficits in the recognition of particular emotions (Loughland et al., 2002a; Green et al., 2003) and underlines the assumption that social perception in schizophrenia may be controlled by early restriction of input to visual cortex (Bestelmeyer et al., 2006; Green et al., 2008). Others have pointed out that during visual exploration visual information is gathered to test a-priori hypotheses and beliefs about the world (Friston et al., 2012). This model is based on the assumption that biological systems maximize the Bayesian evidence for their model of the world through an active sampling of sensory information (Friston et al., 2012). With respect to schizophrenia this suggests that scanpath abnormalities in patients reflect attention allocation and shifting driven by aberrant or more uncertain beliefs about the world compared to that of healthy individuals (Fletcher and Frith, 2009).

In contrast to this assumption, abnormalities of saccade frequencies in patients have also been reported from studies using abstract (Manor et al., 1999; Obayashi et al., 2003) and geometric stimuli (Kojima et al., 1989, 1990, 1992) which differed in complexity but were free of emotional content. Minassian et al. (2005) therefore suggested a general impairment of visual scanning and exploration independent of the semantic content of an image that may be regarded as a stable behavioral marker associated with schizophrenia (Bestelmeyer et al., 2006; Benson et al., 2012).

In the present study we were interested in spatial and temporal aspects of visual exploration behavior of social daily life scenes in patients with schizophrenia. Stimuli were taken from the Integrated Psychological Training (IPT) program for patients with chronic schizophrenia (Roder et al., 1997). One part of the IPT focuses on remediation of social perception by improving visual exploration strategies, training patients to systematically collect visual information from a given visual scene prior to drawing conclusions and interpreting its content. We used advanced scanpath analyses including examination of temporal changes in exploration behavior, cluster analyses, comparisons of attentional landscapes (Pomplun et al., 1996) and scanpath similarities. We also studied whether the stimuli's cognitive complexity and emotional strain had different effects on exploration strategies in patients and controls and whether physical properties of fixation locations differed between groups.

# **MATERIALS AND METHODS**

#### **SUBJECTS**

Thirty-two psychopathologically stable patients from in- and out-services from two sites, the University Hospital of Luebeck (13 males; 2 females) and the University Hospital of Hamburg (9 males; 8 females) met DSM-IV-criteria for schizophrenia (American Psychiatric Association, 1994; **Table 1**). Patients did



not differ between sites for age [*t(*30*)* = 0*.*204, *p* = 0*.*232], mean illness duration [*t(*30*)* = −0*.*59, *p* = 0*.*56] or mean age at onset of first psychotic symptoms [*t(*30*)* = 0*.*80, *p* = 0*.*43]. Diagnoses were established using the German version of the Mini-International Neuropsychiatric Interview (M.I.N.I., Sheehan et al., 1998). Symptoms were assessed on the Positive and Negative Syndrome Scale (PANSS, Kay et al., 1987). Patients showed mild positive syndromes (∼20th percentile), mild negative syndromes (∼15th percentile) and a mild general psychopathology (∼25nd percentile). The mean PANSS difference score of −1.45 (*SD* = 5*.*97) indicates a predominance of negative symptoms in patients.

All patients were on medication, usually antipsychotics (olanzapine *N* = 6, quetiapine *N* = 5, amisulpride *N* = 4, risperidone *N* = 3, clozapine *N* = 2, ziprasidone *N* = 2, phenothiazine *N* = 2, flupentixole *N* = 1). Some patients additionally received antidepressants (venlafaxine *N* = 2, citaloprame *N* = 2, paroxetine *N* = 1, sertraline *N* = 1, trimipamine *N* = 1). All patients were off benzodiazepines for at least 48 h prior to testing.

Thirty-three control subjects (20 males, 13 females) with no reported history of a major psychiatric disorder were recruited to match for age with the patient group [range 24–61 years, mean age in Luebeck (*N* = 16): 35.1 years ±9.9; Hamburg (*N* = 17): 36.0 years ±11.0; *F*group*(*1*,* <sup>61</sup>*)* = 1*.*0, *p* = 0*.*66, *F*site*(*1*,* <sup>61</sup>*)* = 0*.*00, *p* = 0*.*99, no interaction of GROUP × SITE]. Inclusion criteria for all participants comprised: (1) age 18–65 years, (2) normal or corrected-to-normal vision, (3) no current or reported substance dependency and no substance abuse during 4 weeks prior testing, (4) no known systemic or neurological disease. Visual acuity and color vision were assessed using Landolt Ring charts and the Ishihara Color test. Each participant gave written informed consent after having carefully been informed about the study. The study was approved by the local ethics committees of the universities of Luebeck and Hamburg.

#### **STIMULI**

Six color photos depicting social daily life scenes were chosen from the "Integrated Psychological Therapy Program for Patients with Schizophrenia" (Roder et al., 1997). Photos were selected according to their ratings on the two dimensions "cognitive complexity" and "emotional strain" provided by Roder et al. (1997). Complexity and emotional strain scores for each photo were calculated by multiplying the percentage values of each of the evaluation categories reported by Roder et al. (1997) with a weighting factor (low = 1, moderate = 2, high = 3), summing these up and dividing the sum by 3, **Figure 1**.

Photos were presented in randomized order for 20 s each and subjects were instructed to look at the photos as if they were spectators of the depicted scene. After each presentation, participants were asked to rate their emotional reaction to the photos on a 3-item rating scale ("none," "moderate," "strong") by clicking a button on a keypad.


Additionally, participants were asked whether they felt they had enough time to look at the photo and whether they had been able to grasp the photo's content on a 3 point scale ("hardly," "partly," or "completely").

#### **RECORDING PROCEDURE**

At both sites, assessment took place in a dimly lit room. In Luebeck, participants were seated 180 cm in front of a 151 × 120 cm screen, which equates to an angle of 45*.*3 × 36*.*9◦. In

from the evaluations provided by Roder and co-workers. Higher scores indicate a higher cognitive complexity resp. emotional strain (see Materials and Methods).

Hamburg, participants were seated 57 cm in front of a 36*.*3 × 27 cm CRT monitor (19 inch) comprising a visual angle of 35*.*3 × 26*.*6◦. Stimulus resolution was 1024 × 768 pixels at both sites. Subjects were instructed not to move their head during the recording. Eye and head movement recordings at both sites were assessed with the same device using a video-based EyeLink I eye tracking system (250 Hz, SR Research Ltd., Ottawa, ON, Canada). An additional camera tracked four infrared markers mounted on the visual stimulus display for head motion compensation and true gaze position tracking.

#### **DATA ANALYSIS**

A semi-automatic computer program written in MatLab® (R2010a, The Mathworks Inc., Natic, MA, USA) was used to read and calibrate the eye tracking data. Eye position data was filtered by a 100 Hz Gaussian filter. Saccadic eye movements were detected by identifying an initial eye velocity above 30◦/s with its peak velocity occurring within a time window of 60 ms. Beginning and end of a saccade were defined as the points at which velocity crossed 20◦/s with a minimum saccade amplitude of 0.3◦. All saccades, blinks and artifacts were detected and checked manually. Saccades with amplitudes *<*0.6◦ were classified as corrective saccades and excluded from amplitude analyses since they reflect the relocation of an object of interest rather than a complete attentional shift to a new location. The following parameters were determined for each photo: location, number and duration of fixations, number and amplitudes of saccades and the resulting scanpath lengths as the sum of saccade amplitudes during the 20 s scanning time. Additionally, the exploration time of 20 s was divided into four intervals of 5 s each to test for changes over the exploration time.

#### **CLUSTER ANALYSES**

To investigate whether possible disturbances of fixation and saccade behavior led to alterations of areas of interest in patients compared to controls, fixation locations were used in N-2 cluster analyses. Euclidian distances of all fixations per photo were calculated and linkage parameters were obtained using Ward's method (Bortz, 1999). Cluster and distance matrices were calculated for cluster solutions starting at 2 clusters up to N-1 clusters with *N* = number of fixations per photo. This method allows calculating distances within and between clusters for each cluster solution. The minimum of the ratio of distances within/between clusters was considered as the optimal cluster solution. The following parameters were obtained for each photo: total number of clusters, total number of fixations per cluster, total fixation time within each cluster (ms), and the total number of changes between clusters.

#### **ATTENTIONAL LANDSCAPES**

To examine the "where" question, thus, whether patients allocated attention to different locations than controls, fixations on photos of each subject were weighted by their duration and smoothed with a 2D Gaussian function with σ = 1◦ visual angle. Resulting maps are known as "attentional landscapes" (Pomplun et al., 1996). These maps were compared between healthy controls and patients using *T*-tests, **Figure 5**. To correct for multiple comparisons, the significance threshold was set to *p <* 0*.*001.

#### **SIMILARITIES BETWEEN SCANPATHS**

Fixation sequences of each subject were compared to each other subject in the combined control-patient group using the Needleman–Wunsch algorithm (Needleman and Wunsch, 1970) implemented in a modified version of the ScanMatch toolbox (Cristino et al., 2010). Photos were split into areas of a 12 × 8 grid and fixations were assigned to these grid areas. Sub matrices were calculated using the twofold standard deviation of the mean saccade amplitude per photo divided by grid size. Similarity scores for each subject compared to each of the 64 other subjects were determined resulting in similarity matrices of 65 × 65 similarity scores for each photo. Subsequently, median similarity scores in each subject were defined for their scanpath similarity with that of the control group and that of the patient group, for each photo. Thereafter, difference scores between these two similarity scores were calculated for each subject and photo. Positive values indicate a stronger scanpath similarity to the control group whereas negative values indicate a stronger scanpath similarity to the patient group. In order to test sensitivity of scanpath similarity values discriminant analyses were performed using SPSS.

#### **PHYSICAL PROPERTIES**

To describe the physical properties of the photos that determined sensory information the following parameters were defined: luminance, luminance contrast chromaticity contrast and static contrast (Machner et al., 2012). RGB values of the photos were transferred to YCbCr color space in order to separate luminance and chromaticity. Contrasts were calculated by comparing a local patch of 2 × 2◦ around fixation to a global patch around fixation of the same size. For static contrasts patches were compared using black/white images of the photos which were transferred by edge detection using the canny method implemented in Matlab® with a sigma of 3 pixels.

#### **STATISTICAL ANALYSES**

All statistical procedures were performed using the software package SPSS (21.0.0.1, IBM Inc, New York/USA). Analyses of variance (ANOVA) and *T*-tests were used for group comparisons on metric data level, e.g., eye movement parameters. Three-way repeated measurement analyses of variance (ANOVAs: PHOTO × GROUP × SITE) were used to test for within-subject effects, i.e., differences between photos, and for between-subject effects, i.e., group and site, in eye movement parameters. Additionally, Three-Way ANOVAs (TIME × GROUP × SITE) were performed for each photo separately whenever changes over the time course of photo presentation were of interest. There were no significant interactions of GROUP × SITE for any parameter of interest (*p >* 0*.*15) indicating that group differences were similar across sites. Based on these analyses, patient and controls groups were collapsed across sites. Tables with means and standard errors of all parameters of interest from each site as well as from the combined sample are available as supplementary material. Results from Two-Way ANOVAs (PHOTO × GROUP) will be reported that were followed by *post-hoc t*-tests or One-Way ANOVAs in each group separately. Greenhouse-Geisser correction of degrees of freedom was applied when repeated-measurement factors had two or more levels. Corrected *p*-values and effect sizes indicated as partial Eta (η<sup>2</sup> *<sup>p</sup>*) for ANOVAs and Cohen's *d*' for *T*-tests (Cohen, 1988) will be reported.

For group comparisons of ordinal data, e.g., subject's emotional reaction ratings, we performed non-parametric Mann-Whitney-U tests. Spearman's Rho was used to test for correlations between eye movement parameters and data on ordinal level, i.e., subjects' ratings and clinical data (PANSS scores, time since first psychotic symptom, age of onset).

## **RESULTS**

#### **GENERAL DIFFERENCES BETWEEN PHOTOS**

Repeated measures ANOVA showed that photos differed across groups with respect to the number of fixations [*F*photo*(*5*,* <sup>61</sup>*)* <sup>=</sup> <sup>15</sup>*.*39, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*201], mean fixation time [*F*photo*(*5*,* <sup>61</sup>*)* <sup>=</sup> <sup>9</sup>*.*19, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*131], mean saccade amplitude [*F*photo*(*5*,* <sup>61</sup>*)* <sup>=</sup> <sup>56</sup>*.*03, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*479], and scanpath length [*F*photo*(*5*,* <sup>61</sup>*)* <sup>=</sup> <sup>36</sup>*.*62, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*375], **Figure 2**. Photo differences were also found for the parameters obtained in cluster analyses including the total number of clusters [*F*photo*(*5*,* <sup>62</sup>*)* <sup>=</sup> <sup>8</sup>*.*97, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*01, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*126], fixations per cluster [*F*photo*(*5*,* <sup>62</sup>*)* <sup>=</sup> <sup>5</sup>*.*99, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*088], total fixation time per cluster [*F*photo*(*5*,* <sup>57</sup>*)* <sup>=</sup> <sup>18</sup>*.*54, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*245],

and changes between clusters [*F*photo*(*5*,* <sup>62</sup>*)* = 7*.*65, *p <* 0*.*001, η2 *<sup>p</sup>* = 0*.*110], **Figure 3**. Scanpath similarities also differed between photos [*F*photo*(*5*,* <sup>59</sup>*)* <sup>=</sup> <sup>3</sup>*.*2, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*014, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*051], **Figure 6**.

#### **GROUP DIFFERENCES IN FIXATION FREQUENCY AND MEAN FIXATION TIME**

Across photos, patients made one-fifth fewer fixations per photo than controls [*patients*: 51.83 (*SD* = 8*.*91), *controls:* 64.25, (*SD* = <sup>8</sup>*.*75), *<sup>F</sup>*group*(*1*,* <sup>61</sup>*)* <sup>=</sup> <sup>31</sup>*.*03, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*337, **Figure 2A**] and mean fixation time was longer in patients than controls [*patients*: 269 ms (*SD* = 48), *controls:* 221 ms (*SD* = 32), *<sup>F</sup>*group*(*1*,* <sup>61</sup>*)* <sup>=</sup> <sup>24</sup>*.*23, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*284, **Figure 2B**]. These group differences did not differ between photos (no interactions GROUP × PHOTO).

To test for changes in mean fixation time over the exploration time course, exploration time was divided into four intervals (0–5, 5–10, 10–15, and 15–20 s). Besides the expected group effect [*F*group*(*1*,* <sup>63</sup>*)* <sup>=</sup> <sup>24</sup>*.*71, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *p* = 0*.*282], mean fixation time differed between time intervals [*F*time*(*3*,* <sup>63</sup>*)* <sup>=</sup> <sup>18</sup>*.*68; *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*229]. This time effect differed between groups [*F*time×group*(*3*,* <sup>63</sup>*)* <sup>=</sup> <sup>2</sup>*.*96, *<sup>p</sup>* <sup>=</sup> <sup>0</sup>*.*045, <sup>η</sup><sup>2</sup> *p* = 0*.*045, **Figure 4A**]. Exploring the time effect in each group separately showed significant changes in mean fixation time in both

**FIGURE 4 | Fixation durations (A) and mean saccadic amplitude (B) over the time course of free visual exploration of daily life scenes in patients with schizophrenia (***N* **= 32) and healthy controls (***N* **= 33).** The exploration time of 20 s was divided into four intervals of 5 s each. *Post-hoc t*-tests confirmed significant group differences in all four time intervals [0–5 s: *t(*63*)* = −4*.*41,*p <* 0*.*001, *d* = −1*.*09; 5–10 s: *t(*63*)* = −3*.*17, *p* = 0*.*002, *d* = −0*.*79; *10–15 s*: *t(*63*)* = −4*.*15, *p* = 0*.*001, *d* = −1*.*03; 15–20 s: *t(*63*)* = −5*.*99,*p* = 0*.*001, *d* = −1*.*49]. Indicated are means with standard errors, for *post-hoc* comparisons between groups: ∗*p <* 0*.*05, ∗∗*p <* 0*.*01, ∗∗∗*p <* 0*.*001.

groups [*patients*: *<sup>F</sup>*time*(*3*,* <sup>29</sup>*)* <sup>=</sup> <sup>10</sup>*.*63; *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*255;*controls: F*time*(*3*,* <sup>30</sup>*)* <sup>=</sup> <sup>10</sup>*.*62; *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*01, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*249]. Comparisons of regression coefficients of fixation duration on time intervals revealed that mean fixation time increase was larger in patients than controls [*t(*63*)* = −3*.*6, *p* = 0*.*001, *d* = −0*.*89].

#### **GROUP DIFFERENCES IN MEAN SACCADIC AMPLITUDE AND SCANPATH LENGTH**

Both groups did not differ on the number of corrective saccades (*p* = 0*.*072) and corrective saccades were excluded from analyses of mean saccadic amplitude. Across photos, mean saccadic amplitude in patients was smaller than in controls [*patients:* 3*.*90◦ (*SD* = 0*.*85), *controls:* 4*.*33◦ (*SD* = 0*.*77), *F*group*(*1*,* <sup>61</sup>*)* = 7*.*64,*p <* 0*.*01, η<sup>2</sup> *<sup>p</sup>* = 0*.*0111, **Figure 2C**]. The scanpath length reflecting the sum of saccade amplitudes was shorter in patients than controls [*patients:* 256◦ (*SD* = 57), *controls:* 349◦ (*SD* = 59), *<sup>F</sup>*group*(*1*,* <sup>61</sup>*)* <sup>=</sup> <sup>38</sup>*.*96, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*001, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*390, **Figure 2D**]. These group differences did not differ between photos (no interactions of GROUP × PHOTO). Over the time course of exploration, mean saccadic amplitude decreased in both groups [*F*time*(*3*,* <sup>61</sup>*)* = 4*.*79, *p <* 0*.*01, η<sup>2</sup> *<sup>p</sup>* = 0*.*073] with significant smaller amplitudes in patients compared to controls [*F*group*(*1*,* <sup>63</sup>*)* = 7*.*49, *p <* 0*.*01, η2 *<sup>p</sup>* = 0*.*106] but there was no difference in change of mean saccadic amplitude over time between groups (no interaction of TIME × GROUP), **Figure 4B**.

#### **GROUPS DIFFERENCES IN CLUSTER ANALYSES**

Across photos, patients revealed fewer fixation clusters, i.e., areas of interest, than controls [*patients:* 11.39 (*SD* = 1*.*72), *controls:* 12.64 (*SD* <sup>=</sup> <sup>1</sup>*.*52), *<sup>F</sup>*group*(*1*,*62*)* <sup>=</sup> <sup>8</sup>*.*97, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*01, <sup>η</sup><sup>2</sup> *p* = 0*.*126, **Figure 3A**]. Patients also made fewer fixations per cluster [*patients:* 4.72 (*SD* = 0*.*52), *controls:* 5.21 (*SD* = 0*.*46), *<sup>F</sup>*group*(*1*,* <sup>63</sup>*)* <sup>=</sup> <sup>16</sup>*.*37, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*01, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*206, **Figure 3B**]. Patients had a longer total fixation time within each cluster [*patients:* 1481.19 ms (*SD* = 248*.*93), *controls:* 1311.07 ms (*SD* = 163*.*78), *<sup>F</sup>*group*(*1*,* <sup>63</sup>*)* <sup>=</sup> <sup>10</sup>*.*66, *<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*01, <sup>η</sup><sup>2</sup> *<sup>p</sup>* = 0*.*145, **Figure 3C**], and

**single subject.** Fixations were weighted by their duration and smoothed with a 2D Gaussian function with σ = 1◦ visual angle (see Materials and Methods). **(B)** Attentional landscape applied to the original photo with highlighted fields indicating areas of most intense attention allocation. **(C–F)** Results from group comparisons of attentional landscapes between patients and controls with blue areas indicating longer and red areas indicating shorter attention allocation in patients compared to controls (*p <* 0*.*001). For photos "Face" and "Coffee Break" attentional landscapes did not differ between groups.

made fewer changes between clusters than controls [*patients:* 31.94 (*SD* = 6*.*61), *controls:* 37.92 (*SD* = 8*.*06), *F*group*(*1*,* <sup>62</sup>*)* = 10*.*69, *p <* 0*.*01, η<sup>2</sup> *<sup>p</sup>* = 0*.*147, **Figure 3D**]. For all these effects, group differences did not differ between photos (no interaction PHOTO × GROUP).

#### **ATTENTIONAL LANDSCAPES**

Areas to which patients allocated attention differently from controls were revealed by comparisons of attentional landscapes, **Figure 5**. The most pronounced group difference was observed for the photo with the highest cognitive complexity score, i.e., "Shopping Street," where patients showed significantly longer and intense attention allocation to central regions of the scene, especially faces of pedestrians, whereas controls allocated attention significantly longer to details in the periphery. While controls fixated peripheral details of the scenes "Volleyball," "Airport," and "Computer Room" significantly longer than patients no group differences were found related to photos "Face" and "Coffee Break."

#### **SIMILARITIES BETWEEN SCANPATHS**

Difference scores of scanpath similarities were quite homogenous within each group of patients and controls, **Figure 6**, but differed significantly between groups [*F*group*(*1*,* <sup>63</sup>*)* = 45*.*2,

negative values indicate a stronger scanpath similarity to the patient group (see Materials and Methods), *p <* 0*.*001 for group differences in each photo. *p <* 0*.*001, η<sup>2</sup> *<sup>p</sup>* = 0*.*418]. However, there was no group differ-

ence between photos (no interaction of PHOTO × GROUP). Discriminant analysis yielded an eigenvalue of 0.826, indicating that between-group variance was larger than within groupvariance [Wilk's Lambda <sup>=</sup> 0.548, <sup>χ</sup><sup>2</sup> *(*6*)* = 36*.*14, *p <* 0*.*001]. Group centroids were quite distinct (*patients:* −0.91, *controls:* 0.88). Eighty-five percentage of the subjects were classified correctly to their group whereas five subjects of either group were misclassified.

# **PHOTO PROPERTIES** *Physical properties*

The physical properties of fixation locations differed between photos with respect to luminance [*F*photo*(*5*,* <sup>62</sup>*)* = 378*.*83, *p <* 0*.*001, χ<sup>2</sup> *<sup>p</sup>* = 0*.*859], luminance contrast [*F*photo*(*5*,*62*)* = 37*.*51,*p <* 0*.*001, η<sup>2</sup> *<sup>p</sup>* = 0*.*377], chromaticity contrast [*F*photo*(*5*,* <sup>62</sup>*)* = 386*.*66, *p <* 0*.*001, η<sup>2</sup> *<sup>p</sup>* = 0*.*862], and static contrast [*F*photo*(*5*,* <sup>62</sup>*)* = 97*.*79, *p <* 0*.*001, η<sup>2</sup> *<sup>p</sup>* = 0*.*612] but we did not observe any group differences for any of these physical properties.

#### *Cognitive complexity and emotional strain*

For more detailed analyses of the effects of photo contents on visual exploration behavior, linear regression analyses of "cognitive complexity" and "emotional strain" on mean fixation time and mean saccadic amplitude were calculated for each subject using the robust-fit function within Matlab® and were then compared between groups. Regression slopes showed that in both groups mean fixation time significantly decreased with higher cognitive complexity [*patients: t(*32*)* = −2*.*7, *p <* 0*.*01, *d* = 0*.*51, *controls*: *t(*32*)* = −7*.*0, *p <* 0*.*001, *d* = 1*.*21] and increased with higher emotional strain [*patients: t(*31*)* = 4*.*1, *p <* 0*.*001, *d* = 0*.*75, *controls: t(*32*)* = 3*.*6, *p <* 0*.*01, *d* = 0*.*63]. With respect to mean saccadic amplitude, we found no relationship to cognitive complexity for neither group [*patients*: *p* = 0*.*13, *controls*: *p* = 0*.*87] but a significant decrease of mean saccadic amplitude with increasing emotional strain of the photos' content in both groups [*patients*: *t(*31*)* = −3*.*8, *p* = 0*.*001, *d* = 0*.*66, *controls*: *t(*32*)* = −2*.*5, *p* = 0*.*02, *d* = 0*.*43]. In all these analyses, there were no differences between patients and controls.

#### *Subjective ratings of the photos*

Patients indicated having experienced more fear during exploration of the photos than controls [*Z(*65*)* = −0*.*354; *p <* 0*.*001] but no differences between groups were found for tension, sadness, joy, aggression, for having had enough exploration time and for having grasped the content of the photo. In patients, the following correlations between eye movement parameters and subjective ratings were observed: larger saccade amplitudes were correlated with a higher rating on having grasped the photo's content (*r* = 0*.*474, *p* = 0*.*006), and with a stronger experience of tension (*r* = 0*.*470, *p* = 0*.*007). In contrast, we did not find any correlations between emotional ratings and eye movement parameters in controls.

#### *Clinical data*

Relations between clinical data (**Table 1**) and eye movement parameters were found solely for the PANSS difference score which was correlated with mean fixation time (*r* = −0*.*44; *p* = 0*.*014) suggesting that a higher predominance of negative symptoms over positive symptoms was related to longer fixation durations.

#### **DISCUSSION**

The present study used an advanced approach to examine free visual exploration of social situations in patients with schizophrenia. Our results of fewer but longer fixations as well as smaller saccades resulting in shorter scanpaths in patients than controls are in line with earlier reports (Phillips and David, 1997, 1998; Williams et al., 1999; Loughland et al., 2002b; Unema et al., 2005; Bestelmeyer et al., 2006; Benson et al., 2007). Reduced fixation frequency could have nonetheless resulted in the same number of areas of interest (i.e., clusters) in patients compared to controls if patients would have made fewer fixations within each cluster. However, cluster analyses revealed that patients fixated fewer areas of interest, made fewer fixations per cluster but had a longer total fixation time within each cluster and also made fewer changes between clusters compared to controls. Following the conception of different information processing levels described by Craik and Lockhart (1972), one explanation for longer fixations in patients suggest that they were more deeply involved in semantic processing of fixated features than controls (Velichkovsky, 2002). According to this conception, preliminary processing stages are concerned with the analysis of physical or sensory features, while later stages are involved in matching the input against stored abstractions from past learning, i.e., pattern recognition and the extraction of meaning (Craik and Lockhart, 1972). Alternatively, longer fixations in patients may reflect a more general problem of disengaging attention. Attentional landscape analyses (**Figure 5**) illustrate those areas with longer (blue) or shorter (red) fixations in patients compared to controls. Differences between groups were most evident for the photo of the highest cognitive complexity (i.e., Shopping Street) in which patients remained longer on features in the center of the scene, whereas controls spent more time on exploring peripheral details. For three other photos, "Volleyball," "Airport," and "Computer Room," we identified areas with more intense attention allocation to peripheral details in controls compared to patients.

From these results one may conclude that patients gathered less visual information from the scenes than controls. However, nearly all patients reported to have had enough time to grasp the content of the scenes suggesting that patients were not aware of their restricted visual exploration behavior. Notably, although mean fixation times were prolonged in patients, fixation durations in both groups were within the normal range of 200–350 ms (*patients:* 269 ms, *controls:* 221 ms) that has been reported in healthy individuals during free exploration. Fixation durations are determined by a range of different factors including information processing, cognitive processes or eye movement preprogramming so that fixation durations during scene exploration may vary from 100 ms to several seconds (Zingale and Kowler, 1987; Groner and Groner, 1989; Pannasch et al., 2011).

More recently, Pannasch et al. (2011) showed that during free scene exploration, fixation durations in healthy individuals are highly under direct control of stimulus information, especially when a focal-processing mode is active. This mode refers to the parafoveal attentional field in which scanning saccades with amplitudes smaller than 5◦ are used for more detailed information processing of a scene's content (Velichkovsky et al., 2005). In contrast, the ambient-processing mode is accompanied by saccades larger than 5◦ and is thought to serve for orientation and information processing about spatial arrangements of undifferentiated visual cues (Trevarthen, 1968; Pannasch and Velichkovsky, 2009). While the focal-processing mode has been attributed to the more ventral fronto-parietal network for visual attention, the ambient-processing mode has been associated with the dorsal fronto-parietal attentional network (Corbetta et al., 2008; Pannasch et al., 2011; Marsman et al., 2012). Following this model (Pannasch and Velichkovsky, 2009; Pannasch et al., 2011), our findings of smaller mean saccadic amplitudes and longer fixation times in patients suggest that during free visual exploration of social scenes patients use a higher percentage of focal-processing than controls. Notably, saccade amplitude was correlated positively with the rating on the extent of having grasped the content of the scene in patients (but not controls), implying that those patients who used smaller saccades realized to some extent that they had missed some aspects of the scene's content. From a brain systems perspective our findings imply that during visual exploration processing patients rely more on the ventral fronto-parietal attentional network but less on the dorsal fronto-parietal attentional network compared with controls. In schizophrenia disturbances of visual information processing along the magnocellular visual pathway and the dorsal visual stream have been related to impaired use of visual gain control and integration resulting in difficulty of modulating neural responses to take advantage of surrounding context during visual perception (Butler et al., 2008). Disturbed bottom-up visual sensorimotor information processing along the dorsal visual stream to parietal association cortex has been also reported from pursuit eye movement studies in schizophrenia using moving targets (Lencer et al., 2010, 2011). Another possible explanation for a more focal-processing mode in patients with schizophrenia comes from visual search studies that have investigated visual exploration under high attentional load, in contrast to free visual exploration used in the present study (Elahipanah et al., 2011a,b). Results from these studies imply narrower visual span size in patients, i.e., the area of the visual field from which information is extracted, especially with moving targets. This visual search dysfunction has also been related to disturbances of the dorsal visual stream (Elahipanah et al., 2011a,b).

#### **THE WHEN AND WHERE QUESTION**

Temporal analyses showed increasing fixation durations and decreasing saccade amplitudes over the exploration time course in both groups reflecting a more "perceptive scanning" with shorter fixations in the first 5 s followed by longer fixations with allocation of attention to specific features of a scene for more "semantic/metacognitive" processing. Notably, this increase of fixation duration over time was larger in patients than controls (GROUP × TIME interaction) underlining the hypothesis that patients got more deeply involved in cognitive processing of details of the scene over the time course than controls.

Scanpath pattern analyses allow for the examination of fixation sequences, thus when and where a subject shifts attention during visual explorations. Here, analyses of scanpath similarities showed that within-group scan patterns were quite similar while scanpath patterns differed significantly between groups. Using scanpath similarities in discriminant analysis resulted in 85% of subjects being classified correctly to either the patient or the control group. Together, these findings underline the hypothesis of a specific exploration behavior in patients that differs considerably from that in controls independently of the content of a scene.

#### **EFFECTS OF PHOTO PROPERTIES ON EXPLORATION BEHAVIOR**

We found clear differences of fixation, saccade and cluster parameters between photos across groups. The scene's cognitive complexity and emotional strain were shown to have a considerable influence on scanpath patterns in both groups. In line with previous reports, fixation durations decreased with increasing complexity and fixation durations increased with increasing emotional strain while saccade amplitudes decreased with increasing emotional strain (Schrammel et al., 2009; Bradley et al., 2011). The latter observation implies a more focal-processing mode when emotional contents were explored. However, despite these obvious differences between photos, the effects of cognitive complexity and emotional strain on fixation duration and saccade amplitude did not differentiate between patients and controls (no interactions of GROUP × PHOTO in ANOVAs). This finding indicates that patients were able to adapt their visual exploration strategies to changes in cognitive complexity and emotional strain of social scenes similarly to controls. Consistent with this observation, patients also did not differ from controls in their ratings on the emotional reactions to the scenes except for anxiety which was more pronounced in patients than controls. In patients, a stronger experience of tension was correlated with larger saccade amplitudes but no further associations between emotional ratings and oculomotor parameters were observed in patients and none in controls.

Physical properties of fixation locations also did not differ between groups underlining the notion that alterations in fixation duration and saccade amplitude in patients occur independently of the stimulus' content but rather reflect a general deficit of visual information processing as has been shown with abstract scenes or stimuli free of emotional content (Kojima et al., 2001; Bestelmeyer et al., 2006; Benson et al., 2012). Alterations in visual exploration strategies have therefore been discussed as possible biological markers for schizophrenia (Kojima et al., 2001; Bestelmeyer et al., 2006; Benson et al., 2012).

#### **LIMITATIONS AND IMPLICATIONS FOR FUTURE STUDIES**

First, we only selected six different photos of the IPT program so that stimuli material is limited. Second, although evaluations on cognitive complexity and emotional strain provided by the authors of the IPT clearly differentiated between photos they may not have been valid enough to detect differences of photo properties on exploration behavior between patients and controls. Third, despite the fact that our results of basic fixation and saccade parameters are consistent with previous reports on visual scanning behavior in chronically ill patients on longterm stable antipsychotic medication (Phillips and David, 1997, 1998; Williams et al., 1999; Loughland et al., 2002b; Bestelmeyer et al., 2006), we cannot exclude the possibility that exploration alterations in patients were influenced by medication. Follow-up studies assessing visual scanpath before and after treatment are required to evaluate the specific effect of antipsychotics on free visual exploration. Fourth, fixation durations in healthy individuals have been linked to the depth of information processing (Craik and Lockhart, 1972; Velichkovsky, 2002). However, it is difficult to determine whether increased fixation durations in patients reflect deeper information processing in patients compared to controls or whether this is due to a general slowing of processing speed (Dickinson et al., 2007) resulting in reduced gathering of visual context information. Future scanpath studies are needed that additionally compare the objective assessment of information taken from a scene between patients and controls. Fifth, special effort was undertaken to ensure similar testing conditions at both study sites, i.e., identical stimulus material and testing procedure, same eye movement recording device. However, as for most multi-site studies, there also were some differences between sites, i.e., stimulus display size was larger in Luebeck than Hamburg. Separate site specific analyses showed significant group differences for all eye movement parameters of interest with some very few exceptions (see supplementary material). In these cases, the significance threshold was not reached in either the Luebeck or the Hamburg sample probably due to lack of power in the site specific samples. In favor of this hypothesis, there was no significant interaction of GROUP × SITE for any parameter of interest combining the samples from both sites. Reducing the variance within the groups of patients and controls resulted in highly significant differences between patients and controls. We are aware of the fact that we might have ignored certain effects of stimulus display size on exploration behavior. Therefore, future studies are needed to examine the effects of stimulus display size, e.g., large vs. small visual field presentation, on free visual exploration in patients with schizophrenia.

# **CONCLUSIONS**

During free visual exploration in daily life situations, patients with schizophrenia seem to generally use a more focal-processing mode with longer fixations on distinct features in the center of a scene in contrast to a more ambient-processing of context information in healthy individuals. The question whether this altered visual exploration strategy in schizophrenia supports a model of how patients require more sensory evidence to inform more uncertain beliefs about the world (Friston et al., 2012) or whether the more focal-processing mode reflects a general deficit of impaired context information processing during visual perception that represents a biomarker for schizophrenia (Butler et al., 2008; Benson et al., 2012) should be subject to future studies. Despite this alteration, patients appear able to adapt their visual exploration strategies to changes in cognitive complexity, physical properties, and emotional strain similarly to healthy participants. It remains an open question whether and how cognitive remediation

# **REFERENCES**


48, 1544–1553. doi: 10.1111/j.1469- 8986.2011.01223.x


programs such as IPT are successful in modifying the focalprocessing mode in patients into a more ambient-processing mode.

# **ACKNOWLEDGMENTS**

We thank Silke Zapf for her assistance in eye movement assessments and analyses at the University of Luebeck. We are also grateful to Volker Roder for supplying the stimulus material. Special thanks go to Rolf Verleger, University of Luebeck, for advice in statistical methods and for his comments on an earlier draft of this manuscript. This work was supported by a Young Investigator Grant to Matthias Nagel by the University of Luebeck (J06-2002).

# **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/ fpsyg.2013.00737/abstract


analysis: using eye fixations as events in functional magnetic resonance imaging to reveal cortical processing during the free exploration of visual images. *Hum. Brain Mapp.* 33, 307–318. doi: 10.1002/hbm.21211


highs of a framework for memory research. *Memory* 10, 405–419. doi: 10.1080/09658210244000234


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 May 2013; accepted: 23 September 2013; published online: 11 October 2013.*

*Citation: Sprenger A, Friedrich M, Nagel M, Schmidt CS, Moritz S and Lencer R (2013) Advanced analysis of free visual exploration patterns in schizophrenia. Front. Psychol. 4:737. doi: 10.3389/fpsyg. 2013.00737*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Sprenger, Friedrich, Nagel, Schmidt, Moritz and Lencer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Combined visual and motor disorganization in patients with schizophrenia

#### *Anne Giersch1 \*, Hélène Wilquin2,3, Rémi L. Capa1 and Yvonne N. Delevoye-Turrell 3,4*

*<sup>1</sup> Department of Psychiatry, INSERM U1114, University Hospital of Strasbourg, Strasbourg, France*

*<sup>2</sup> LPCLS EA3278, Aix Marseille University, Aix-en-Provence, France*

*<sup>3</sup> URECA, University Lille Nord de France, Lille, France*

*<sup>4</sup> CNRS, MESHS, USR 3185, Lille, France*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Elisa C. Dias, Nathan Kline Institute for Psychiatric Research, USA Sebastian Walther, University Hospital of Psychiatry, Switzerland*

#### *\*Correspondence:*

*Anne Giersch, Département de Psychiatrie I, Centre Hospitalier Régional Universitaire de Strasbourg, INSERM U1114, Hôpital Civil, 1 Place de l'Hôpital, F-67091 Strasbourg, Cedex, France e-mail: giersch@unistra.fr*

Cognitive impairments are difficult to relate to clinical symptoms in schizophrenia, partly due to insufficient knowledge on how cognitive impairments interact with one another. Here, we devised a new sequential pointing task requiring both visual organization and motor sequencing. Six circles were presented simultaneously on a touch screen around a fixation point. Participants pointed with the finger each circle one after the other, in synchrony with auditory tones. We used an alternating rhythmic 300/600 ms pattern so that participants performed pairs of taps separated by short intervals of 300 ms. Visual organization was manipulated by using line-segments that grouped the circles two by two, yielding three pairs of connected circles, and three pairs of unconnected circles that belonged to different pairs. This led to three experimental conditions. In the "congruent condition," the pairs of taps had to be executed on circles grouped by connecters. In the "non congruent condition," they were to be executed on the unconnected circles that belonged to different pairs. In a neutral condition, there were no connecters. Twenty two patients with schizophrenia with mild symptoms and 22 control participants performed a series of 30 taps in each condition. Tap pairs were counted as errors when the produced rhythm was inverted (expected rhythm 600*/*300 = 2; inversed rhythm *<*1). Error rates in patients with a high level of clinical disorganization were significantly higher in the non-congruent condition than in the two other conditions, contrary to controls and the remaining patients. The tap-tone asynchrony increased in the presence of connecters in both patient groups, but not in the controls. Patients appeared not to integrate the visual organization during the planning phase of action, leading to a large difficulty during motor execution, especially in those patients revealing difficulties in visual organization. Visual motor tapping tasks may help detect those subgroups of patients.

**Keywords: schizophrenia, motor control, disorganization symptoms, visual organization, visual perception, visual grouping**

# **INTRODUCTION**

Many studies have shown that patients with schizophrenia have a difficulty organizing visual information through space (Place and Gilmore, 1980; Wells and Leventhal, 1984; Silverstein et al., 2000; Must et al., 2004; Uhlhaas et al., 2006; Kurylo et al., 2007; Giersch and Rhein, 2008; van Assche and Giersch, 2011). These deficits are often correlated to the clinical symptom of disorganization, and might even reflect this symptom [review in Silverstein and Keane (2011)]. The possibility to objectify this clinical symptom is of importance since it is related to the loosening of associations that has been proposed as a core feature in schizophrenia (Bleuler, 1911). However, cognitive impairments observed in stabilized patients are of relatively small amplitude. Contrary to brain-damaged patients, the results in stabilized patients with schizophrenia do not lead to all-or-none phenomena, making it difficult to relate cognitive impairments and clinical symptoms, or to derive subgroups of patients according to their cognitive difficulties. Yet, some cognitive deficits might be specific to schizophrenia (Loughland et al., 2002; Lee et al., 2013), and they are often correlated to clinical symptoms (Silverstein and Keane, 2011). We reasoned that neuropsychological explorations may need to be further adapted to the specific difficulties encountered in schizophrenia. Second, a one-to-one correspondence is unlikely to exist between cognitive disorders and clinical symptoms. Cognitive disorders more likely interact with one another. Here, we illustrate this idea by elaborating a new test based on known impairments that might characterize schizophrenia and some of its associated symptoms. We devised a task that requires both (1) the perceptual organization of information, and (2) the motor organization of sequences of action plans through time and space. Our aim was to assess how the impairments in these fields conjointly contribute to the difficulties observed in patients, and whether the visuo-motor coordination constraints imposed by the task would lead to larger impairments than usual in patients with schizophrenia displaying mild symptoms, specifically those characterized by the clinical symptom of disorganization.

#### **ORGANIZATION OF A MOTOR SEQUENCE**

We have shown previously that patients with schizophrenia display difficulties in organizing and smoothly executing motor actions when those actions are sequential (Delevoye-Turrell et al., 2007). Even when performing a task as simple as lifting and gripping an object, our previous results suggest that patients with schizophrenia fail to plan their actions in advance. Contrary to controls, they are not disturbed by a secondary task during the planning phase, but use more resources than healthy volunteers during the motor execution phase of movements (Delevoye-Turrell et al., 2006). Other studies have also reported a difficulty in planning actions in a coherent way (Jogems-Kosterman et al., 2001; Zalla et al., 2006; Grootens et al., 2009). However, to the best of our knowledge, previous studies have not shown any correlation between clinical disorganization and the ability to plan sequences of motor actions. In the present study, we included a manipulation of those visual factors that are known to be affected by clinical disorganization in order to increase the sensitivity of our task to this clinical factor.

#### **ORGANIZATION IN VISUAL PERCEPTION**

Mechanisms helping to organize visual information are diverse, and intervene at different stages of visual organization. Visual information is first extracted locally and decomposed into color, orientation, or luminance information. This information then needs to be correctly integrated or separated in order to derive the shapes of objects and the structure of the considered visual scene. Many automatic grouping mechanisms have been described, e.g., grouping by color, by similarity, by uniform connectedness (Wertheimer, 1923/1950; Palmer and Rock, 1994). These grouping mechanisms allow us to attribute object parts to the same object and to put together different objects that belong to a same group, e.g., the trees in a forest. Our results and those reported in the literature show that mildly symptomatic patients with schizophrenia benefit as much as controls from automatic grouping (Chey and Holzman, 1997; Silverstein et al., 1998; Giersch and Rhein, 2008; van Assche and Giersch, 2011), at least when grouping cues are unambiguous and when they do not lead to spurious grouping (see Kurylo et al., 2007; Silverstein and Keane, 2011 for limits to the benefits of automatic grouping in schizophrenia).

In addition to automatic grouping mechanisms, more integrated cognitive mechanisms provide a function of "re-grouping" of visual elements (van Assche and Giersch, 2011) This mechanism would give the means to the brain to explore a visual scene by foveating different objects successively, without being restrained by the fact that these objects belong to different groups of objects. We have suggested that patients with schizophrenia are in great difficulty when having to "re-group" items in visual perceptual tasks, and shown that this effect is correlated with clinical disorganization (van Assche and Giersch, 2011). To further assess the possible consequences of regrouping deficits in schizophrenia, here we question the possibility that "re-grouping" is also involved in the planning and execution of a sequence of motor actions that requires participants to point toward objects of different groups. Indeed, when pointing to a series of visual targets, these targets can be considered individually (i.e., one by one) or as a series of sub-elements of a greater whole (two adjacent targets composing a group and guiding sequential pointing actions). The grouping of adjacent targets might then affect the planning and fluent execution of the motor sequences.

The possibility that visual information can impact action planning and execution is supported by several arguments, which are presented in the next section.

# **VISUAL PERCEPTION AND ACTION**

A growing number of empirical findings now support the involvement of perception in action processing, and has led to the ideomotor theory (Prinz, 1987; see Nattkemper et al., 2010; Shin et al., 2010, for reviews). According to this theory, the brain would use perceptual representations of action-effects for optimal planning of motor actions. For example, if one plans to manually point toward visual targets, the sensory feedback expected to result from this action is processed in advance of action initiation, i.e., during motor planning. During action execution, participants can then check whether the actual sensory feedback corresponds to what was expected. As a consequence, the correspondence between the action and sensory feedback has a facilitation effect on action, as shown by a number of studies [review in Shin et al. (2010)]. For example, Kunde (2001) used a task during which key presses were followed by tones of varying intensity. The results showed that the correspondence between the intensity of the key presses and the loudness of the tones facilitated the responses. Even when the relationship between the action and its consequence is less straightforward, it can be learned and used to optimize performance (Elsner and Hommel, 2004). In the present study, participants had to execute pairs of pointing actions toward distinct visual targets, and we manipulated the correspondence between the organization of the motor sequence on the one hand and the perceptual organization of visual targets on the other hand. The manipulation of this correspondence was then used in the framework of the ideomotor theory to assess to what extent patients with schizophrenia are able to include visual representations for motor planning and to what extent their difficulties in organizing perceptual information may be revealed in a task that requires in addition the execution of sequences of motor actions.

## **PARADIGM AND PREDICTIONS**

In the present study, we describe a manual-pointing task in which participants were required to tap on 6 circles (the visual targets) that were presented on a tactile screen (**Figure 1**). The participants' task was to tap successively and clockwise on the targets following the rhythm of a series of auditory stimuli. To induce the perception of an auditory structure, we used temporal proximity, which is known as a fundamental principle of organization in auditory perception (Wertheimer, 1938). Two sounds separated by a short interval defined a pair, and successive pairs were separated by a longer interval. Since participants were instructed to tap in synchrony with the sounds, the rhythmic organization of the auditory stimuli induced a rhythmic motor execution of pairs of taps on adjacent circles, which required participants to structure the motor elements through space and through time.

When participants performed the pairs of taps on 6 isolated circles (without any visual grouping cues), the motor outputs were organized in pairs but the perceptual representation of the visual targets was not. A perceptual visual structure was induced by using connectors, which linked adjacent circles together in order to create visual perceptual pairs. This visual structure globally matched the auditory structure, inasmuch both organizations led to the perception of pairs: sound pairs on the one hand, and visual pairs on the other hand. Consequently, the perceptual organization of the circles could hypothetically affect the motor output, depending on the grouping abilities of the participants. Indeed, when pairs of taps were executed on pairs of connected targets the action structure matched the automatic visual organization. On the other hand, when the pairs of taps were to be executed on unconnected targets, participants had to match their pair of taps with "re-grouped" targets (**Figure 1**). We hypothesized that patients would show impairments related to their difficulties in visual "re-grouping," i.e., we expected patients to have difficulties when the task required to point to successive visual targets belonging to different groups. Inasmuch the difficulty related to "re-grouping" is related to clinical disorganization, we also predicted that this pattern of results should be observed predominantly in those patients suffering from clinical disorganization.

# **METHODS PARTICIPANTS**

Twenty-two stabilized outpatients participated in the study. They met the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition)'s criteria for schizophrenia (American Psychiatric Association, 1994). The diagnosis was based on a semi-structured interview (the Mini International Neuropsychiatric Interview, Sheehan et al., 1998), and was established by a senior psychiatrist of the University Psychiatry Department. Patients were 9 women and 13 men (mean age = 40.2 years, *SD* = 8*.*2; schooling years = 11.5, *SD* = 2*.*4).

The patients' performances were compared to those obtained in 22 controls (11 women and 11 men; mean age = 40.9 years, *SD* = 7*.*7; schooling years = 13.2, *SD* = 2). The two groups did not differ in age (*F <* 1). They did differ, however, in the number of schooling years [*F(*1*,* <sup>42</sup>*)* = 6*.*2, *p <* 0*.*05).

Regarding clinical characteristics, mean disease duration was 14.9 years (*SD* = 7*.*6). Symptoms were assessed through the use of the Positive and Negative Syndrome Scale (PANSS, Kay et al., 1987). Patients presented a mean score of 16.8 (*SD* = 5*.*9) for the Positive Subscale, 22.1 (*SD* = 6*.*3) for the Negative Subscale and 36.5 (*SD* = 9*.*8) for the Global Psychopathology Subscale, leading to a total mean score of 75.5 (*SD* = 17*.*2). Overall, these results indicated that the patients were within the normal-to-mild range, and were relatively asymptomatic.

Individual PANSS scores were also converted into a Positive factor (sum of P1, P3, G9 = 8.2, *SD* = 3*.*5), a Negative factor (sum of N1, N2, N3, N4, N5, N6, N7, G7, G13, and G16 = 20.3, *SD* = 6*.*3) and a Disorganization factor (sum of P2, N5, G10, and G11 = 9.6, *SD* = 3*.*1), according to the classification proposed by Lépine et al. (1989).

We were especially interested in the disorganization factor, and thus, we divided patients in two subgroups, one with a disorganization score above or equal to 10 (*N* = 10) and another group with a disorganization score below 10 (*N* = 12). This cutoff was set in order to divide the group of patients in two (this could not be totally achieved, due to the fact that 4 patients had a disorganization score of 9). These two groups differed in the number of schooling years [10.4 *SD* 1.6 for the group with a high disorganization score vs. 12.5 *SD* 2.5 for the group with a low disorganization score, *F(*1*,* <sup>20</sup>*)* = 5*.*1, *p <* 0*.*05]. We thus, divided the group of controls according to their schooling years also, in order to evaluate whether there was an impact of schooling years on the performance patterns. The 10 controls with the lowest number of schooling years had a similar education level as the patients from the group with a high disorganization score [10.4 in patients vs. 11.2 in controls, *F(*1*,* <sup>18</sup>*)* = 2*.*2, *n.s*.]. It is to note that the two subgroups of controls revealed very similar performances profiles, indicating that the education level did not impact result patterns. Our findings were in fact similar whether we divided the group of controls or not. Thus, for the sake of simplicity, the data presented in the following section corresponds to the means averaged across the two subgroups of controls.

All patients were taking medication, and doses were converted in chlorpromazine equivalents (mean dose = 272 mg, *SD* = 237). Patients received a neuroleptic treatment, either typical (*N* = 7) or atypical (*N* = 15). Six patients were also administered with an antiparkinsonian treatment.

All participants gave written informed consent prior the beginning of the study, consistently with the Declaration of Helsinki's recommendations. This project was approved by the local ethics committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale d'Alsace IV). Any history of neurological disorder (meningitis, brain injury, cerebrovascular accident), generalized anaesthesia within the past 3 months, or recent drug abuse, was considered as exclusion factors. Participants treated by benzodiazepines were also discarded from this study. In addition, four participants were excluded from the initial group of 24 participants because they did not follow instructions (2 patients and 2 controls).

#### **EQUIPMENT AND STIMULI**

Participants were seated comfortably on a chair in front of a touch screen (Elo Touch, 23 × 36 × 30 cm), which was placed on a narrow support at knee-height. The participants' task was to produce sequential-pointing movements to visual targets, with their dominant index finger, in synchrony with a series of beeps produced by the computer. The visual targets consisted in 6 equidistant outlined black circles arranged in the form of a hexagon. Participants were asked to point each circle (diameter 1.2◦), one after the other, starting with the bottom-right circle (**Figure 1**). Participants were required to point each target (touching the screen) following a clockwise direction until they had reached the end of the trial duration (13 s). The auditory rhythmic sequence of beeps was created using Audacity software. Three different visual images were created using CorelDraw software. One was with the 6 circles but without connecters. The two others were with connecters, i.e., with line-segments linking neighboring circles by pairs (**Figure 1**). There were two possible locations for the connecters and both were used an equivalent number of times. Participants were instructed to be as accurate as possible both spatially and temporally. The connecters being presented right from the start of each trial, they emphasized a first grouping structure to the task, based on the perceptual principles of visual grouping.

# **EXPERIMENTAL CONDITIONS**

The participants' task was to tap each visual target in synchrony with the sounds emitted by the computer. Alternating rhythmic patterns were used, i.e., inter-stimulus intervals of 300 and 600 ms. These alternating intervals imposed an auditory grouping and emphasized the need to re-group motor actions. Sounds separated by a short interval (300 ms in our paradigm) induce the perception of grouped pairs of sounds because they are close in time and separated from other pairs by a longer interval (600 ms in our paradigm). Since participants were required to tap in synchrony with the sounds, this led to the execution of pairs of taps close together in time, the two taps of a pair being ideally separated by a time interval of 300 ms, and successive pairs of taps being ideally separated by a time interval of 600 ms. The 300/600 ms intervals were chosen because when combined they are close to the preferred tempo of Inter Response Intervals of 500 ms which corresponds to the spontaneous tempo reported in the tapping literature (Repp, 2005). In addition, we knew from previous studies (Ameller et al., 2011; Turgeon et al., 2012) that patients with schizophrenia are able to perform the task at this speed. Finally, the contrast between 300 and 600 ms intervals was strong enough to yield the unambiguous perception of rhythmic grouping. It should be noted that this auditory grouping manipulation has already been applied to taps executed in one unique location; the originality of our paradigm is to ask participants to tap on distinct visual targets spread out throughout the action space and thus, requiring participants to re-group motor elements not only through time but also through space.

The 300/600 ms rhythmic pattern was used both with stimuli without and with connecters, leading to three conditions:


#### **EXPERIMENTAL DESIGN**

The experiment lasted about 1 h. First, all participants performed a familiarisation phase in which the different rhythms and images were presented. Participants were trained to synchronize their pointing taps with a trial requiring no visual grouping and equivalent time intervals between tones before being trained with alternating rhythms.

Each trial consisted in a *listening phase* (≈4.5 s), a *waiting phase* without sound, which gave time for the participants to get ready to point to the first circle [they prepared their finger just above the first circle (bottom right)] (≈3 s) and a *test phase* of 30 taps (≈13 s). Each participant performed a sum-total of 48 trials.

Participants started with the Neutral condition (without connecters) and performed 2 blocks of 4 trials each. This was followed by the Visual Grouping conditions. Each congruent and noncongruent condition included 4 blocks of 5 trials each (2 blocks with the image A and 2 with the image B, see **Figure 1**). These 8 blocks were randomized.

# **STATISTICAL ANALYSIS**

For each subject and each trial, we measured both the Inter Response Interval and the asynchrony between the tap and the sound. The spatial location of the tap was also measured allowing us to distinguish those taps directed toward connected and unconnected targets. We were in this way able to check whether participants followed the rhythm correctly.

In a first analysis, we measured the Inter Response Interval (IRI in ms), which is the time interval between the onsets of two successive taps produced by the participants, and which is a parameter that is commonly used in the tapping literature (Repp, 2005). From this IRI measure, we then calculated the mean ratio across trials for each participant and under each condition. This ratio referred to the participants' capacity to maintain the metric organization of the auditory rhythmic sequence (i.e., alternance of short and long intervals lasting 300 and 600 ms, respectively) and was calculated following the equation: IRIn/IRIn-1, where n refers to the long interval (600 ms), and n-1 to the short interval (300 ms). The ratio should thus be equal to 2. When this ratio was inverted (*<*1), we considered the trial as bad. Following this rule, we then calculated for each subject and under each experimental condition, the percentage of trials considered as bad (the number of bad trials divided by the total number of trials in a condition), a calculation that we will refer to in the following as "rate of errors."

In a second analysis we aimed to evaluate the use of anticipatory mechanisms. We used here another typical parameter that is used in tapping experiments, i.e., the tap-tone synchronization accuracy (applied on correct trials only). To that aim, we calculated the asynchrony, which was taken as the time interval between the peak force of each tap and the start of the nearest tone (in ms).

A crucial aspect of such a sensorimotor synchronization task is the predictability of the external beep, which arises from its regular recurrence. It is this predictability that allows good synchronization between tap and sound. This is specifically the feature that distinguishes sensorimotor synchronization from a simple reaction time task, for which the response is made as quickly as possible after the sound and thus, is characterized by a delay *>*180 ms. In rhythmic tapping, the ability to anticipate the beep occurrence leads to asynchronies between tap and sound that are close to zero or even negative in healthy volunteers. Tap-tone asynchronies can thus be used to evaluate the quality of predictive timing. In addition, in the present study, our task imposed an organization of the taps by pairs, and this means that predictive timing might differ according to the order of the taps. We thus, distinguished the tap-tone asynchronies for the 1st and 2nd tap of a pair.

We used Analyses of Variance (ANOVA) with repeated measures on both error rates (rates of trials with an inverted ratio) and tap-sound asynchronies, with Group as a between-group factor and Experimental Condition as a within group factor (neutral, congruent and non-congruent conditions). For tap-sound asynchronies, an additional within-group factor was the tap Order within the sequences of two taps (first vs. second). Significant interactions were decomposed by means of Tukey HSD *post-hoc* tests and Sub-Analyses of Variance.

# **RESULTS**

#### **ERROR RATES**

The most striking effect was the large amount of errors characterizing motor performances in the patients with schizophrenia. The interval ratio of patients was inverted, i.e., below 1, in 18% of the trials (averaged across conditions), which was significantly higher than the rate of inversions of 5% observed in the controls [*F(*1*,* <sup>42</sup>*)* = 11*.*1, *p <* 0*.*005].

The error rates were highly variable in patients, but were unusually large in half of the patients in the non-congruent condition. In this condition, 11 patients (vs. 0 control) revealed an error rate above 19%, i.e., 3*SD* above that measured in the controls (4.7% *SD* 4.7) (see distribution in **Figure 2**). The other half

of the patients (*N* = 11) showed an error rate below 10% which was within the normal range (**Figure 2**). The two subgroups of patients differed significantly only on the score of clinical disorganization [*F(*1*,* <sup>20</sup>*)* = 4*.*7, *p <* 0*.*05]. These results were confirmed by the presence of significant correlations between tapping performances and clinical scores in the patients group as a whole. More specifically, a significant positive correlation was revealed between the score of clinical disorganization and the rate of errors in the non-congruent condition (*r* = 0*.*44, *N* = 22, *p <* 0*.*05). There were no other significant correlations with clinical scores, neither with the sub-scales of the PANSS nor with the positive or negative Lepine factors. All further analyses on motor performance scores were hence conducted taking into account the disorganization subgroups.

It is to be noted that when the group of patients was not divided, there was only a tendency toward a significant interaction between groups and experimental conditions [*F(*2*,* <sup>84</sup>*)* = 2*.*6, *p* = 0*.*076], which might be due to the heterogeneity of performance in the patients. Nevertheless, the Tukey HSD *post-hoc* analysis showed that in patients errors were more frequent in the non-congruent (22.1%) than in the congruent condition (14.8%), *p <* 0*.*05. This effect was not observed in controls (4.7 vs. 3.0%, *p >* 0*.*9).

In the following, we compared the performance levels across three groups of individuals (controls; patients with a high disorganization score; patients with a low disorganization score) as a function of the experimental conditions.

There was a significant interaction between groups and experimental conditions for the rate of errors [*F(*4*,* <sup>82</sup>*)* = 4*.*8, *p <* 0*.*005]. The HSD Tukey *post-hoc* analysis showed that patients with a high score of disorganization had a higher error rate in the non-congruent condition (31.6%) than in the two other conditions (18.2% in the neutral condition, *p <* 0*.*005; 19.7% in the congruent visual grouping condition, *p <* 0*.*05, **Figure 3**). These results were confirmed by means of sub-analyses that were conducted in each group and aimed at comparing performance levels across conditions. The sub-analyses not only confirmed the HSD Tukey *post-hoc* test but also revealed some effects in controls and in patients with a low disorganization score. The effect of experimental conditions was significant in the group with a low disorganization score [*F(*2*,* <sup>22</sup>*)* = 3*.*7, *p <* 0*.*05], due to a lower error rate in the congruent than in the neutral condition (10.6 vs. 17.1%, *p <* 0*.*05 in the HSD Tukey *post-hoc* analysis following the sub-analysis in patients with a low disorganization score). In the control group, the effect of experimental conditions tended toward significance only, *F(*2*,* <sup>42</sup>*)* = 3, *p* = 0*.*06. As in the group with a low disorganization score, the error rate tended to be lower in the congruent than in the neutral condition (3 vs. 7.3%, *p* = 0*.*05).

Sub-analyses were also used to compare performance levels across groups of participants in each experimental condition. In the non-congruent condition, the group effect was significant [*F(*2*,* <sup>41</sup>*)* = 11*.*8, *p <* 0*.*001]. The HSD *post-hoc* analysis showed that patients with a high disorganization score made more errors than both controls (*p <* 0*.*001) and patients with a low disorganization score (*p <* 0*.*05). These results consistently confirm the performance decrement in the non-congruent condition in disorganized patients relative to the two other groups.

In the congruent condition, the group effect was significant [*F(*2*,* <sup>41</sup>*)* = 5*.*6, *p <* 0*.*01] and the HSD Tukey *post-hoc* analysis showed that patients with a high disorganization score made more errors than controls (*p <* 0*.*01). The group effect did not reach significance level in the neutral condition, *F(*2*,* <sup>41</sup>*)* = 2*.*7, *p* = 0*.*076.

#### **TAP-SOUND ASYNCHRONIES**

The recording of tap-sound asynchronies was aimed at checking to which extent participants benefited from the sound regularities and planned their tap in advance. Results on this parameter were identical in patients with a high and low score of disorganization, as illustrated in **Figure 4** and thus, data was averaged across subgroups of patients.

A first analysis of variance conducted on tap-sound asynchronies showed a triple significant interaction between group (patients vs. controls), the tap order (first vs. second), and the experimental conditions (neutral, congruent, non-congruent), *F(*2*,* <sup>84</sup>*)* = 4*.*3, *p <* 0*.*05. In all participants, the tap-tone asynchrony was longer for the second than for the first tap, reflecting the rhythmic structure of the motor sequences. In patients, the tap-tone asynchrony increased by 41 ms between the first and the second tap, *F(*1*,* <sup>21</sup>*)* = 23*.*6, *p <* 0*.*001. In controls, it increased by a similar amount, i.e., by 41 ms, *F(*1*,*21*)* = 82, *p <* 0*.*001. In the following sections, we decompose the 3rd level interaction by distinguishing the first vs. second tap of a sequence of two.

#### *First tap*

For the first tap, tap-sound asynchronies were larger in patients than in controls [32 vs. 16 ms, *F(*1*,* <sup>42</sup>*)* = 5*.*7, *p <* 0*.*05]. However, this effect did not interact with the experimental conditions (*F <* 1).

#### *Second tap*

For the second tap, we observed an interaction between group and experimental conditions [*F(*2*,* <sup>84</sup>*)* = 10*.*3, *p <* 0*.*001). Decomposing this interaction showed opposite effects of the presence vs. absence of connecters in the patients and in the controls.

In controls, the tap-sound asynchronies were shorter in the presence of connecters than in the absence of connecters. This was reflected in a significant effect of experimental conditions [*F(*2*,* <sup>42</sup>*)* = 3*.*8, *p <* 0*.*05]. The effects were not strong enough to yield significant effects in the HSD Tukey *post-hoc* analysis, but sub-analyses showed that tap-sound asynchronies were shorter in

**FIGURE 4 | Mean tap-tone asynchrony (with standard error of the means, SEM) on the 2d tap, for correct trials averaged across participants in each group (patients with a low and high level of disorganization, and controls) as a function of the experimental condition (neutral, non-congruent, congruent).**

the non-congruent than in the neutral condition [53 vs. 66 ms, *F(*1*,* <sup>21</sup>*)* = 4*.*4, *p <* 0*.*05], and that they tended to be shorter in the congruent than in the neutral condition [54 vs. 66 ms, *F(*1*,* <sup>21</sup>*)* = 4*.*2, *p* = 0*.*051].

These effects were reversed in patients. The tap-sound asynchronies were longer in the presence of connecters than in the absence of connecters. There was a significant effect of experimental conditions [*F(*2*,* <sup>42</sup>*)* = 6*.*6, *p <* 0*.*005], and the HSD Tukey *post-hoc* analysis showed that both the congruent and non-congruent conditions yielded longer tap-sound asynchronies than the neutral condition (77 ms in the congruent and 82 ms in the non-congruent conditions, vs. 60 ms in the neutral condition, *p <* 0*.*05 and *p <* 0*.*005, respectively).

It is to be noted that tap-sound asynchronies were no larger in the bad trials, i.e., in the inverted trials (data not shown).

#### **CORRELATIONS BETWEEN PERFORMANCE AND TREATMENT**

There were no significant correlations between performance and the dosage of neuroleptics (in chlorpromazine equivalent). In addition, the number of patients treated with typical neuroleptics was equivalent in both subgroups of patients (4 and 3). Finally, the dosage of neuroleptics did not differ significantly between the two subgroups of patients [*F(*1*,* <sup>20</sup>*)* = 1*.*3, *p >* 0*.*25]: 219 vs. 335 mg in the groups with the low and the high scores of disorganization, respectively. The non-significant differences in dosage were due to one patient having a higher dose than all others (1200 mg). The dosage was strictly identical between groups when this patient was excluded from the analyses: 219 vs. 239 mg. In the present paper, the subject was included in all analyses because when excluded result patterns were identical in all points. For example, the error rates remained abnormally high in the patient group for the non-congruent condition (30% without vs. 31.6% with this subject).

Correlations with clinical disorganization scores have been described above. There were no other significant correlations with clinical scores.

# **DISCUSSION**

Our experimental design was aimed to explore the impact of visual grouping on the execution of a sequence of motor actions. In the present study, the participants' task was to produce pairs of taps on a touch screen in synchrony with alternated auditory rhythms. The results indicated that the connecters significantly impacted the planning and execution of the tap sequences both in controls and patients. More specifically, in healthy participants, there was a significant effect of the presence of connecters on tap-tone asynchrony, with a slight improvement in timing performances indexed through the reduction of tap-tone asynchronies. The amplitude of this effect was small in controls, but it contrasted with a large impairment in patients with schizophrenia. Indeed, not only the presence of connecters did not help but it was detrimental for all patients, whatever the degree of clinical disorganization. In addition, there was a specific difficulty in the non-congruent condition in those patients characterized by high scores of clinical disorganization. Overall, we suggest this easy non-verbal task may provide the means to gain a better understanding of the relationship between the disorganization symptoms and the use of cognitive representations for the planning and execution of structured motor behavior.

Several trivial explanations can be discarded. The results reported in the present study cannot be attributed to a generalized deficit in schizophrenia. Indeed, a generalized deficit results in amplified or reduced effects, as compared to those observed in controls because, e.g., patients are more sensitive to task difficulty than controls, irrespective of the mechanisms involved (Knight and Silverstein, 2001). A generalized deficit can hardly explain the opposite effects that we reported for tap-tone asynchronies in patients and in controls. It can neither predict the large congruency effects that we report in the disorganized patients, since such effects are in fact absent in the control group. A second point of importance is that we were able to eliminate an explanation in terms of neuroleptic treatment, inasmuch as both subgroups of patients had a similar treatment dosage and treatment type (typical vs. atypical) even though they differed markedly on performance profiles. Finally, the timing parameters that were computed confirmed that overall the patients were able to perform the task adequately: the tap-sound asynchronies were way below the latencies expected for simple reaction times (*<*200 ms), confirming that all three groups anticipated the timing of occurrence of the sounds to some extent. All participants used the auditory information to plan the sequential actions and thus, to perform in a predictive fashion following the auditory rhythm as instructed. This pattern of results was observed even in those patients who were characterized by high error rates (i.e., those trials produced with an inversed ratio), suggesting that difficulty was not in the motor task in itself.

Overall, these findings suggest that our paradigm may help to uncover a more selective impairment than a general planning/sequencing deficit. In the following, we decompose the different mechanisms involved in controls, and then suggest an interpretation of the impairments in relation to schizophrenia.

In the present study, because of the timing constraints, the task required participants to organize the different visual elements of action within pairs in order to prepare sequences of taps in a predictive manner. Tap-sound asynchronies indexed anticipation and planning, and tap-sound asynchronies for the second tap were shorter when connecters were present rather than absent. These effects suggest that connecters help controls to plan their sequence of taps more efficiently. This may be the case because under such conditions there is an organization correspondence between visual space and motor representations. This would be similar to congruency effects between action and sensory feedbacks, as described in the literature within the ideomotor theory (Prinz, 1987; Hommel et al., 2001; Shin et al., 2010). This would allow participants to know in advance where to expect connecters, in relation to the visual targets they need to point to. Even if this knowledge is not conscious, it would help participants to benefit from the congruency between expectation and true sensory feedback and thus, help to improve performance accuracy throughout the trial. Moreover, this might represent a possible mechanism allowing participants to avoid a deleterious effect of the non-congruent condition. It is indeed striking that control participants were as efficient in the congruent and in the noncongruent conditions, both for rates of errors and for tap-sound asynchronies. This is in marked contrast with what is observed in visual perception tasks for which a performance advantage is very commonly observed when the task is directed to a unique object (or group of objects) compared to that seen when the task is directed toward distinct, non-grouped objects (Duncan, 1984; Egly et al., 1994; Beck and Palmer, 2002; Palmer and Beck, 2007; van Assche et al., 2012). But it is the case that in the present task, participants benefited from the regularity of the organization to guide motor outputs. The most likely explanation is that healthy participants learned the correspondence between the tap pairs and the visual organization of the visual targets. In the non-congruent condition, this might have been facilitated by the fact that healthy participants are able to build a representation that binds unconnected targets together (Giersch and Rhein, 2008; van Assche et al., 2012). Inasmuch participants "regrouped" unconnected targets, they quickly learned to expect an absence of connecters between those visual targets that were to be tapped together. This mechanism may have helped them overcome a possible motor-perceptual correspondence difficulty in the non-congruent condition. It would account for the advantage provided by connecters in the controls, and explain the lack of congruency effect.

In contrast to controls, the patients' tap-sound asynchronies increased when connecters were present. These results suggest that connecters further impaired the patients' ability to plan efficiently their tap sequence. It is the case that patients have been shown to be selectively affected in tasks requiring the anticipatory planning of sequences of actions (Jogems-Kosterman et al., 2001; Delevoye-Turrell et al., 2007; Grootens et al., 2009), and our results are consistent with this literature. Even though the taptone asynchronies were small and indicated that patients were planning their tapping sequence in a predictive manner up to a certain extent, asynchronies recorded in patients were much longer than those observed in controls in the presence of connecters. This suggests a difficulty in the patients to integrate the visual organization within the motor plan of the tap sequences. As a consequence patients did not benefit from the organization correspondence between visual information and motor representations. Moreover, if the visual organization is not integrated during the motor planning phase, visual information would then need to be processed during motor execution, which would increase furthermore the cognitive load of the motor task. This might have led patients to be more sensitive to the effects of congruency. The lack of congruency effects on tap-tone asynchrony certainly suggests that the conflict arising in the non-congruent condition affects motor execution but not motor planning.

The impact of the non-congruent condition is similar to those impairments previously reported in visual perceptual tasks (Giersch and Rhein, 2008; van Assche and Giersch, 2011; Giersch et al., 2012), which showed that patients have a difficulty with pairs of objects when these objects are not only individualized but belong to different groups. The present results suggest that this specific cognitive impairment in the perceptual domain directly impacts performance levels in the motor domain. The difficulty in the non-congruent condition is thus observed in those patients having a difficulty with unconnected figures belonging to different groups, i.e., patients with a high score of disorganization. It is possible that because patients do not integrate visual organization within their motor plan, they do not use connecters as landmarks to stabilize and/or improve their tapping accuracy. This would lead patients to switch more easily from the noncongruent to a congruent tapping mode, leading to an inversion of the alternated rhythm. It is this combination of planning and visual organization deficits that might account here for the large impairments reported in disorganized patients specifically in the non-congruent condition.

The present findings now need to be confirmed in larger groups of patients. However, our results suggest that the approach to combine a need to organize factors of different sensory modalities within a simple motor task may be a good lead in the development of more sensitive tools for the exploration of cognitive disorders and their relationship with clinical symptoms. It is noteworthy that the impairment on tap-tone asynchrony concerns all patients, contrary to the effect of the non-congruent condition on error rates. This suggests that the two variables tap into different mechanisms that contribute differently to clinical symptoms. As we have seen above, the lengthening of the tap-tone asynchrony suggests a difficulty in action planning, and this might be of significance at a clinical level. If patients cannot plan efficiently their actions, this means a loss of control on their behavior. Such a loss of control might pave the way for delusions of control, i.e., for the belief that one's own actions are controlled by an external force (Wilquin and Delevoye-Turrell, 2012). The impairment in planning described in our study might add up to those deficits already proposed to be involved in the emergence of such delusions (Frith, 2005). The effect of the non-congruent condition, on the other hand, appears to be rather related to known visual organization impairments. What remains to be seen, however, is to what extent the clear-cut performance differences observed between the two subgroups of patients correspond to true observable clinical differences. It is the case that clinical disorganization is evaluated on a continuum, whereas the present patterns of results suggest a subgroup division. Hence, further exploration is required to test the clinical validity of the subdivision that was proposed here.

Future studies will also reveal whether the present exploration complements existing measures of executive and motor performances related to disorganization in schizophrenia, e.g., neurological soft signs for motor control (Mohr et al., 1996; Tosato and Dazzan, 2005; Mechri et al., 2009) and the tower of London for executive functions (Greenwood et al., 2011). The tower of London requires participants to move discs from one peg to another by following a number of predetermined rules. All these tasks require both motor coordination and motor planning. What our paradigm brings in addition is the possibility to distinguish planning from execution, and to isolate a condition that represents a peculiar difficulty for patients: regrouping visual items for fluent motor execution. This difficulty makes clinical sense inasmuch that it echoes the loosening of association originally described by Bleuler. Furthermore, it indicates that some associations are more difficult than others to performance for patients suffering from schizophrenia. In the present paradigm, associating items already bound together through automatic grouping mechanisms is not as difficult as associating items that are not only individualized but that belong to different groups of objects. This condition might be specifically difficult for disorganized patients.

This cognitive deficit for motor fluency might have been missed until now because pointing toward objects seems trivial. Indeed, in the present study, the control participants were very efficient in planning and executing rhythmic sequences of motor pointing actions toward unrelated spatial elements that belonged to different groups of objects. This ability is very useful in our multimedia daily experiences. Indeed, the visual environment is often crowded of distractors, and healthy individuals frequently plan pointing actions between objects, whatever their relative locations in space and in time (imagine tapping on the touchpad of your telephone). What usually seems so easy to us might, however, be much more difficult to perform for patients suffering from schizophrenia, and especially those patients characterized with high scores of clinical disorganization. What remains to be seen is whether the impairments observed here in a manual

# **REFERENCES**


pointing task can also explain the patients' difficulties navigating visually through the visual environment, i.e., with ocular rather than manual movements. It has often been shown that patients have a reduced span of exploration when viewing a face or an abstract picture (Gaebel et al., 1987; Kojima et al., 1990; Gordon et al., 1992; Phillips and David, 1997; Loughland et al., 2002; Obayashi et al., 2003; Minassian et al., 2005; Delerue et al., 2010; Delerue and Boucart, 2012). Patients can miss part of a picture. It might be the case that impaired spontaneous exploration of natural scenes is due at least in part to a difficulty in planning and executing sequential eye movements between unrelated parts of a visual image.

# **ACKNOWLEDGMENTS**

The research reviewed in this manuscript was supported by the French National Institute for Health and Medical Research (INSERM), the Centre Hospitalier Régional Universitaire of Strasbourg, and the French National Research Agency (ANR, n◦ ANR-10-BLAN-1903-01).

do not preserve automatic grouping when mentally re-grouping figures: shedding light on an ignored difficulty. *Front. Psychol.* 3:274. doi: 10.3389/fpsyg.2012.00274


schizophrenia. *Schizophr. Bull.* 13, 261–276. doi: 10.1093/schbul/ 13.2.261


(Amsterdam: Excerpta Medica), Abstract no. 3232.


organization, the disorganization syndrome, and context processing in chronic schizophrenia. *Schizophr. Res.* 43, 11–20. doi: 10.1016/S0920- 9964(99)00180-2


and Kegan Paul), 71–88. doi: 10.1037/11496-005


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 01 February 2013; accepted: 23 August 2013; published online: 18 September 2013.*

*Citation: Giersch A, Wilquin H, Capa RL and Delevoye-Turrell YN (2013) Combined visual and motor disorganization in patients with schizophrenia. Front. Psychol. 4:620. doi: 10.3389/fpsyg. 2013.00620*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Giersch, Wilquin, Capa and Delevoye-Turrell. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Abnormal task modulation of oscillatory neural activity in schizophrenia

#### *Elisa C. Dias <sup>1</sup> \*, Stephan Bickel 1,2, Michael L. Epstein1, Pejman Sehatpour 1,3 and Daniel C. Javitt 1,3*

*<sup>1</sup> Center for Schizophrenia Research, Nathan Kline Institute for Psychiatric Research, Orangeburg, NY, USA*

*<sup>2</sup> Department of Neurology, Albert Einstein College of Medicine, Bronx, NY, USA*

*<sup>3</sup> Division of Experimental Therapeutics, Department of Psychiatry, Columbia College of Physicians and Surgeons, New York, NY, USA*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Oliver Pogarell, University of Munich, Germany Thomas Whitford, University of New South Wales, Australia*

#### *\*Correspondence:*

*Elisa C. Dias, Center for Schizophrenia Research, The Nathan Kline Institute for Psychiatric Research, 140 Old Orangeburg Rd., Bldg. 35, Orangeburg, NY 10962, USA e-mail: dias@nki.rfmh.org*

Schizophrenia patients have deficits in cognitive function that are a core feature of the disorder. AX-CPT is commonly used to study cognition in schizophrenia, and patients have characteristic pattern of behavioral and ERP response. In AX-CPT subjects respond when a flashed cue "A" is followed by a target "X," ignoring other letter combinations. Patients show reduced hit rate to "go" trials, and increased false alarms to sequences that require inhibition of a prepotent response. EEG recordings show reduced sensory (P1/N1), as well as later cognitive components (N2, P3, CNV). Behavioral deficits correlate most strongly with sensory dysfunction. Oscillatory analyses provide critical information regarding sensory/cognitive processing over and above standard ERP analyses. Recent analyses of induced oscillatory activity in single trials during AX-CPT in healthy volunteers showed characteristic response patterns in theta, alpha, and beta frequencies tied to specific sensory and cognitive processes. Alpha and beta modulated during the trials and beta modulation over the frontal cortex correlated with reaction time. In this study, EEG data was obtained from 18 schizophrenia patients and 13 controls during AX-CPT performance, and single trial decomposition of the signal yielded power in the target wavelengths. Significant task-related event-related desynchronization (ERD) was observed in both alpha and beta frequency bands over parieto-occipital cortex related to sensory encoding of the cue. This modulation was reduced in patients for beta, but not for alpha. In addition, significant beta ERD was observed over motor cortex, related to motor preparation for the response, and was also reduced in patients. These findings demonstrate impaired dynamic modulation of beta frequency rhythms in schizophrenia, and suggest that failures of oscillatory activity may underlie impaired sensory information processing in schizophrenia that in turn contributes to cognitive deficits.

#### **Keywords: schizophrenia, oscillations, alpha, beta, AX-CPT, cognitive, motor preparation**

# **INTRODUCTION**

Schizophrenia is a severe brain disorder associated with widespread cognitive deficits that represent a core feature of the disorder. Although deficits have been studied most extensively with regard to higher order constructs such as attention, working memory, or executive processing, deficits are documented as well in basic sensory processing. For example, in the auditory system, patients show deficits in pitch processing and mismatch negativity which correlate with dysfunction in higher cognitive abilities (Javitt, 2009; Leitman et al., 2010; Kantrowitz et al., 2013). In the visual system, patients show impairment on non-linear gain mechanisms, leading to reduced contrast sensitivity particularly within the magnocellular visual system (Butler et al., 2001), as demonstrated using behavioral, neurophysiological, and fMRIbased approaches (Butler et al., 2005; Martinez et al., 2008, 2012). These deficits in sensory processing suggest that patients with schizophrenia are impaired not only in how they process information, but how they experience the world (e.g., Leitman et al., 2007; Butler et al., 2009).

One of the most informative tasks for evaluation of visual sensory-level contributions to higher order cognitive impairment in schizophrenia is the AX-type continuous performance task (AX-CPT; Rosvold et al., 1956). In the AX-CPT, subjects attend to a sequence of individually presented letters and must respond whenever they view a letter "A" followed by a letter "X," while ignoring all other sequences. In the most widely used version of the task ("AX-70" version) the majority of trials consist of AX sequences, so that the critical challenge of the task becomes to withhold response to all other sequences. Patients show reliable deficits both in detection of AX-sequences and rejection of sequences in which an invalid cue (letter other than "A," collectively termed "B"), is followed by a target ("BX" trials), leading to a reduction in a signal-detection measure termed d'-context, which is calculated based upon relative percentages of errors for sequences AX (misses) and BX (false alarms; Cohen et al., 1999; Barch et al., 2001; Dias et al., 2003, 2011; MacDonald et al., 2003; Bickel et al., 2012).

In contrast to their increased error rates on BX trials, patients have relatively intact ability to withhold response to non-target stimuli even if they are preceded by a valid cue ("AY" sequences, in which "Y" represents letters other than "X"), suggesting that performance deficits in this task in schizophrenia primarily reflect impaired ability to utilize cue information to determine whether to press or not press in response to a correct target ("X") stimulus. By contrast, when given a stimulus which by itself requires a non-response (i.e., "Y" target), patients show preserved response inhibition (Cohen et al., 1999; Barch et al., 2001; Dias et al., 2011), showing relatively intact levels of task engagement.

Initial theories of AX-CPT dysfunction in schizophrenia suggested that impairments were due to inability to maintain information during the cue-target interval, in keeping with dopaminergic theories of schizophrenia (Cohen and Servan-Schreiber, 1992). Subsequent studies, however, showed similar deficits both at both short and long ISI, suggesting that impairments resulted primarily from decoding of cue information itself, rather than maintenance over time (Lee and Park, 2005, 2006; Javitt et al., 2007). A similar pattern of deficit is observed in healthy volunteers during acute challenge with the N-methyl-D-aspartate receptor (NMDAR) antagonist ketamine, suggesting that performance deficits on this task may reflect underlying impairment of NMDAR-dependent circuits (Umbricht et al., 2000).

Early theories of AX-CPT dysfunction in schizophrenia also suggested that deficits reflected primarily impairments in ability to withhold responses when the global prepotency, based upon relative probability of AX and BX trials, was to respond. In contrast, we have more recently observed that patients show greatest deficits, compared to controls, in versions of the task when AX probability is low, so the global prepotency is to not respond to a target ("X") stimulus (Dias et al., 2011). The availability of parametric versions of this task, in which the proportions of the different types of trial are systematically varied creating different expectancies of response, permits detailed assessment of mechanisms underlying performance deficits in schizophrenia, using both fMRI and event-related potential techniques.

In fMRI versions of the task, greatest attention has been paid to activation of frontal areas. In these studies, consistent deficits are observed in dorsolateral prefrontal cortex (DLPFC) and anterior cingulate cortex (ACC; Barch et al., 2003; MacDonald and Carter, 2003; Perlstein et al., 2003). However, a more recent meta-analysis shows that in controls there is also activation of posterior visual areas, such as the middle occipital gyrus, that is not observed in schizophrenia patients, suggesting sensory-level disturbance as well (Minzenberg et al., 2009).

The sensory involvement in AX-CPT performance deficits in schizophrenia have more recently been amplified by ERP studies focusing on relative generation of late potentials that localize to higher cortical regions such as anterior cingulate or DLPFC (e.g., N2, P3) as compared to early potentials that localize to sensory brain regions such as striate (V1) and extrastriate (e.g., V3a) visual regions (e.g., P1, N1). The greater temporal resolution of ERP vs. fMRI, moreover, permits greater isolation of responses to cue- vs. target-stimuli.

As expected, significant deficits were observed even in response to cue stimuli, supporting concepts that dysfunction was related to impaired cue utilization (Dias et al., 2011). Moreover, deficits were observed in generation of both late ERP components, such as N2, P3, or contingent negative variation (CNV), that localize to frontal brain regions (e.g., ACC, DLPFC) as well as early ERP components such as P1 and N1 that localize to primary and secondary visual regions (Dias et al., 2011). In connectivity analyses, impaired performance, as reflected in d'-context, was associated with activity both within sensory and frontal brain regions (**Figure 1**). However, frontal deficits were, in turn, driven in part by failures in sensory processing, primarily within the magnocellular visual system, with little additional illness-related impairment in frontal function.

In the present study, AX-CPT deficits in schizophrenia are analyzed using a "time-frequency" approach to EEG data analysis, which utilizes trial-by-trial information to analyze the functional basis of impaired ERP generation. In the time domain approach, responses to individual stimuli or events are analyzed as a function of time, and are measured relative to a pre-stimulus baseline, which is presumed to reflect absence of activity.

By contrast, frequency domain approaches are based upon the concept that the brain is constantly engaged in ongoing, rhythmic activity, and that individual sensory events or cognitive demands not only induce new brain activity but also modulate ongoing brain activity. The amplitude of scalp-recordable rhythmic activity is dependent upon both the amplitude of activity within local oscillatory ensembles in the brain, and the degree to which ensembles are synchronized in time, with greater synchrony leading, in general, to greater scalp-detectable activity.

In order to quantify event-related effects on oscillatory processes, two separate measures are extracted in frequency domain approaches: first, the response power across EEG frequency bands for each individual event, termed "induced" power and second, the consistency of response across stimuli, termed "intertrial coherence" (ITC) or phase locking. The combination of these two measures provides a measure of "evoked" power, similar to responses measured in more common time-domain approaches. Increases in induced power may reflect either stimulus-driven activation, or event-related synchronization (ERS) of ongoing activity. In contrast, reductions of single-trial power are presumed to reflect primarily event-related desynchronization (ERD) of ongoing neural activity. Analysis of these single-trial events thus provides information that is not detectable using more traditional time domain approaches.

**FIGURE 1 | Path analysis results in the AX-70 task variant [from Dias et al. (2011), online only materials].** Component variables were overlaid on a schematic brain based upon generator locations derived from source analysis (Dias et al., 2003), monkey intracranial recordings (Dias et al., 2006), and prior fMRI studies (Barch et al., 2003). Arrows reflect significant statistical associations as shown by path analysis, with thickness of arrow representing strength of connection. CMIN/DF of the model was 1.109, and RMSEA was 0.052. For statistics, P1 values were collapsed across A- and B-cues, which were not significantly different (*p >* 0*.*2). For more information, see Dias et al. (2011).

In the frequency domain approach, analyses are typically performed using pre-defined frequency bands which are characterized based upon order of discovery. Ongoing research is attempting to determine functional roles of the different rhythms and their interactions. Thus, the alpha frequency band (8–14 Hz) was first described in the 1930's and has been more recently shown to be related with mechanisms of alertness (Klimesch et al., 1998), attention (Foxe et al., 1998; Worden et al., 2000; Thut et al., 2006; Lakatos et al., 2008; Gould et al., 2011; Rohenkohl and Nobre, 2011), and inhibition (Klimesch et al., 2007).

The beta frequency band, (∼14–30 Hz) has been associated with sensory-motor activity (Pfurtscheller, 1981; Schoffelen et al., 2008), sensory gating (Hong et al., 2008), and attention (Gross et al., 2004). Oscillations in the high frequency gamma band have been proposed as mechanisms for binding different visual features of objects (Gray et al., 1989), attention (Fries et al., 2001), and memory (Tallon-Baudry et al., 1998), and in the lower frequency theta band have been linked to hippocampal functions such as learning and memory (Buzsaki, 2002) and also entrainment of higher frequencies and sensory selection (Lakatos et al., 2008; Schroeder and Lakatos, 2009).

Consistent with this hypothesized role of alpha/beta frequency activity, we have previously demonstrated in a group of healthy individuals that cue stimuli in the AX-CPT elicit task-related reductions in alpha/beta activation during the cue and evaluation period, which are hypothesized to reflect the bringing of these regions "on-line" to permit them to process cue-related information. Similarly, we observed beta reductions over motor regions in response to valid cues, particularly in task versions that were most likely require a motor response (Bickel et al., 2012). The present study analyzes these events in patients with schizophrenia, in order to evaluate patterns of oscillatory analysis underlying impaired cue processing in the AX-CPT as a model system for evaluating mechanisms of overall visual system dysfunction in schizophrenia.

As in our prior analysis in healthy volunteers, separate analyses were performed within 3 discrete time windows: (1) a sensory response period (0–0.2 s) in which activity is driven by external stimuli and is localized to primary sensory regions, (2) a sensory evaluation period (0.2–0.5 s) in which sustained, continued event-related modulations of activity are observed primarily over sensory regions as well, and (3) a preparation interval (0.8– 1.2 s) over which time activity appears modulated primarily by the likelihood of an upcoming target stimulus and requirement to respond. Based upon our prior time-domain analyses (Dias et al., 2011), we hypothesized that patients would show not only reduced stimulus-induced activity during the sensory response period, but also reduced stimulus-induced beta modulation during the sensory evaluation period. These reductions, which are not evident in time-averaged ERP data, provide additional mechanisms regarding neural basis of visual information processing impairments in schizophrenia.

# **METHODS**

#### **PARTICIPANTS**

Thirty patients meeting DSM-IV criteria for schizophrenia (SCZ), and 17 healthy controls (HCs) participated in the study. However, frequency analysis requires long epochs of EEG data, which results in the elimination of many trials, and so several subjects were rejected from the current analysis due to insufficient number of trials, and consequently the final number of subjects for this analysis is lower (18 SCZ, 13 HC). Two additional SCZ had to be removed from some analyses because of excessive line noise in one or more of the tasks.

The patients were diagnosed using the Structured Clinical Interview for DSM-IV and available clinical information, and were recruited from inpatient and outpatient facilities associated with the Nathan Kline Institute. HCs were recruited from the Nathan Kline Institute's Volunteer Recruitment Pool and were free from any SCID-defined Axis I psychiatric disorders. All participants had normal or corrected to normal visual acuity. All but one patient and one control were right-handed, as assessed with the Edinburgh Handedness Inventory (Oldfield, 1971). Participants were not included if they met criteria for substance or alcohol dependence in the last 6 months, or abuse within the last month. This study was approved by the Nathan Kline Institute internal review board. Additional subject characteristics, including neuropsychological scores, are presented in **Table 1**.

#### **PARADIGM**

The AX-CPT has been described in detail in previous papers (Dias et al., 2003, 2011; Bickel et al., 2012). In summary, a stream of cue-probe letter sequences were presented sequentially on a computer screen located 137 cm in front of the subjects' eyes using Presentation software (Neurobehavioral Systems, Inc., Albany, CA). Letters subtended ∼2◦ of visual arc and were presented for 100 milliseconds, white on black, using Helvetica font.

The interval between onset of cue and onset of probe letters (SOA) was 1240 ms, and the interval between successive

#### **Table 1 | Subject characteristics.**


*Abbreviations: BPRS, Brief Psychiatric Rating Scale (Overall and Gorham, 1962); ILS, Independent Living Scales (Loeb, 1996); IQ, intelligence quotient (Ammons and Ammons, 1962); SANS, Scale for the Assessment of Negative Symptoms (Andreasen, 1984); SES, socioeconomic status, measured by 4-factor Hollingshead Scale (Hollingshead, 1975).*

*\*P-values from t-tests.*

cue-probe sequences (ISI) was 1390 milliseconds. Subjects were instructed to respond by pressing a button with their right hand if the cue-probe sequence was the letter "A" followed by the letter "X," while ignoring all other cue-probe sequences. Invalid cues, collectively referred to as "B" and invalid probes, collectively referred to as "Y," were letters other than A and X. The response window following probe presentation was 1 s.

The task used in this study is different from some other versions of AX-CPT, in that a response is only required for the AX condition, whereas in other studies (Cohen et al., 1999; Barch et al., 2001; Braver et al., 2009), one response is required for correct and another for incorrect sequences. Thus, our task creates different expectations of a motor activity that are not present in other studies (Dias et al., 2003, 2011; Bickel et al., 2012).

Variations of the task were created by changing the proportions of each type of trial, generating different expectation of response. Three variants were used, and in each a specific cueprobe sequence was presented with 70% probability, while all other sequences were presented with 10% probability, in pseudorandom order (**Table 2**). Adjusting the prevalence of each cue-probe sequence allowed for selective manipulation of both the local and global prepotency to respond. Thus, the variant in which 70% percent of the trials were sequence A-X (named AX-70) created a global prepotency to respond to the probe X, and the subject had to use the information from the cue "B" to inhibit incorrect responses to the X in B-X sequences. In the variant with 70% of A-Y trials (AY-70), the subjects did not prepare a response following cue A, as most probes were invalid, and thus had to rapidly prepare a response when the target probe X was shown. And in the third variant, BX-70, the subject has mostly invalid trials, but once a cue A was shown, they had a 50% chance of having a response. The probabilities generated by the different task variants are detailed in **Table 2**.

Each subject performed 6 blocks of 93 trials for each of the 3 tasks, totaling 1674 trials. In most cases, task AX-70 was presented first because this was the condition of primary interest. Task order did not significantly affect performance either across or within group [main effect of order: *F(*3*,* <sup>41</sup>*)* = 0*.*25, *P* = 0*.*86; group vs. order interaction: *F(*3*,* <sup>41</sup>*)* = 0*.*76, *P* = 0*.*52]. Subjects took many breaks between blocks and did not report abnormal fatigue during the tasks.

Performance was indexed by a modification of the signal detection theory measure of sensitivity d', called d'context, that calculates the separation between the means of the signal and noise distributions, and is calculated from measurements of the hit rate and false-alarm rate: d' context = *Z*(hit rate in AX trials) − *Z*(false alarm rate in BX trials), where *Z* is the inverse of the cumulative Gaussian distribution (Barch et al., 2001). A higher d' indicates that the signal can be more readily detected, and by restricting the analysis to AX and BX trials, the effect of cue encoding on the task performance can be indexed.

#### **EEG**

Continuous EEG data were originally collected and digitized at 512 Hz with an ActiveTwo system (Biosemi, Amsterdam, the Netherlands) using a 168 channel high-density cap as this montage afforded us the excellent spatial resolution required for source localization of the effects. However, for the purpose of the frequency analysis of scalp data, such a high electrode density is less critical. We therefore reduced the redundancy of the scalp electrode signatures by interpolating the data to a 27 channel virtual montage, referenced to the center of the head, for data analysis using BESA (Brain Electric Source Analysis, MEGIS Software GmbH, www.besa.de, Gräfelfing, Germany; **Figure 2**). By doing so we reduced the influence of a reference electrode. We further combined the data from a subset of virtual electrodes to create larger regions of interest, specified below, to investigate activity over scalp regions of interest based on a priori hypotheses, guided by previous investigations (Dias et al., 2011; Bickel et al., 2012).

Continuous data was epoched from −300 to 1550 milliseconds surrounding the digital tags that marked the onset of the stimuli, and band pass filtered (0.05 to 100 Hz). Trials with activity exceeding ± 120 microvolts, with eye blinks, or with incorrect


*Abbreviations: P(A), probability of cue A; P(X), probability of target X; P(X|A), probability of target X, given cue A.*

responses, were eliminated. Occasional noisy channels were substituted by interpolated data from neighboring channels. Subjects with fewer than 40 correct trials were removed from the analysis.

In most cases, epochs were baseline-corrected to the average of the pre-stimulus period −300 to −100 before stimulus onset, but the data were also analyzed without this correction to examine absolute power. The data were re-referenced to the average reference.

The signal in each individual trial was decomposed into the time-frequency domain using the complex demodulation procedure implemented in BESA for a frequency window of 4–50 Hz. Frequencies were sampled in 2-Hz steps, and time in 25 millisecond steps.

Further analyzes were performed using MATLAB (Mathworks, Natick, MA), with Fieldtrip (http://www*.*ru*.*nl/fcdonders/ fieldtrip/) and custom written scripts. The data for single trials was grouped by subject, to allow statistical analyses of the effects, and then, for the figures, into a grand average of all subjects in each group.

A separate analysis was done, for comparison, by extracting the frequency information from the previously grouped data. This provided the power of the evoked potentials, shown in **Figure 3** for task AX-70, cue A, for both SCZ and HC, as opposed to the induced activity that is observed when analyzing the power of the frequencies in single trials (**Figures 4**, **7**). For the spatial representation of the evoked potentials shown in **Figure 3A**, the data were analyzed in the time domain with a low pass 6th order Butterworth filter at 35 Hz and baseline correction from −100 to 0 ms before stimulus onset.

For illustration purposes only, to better display the effects on the various frequency bands in the same time-frequency plots (**Figures 4**, **7**), and counter the effect of the power reductions in the higher frequencies, the power in each 2 Hz frequency band was multiplied by the frequency it represented. Data analysis was performed on non-multiplied data.

To localize the sources of the activity recorded on the scalp, multi source beamformer (MSBF) was performed using the BESA source coherence module, by calculating the cross spectral density matrix of the specific frequency and time ranges of interest, for the grand mean average of each condition for each group, and computing the output power of the beamformer at each grid location. The figures (**Figures 6**, **9**) show the location of the respective grid maximums plotted on the BESA standard anatomical MRI.

#### **STATISTICAL ANALYSIS**

Frequency boundaries used in this analysis were similar to those used in our previous manuscript (Bickel et al., 2012) and other published literature (Pfurtscheller, 1981; Farmer, 1998; Worden et al., 2000; Kelly et al., 2006; Uhlhaas and Singer, 2010), while avoiding overlap of the frequency ranges. Thus, we analyzed activity in the frequencies bands: theta (4–6 Hz), alpha (8–14 Hz), beta (16–24 Hz), and gamma (30–50 Hz). We did not analyze the data above 50 Hz.

We focused our analysis on three time windows: sensory, evaluation, and preparation (Bickel et al., 2012). The sensory window (0–0.2 s) was meant to capture the initial sensory volley of activity, including sensory ERPs P1 and N1; the evaluation window (0.2–0.5 s) was meant to capture the encoding activity; and the

related potential on the scalp, seen from behind the head, on both healthy controls (top panel) and schizophrenia patients (bottom panel). **(B)** Power of the evoked potential, obtained by extracting the frequency components of the averaged trials. **(C)** *T*-test comparison between patients and controls. The results are masked to only show *t*-values with a *p <* 0*.*05.

**FIGURE 4 | Evaluation window. (A)** Top view of the spatial distribution of beta induced power during the evaluation period, for task AX-70; the red oval represents the scalp region in the occipito-parietal region that was used for the statistical analysis. The data for healthy controls are on top panels and for schizophrenia patients on the bottom panels. **(B)** Time-frequency distribution of activity in the occipito-parietal region following cue A onset (time 0), adjusted for power of the different frequencies (i.e., power multiplied by the specific frequency), for tasks AX-70 (left) and BX-70 (right). The black dotted rectangle represents the time-frequency window used for the analysis of beta power, and the red dotted rectangle represents the window for the alpha power. **(C)** Top view of the spatial distribution of the alpha induced activity for task AX-70.

preparation window (1–1.3 s) captured the activity in preparation for the appearance of the probe.

Regions of interest for analysis were chosen with basis on previous published data. For analysis of the activity of alpha and beta frequency bands over occipito-parietal cortex, data was obtained from an occipital parietal region, including virtual electrodes O1, O2, P7, and P8, of the 27 electrode montage (**Figure 5A**). Preparation beta was obtained from a left antero-medial region, including virtual electrodes FZ and F3, based on previous findings (Pfurtscheller, 1981; Alegre et al., 2006; Ritter et al., 2009; Bickel et al., 2012) (**Figures 7**, **8D**). In addition, also based on results from our exploratory analysis on that study (Bickel et al., 2012), we analyzed the activity in the beta frequency range in the occipito-parietal region. Theta and gamma activities were examined both from an antero-medial region, including virtual electrodes FZ, F1, and F3 and from the occipito-parietal region.

Repeated measures ANOVA, with subsequent paired *t*-tests were used to assess significance. The model applied to each frequency band included task (AX-70, AY-70, or BX-70), cue type (A or B) and period (sensory, evaluation, preparation) as withingroup factors and group (patient, control) as between-group factor. A secondary analysis was done for each period. All tests had a preset level of significance of *p <* 0*.*05. The commercial program SPSS (version 15) was used to compute the statistical tests. For the *t*-test analysis of the evoked time-frequency data, a cluster-based test was used as implemented in the Fieldtrip toolbox (Maris and Oostenveld, 2007). This approach controls for multiple comparison testing when computing statistics across multiple channels, and/or frequency and time points, and is further described in Bickel et al. (2012). A cluster level alpha of 0.05 was used.

#### **RESULTS**

#### **BEHAVIORAL RESULTS**

AX-CPT performance by these subjects has been reported previously as part of a larger study (Dias et al., 2011). In this subgroup, behavioral findings remained unchanged from the larger group, including reduced performance accuracy across all trial types [*F(*1*,*27*)* = 12*.*53, *p* = 0*.*001] and prolonged reaction times [*F(*1*,* <sup>27</sup>*)* = 27*.*65, *p <* 0*.*0001] across tasks (**Table 3**). As in the larger sample, patients showed greatest deficits in the BX-70 version of the task lowest levels of deficit in the AX-70 version, leading to a significant group × task interaction [*F(*2*,* <sup>26</sup>*)* = 6*.*61, *p* = 0*.*005]. As in the larger group, performance deficits across task versions were due both to decreased hits on valid cue, correct target (AX) trials and false alarms to invalidly cued, correct targets (BX) trials.



#### **EEG RESULTS**

ERP responses were analyzed only from trials associated with a correct behavioral response. As previously (Bickel et al., 2012), separate analyses focused on activities within sensory (0–0.2 s), evaluative (0.2–0.5 s) and response preparation (0.5–1.0 s) time windows.

#### *Sensory window (0–0.2 s)*

In the sensory window, the primary averaged activity consisted of a bi-occipital positivity centered at 100 ms (**Figure 3A**). Evoked power of this activity was primarily in the theta frequency range, though it extended into higher frequencies for both groups (**Figure 3B**). Activity in patients was significantly reduced, as shown by *t*-tests between groups, followed by a cluster analysis (**Figure 3C**). The main deficits in patients vs. controls were observed in the initial stage of the response, consistent with the selective reduction of fast (e.g., magnocellular) input to the cortex.

Single-trial analysis showed additional power in the beta frequency range (16–24 Hz), with a significant group effect [*F(*1*,* <sup>27</sup>*)* = 14*.*6, *p* = 0*.*001], and a task by cue interaction [*F(*2*,* <sup>26</sup>*)* = 7*.*369, *p* = 0*.*003]. There were no interactions with group, suggesting that although power was reduced in patients, there was a similar modulation of beta activity between groups in this period. For induced power of oscillations in the alpha and gamma frequency bands there were no effects of task, cue or group, and no significant interactions in this period.

#### *Evaluation window (0.2–0.5 s)*

In the evaluation time window, the main task-related modulations of oscillatory activity consisted of reductions in both alpha (8–14 Hz) and beta (16–24 Hz) frequency ranges, consistent with ERD of ongoing alpha/beta activity. The ERD started shortly after offset of the sensory ERP and persisted for variable duration depending upon task condition, with shortest duration in the AX-70 and AY-70 conditions, and longest duration in the BX-70 condition (**Figure 4B**). ERDs in both the alpha and beta frequency ranges were strongly localized over parieto-occipital scalp, suggesting modulation of activity within primary and secondary visual areas (**Figures 4A,C**).

Despite similar timing of the posterior alpha and beta activity, differential deficits were observed in patients vs. controls for the different frequency bands. In the alpha frequency range, patients and controls showed reductions in single-trial power that were large and that varied little across task versions (**Figure 5B**). On ANOVA, there were no significant main effects of group [*F(*1*,* <sup>27</sup>*)* = 0*.*146, *p* = 0*.*706], cue-type [*F(*1*,* <sup>27</sup>*)* = 0*.*011, *p* = 0*.*917], task [*F(*2*,* <sup>26</sup>*)* = 0*.*220, *p* = 0*.*804] or group X task interaction [*F(*2*,* <sup>26</sup>*)* = 0*.*186, *p* = 0*.*832].

By contrast to lack of cue/task effects or between-group difference in degree of alpha reduction across groups, there were highly significant effects of these parameters on beta ERD during the same evaluation latency range. Patients showed an overall reduction in magnitude of the beta ERD, resulting in a highly significant main effect of group [*F(*1*,* <sup>27</sup>*)* = 10*.*160, *p* = 0*.*004]. In controls, the magnitude of the beta ERD to A cues was smallest in the AY-70 task condition, intermediate in the AX-70 condition and greatest in the BX-70 condition (**Figure 5C**).

An opposite pattern of effects was observed for B cues, leading to a significant task version by cue type interaction [*F(*2*,* <sup>26</sup>*)* = 4*.*801, *p* = 0*.*017], although there were no significant main effects of task [*F(*2*,* <sup>26</sup>*)* = 0*.*603, *p* = 0*.*555] or cue [*F(*1*,* <sup>27</sup>*)* = 1*.*194, *p* = 0*.*284]. Patients showed a loss of this modulation by task, which resulted in a trend toward a 3-way task X group X cue interaction [*F(*2*,* <sup>26</sup>*)* = 2*.*569, *p* = 0*.*096]. Follow up ANOVAS for each group confirmed this finding, with controls showing significantly modulated beta ERD by task and cue [*F(*2*,* <sup>11</sup>*)* = 10*.*790, *p* = 0*.*003], reflecting differential information values of the cues in the different tasks, whereas patients did not [*F(*2*,* <sup>14</sup>*)* = 1*.*204, *p* = 0*.*329].

The pattern of activity in controls mirrored the probability ratio of the response, i.e., the likelihood of having to make a response (sequence AX), given the task (global prepotency) and the cue (local prepotency; **Table 2**), with beta ERD to "A" cues increasing with greater probability of requirement for an upcoming response (**Figure 5C**).

In addition, the BX-70 task, in which controls had the best performance, as indexed by d'-context, also showed a larger difference in beta ERD between cues (CueA – CueB), and this difference was reduced in patients, as was d'-context (**Figure 5D**). This suggests that the beta ERD may carry information differentiating the cues that contributes to task performance.

Given the orientation of visual generators, a reflection of the occipital activity was observed as well over anterior scalp regions (**Figures 4A,C**). The phase of the beta activity in this area was tested and, as expected, this activity was in counter-phase to the occipital activity, suggesting that the measured activity represented opposite poles of the same dipole. Moreover, amplitude of the frontocentral activity varied in parallel to the occipital activity across cue and task conditions, as assessed by Pearson correlations between the beta ERD power in anterior and posterior scalp regions throughout the trial (0–1.2 s). For example, for task AX-70, correlations between anterior and posterior beta ERD were all highly significant (*p <* 0*.*0001) both for patients and controls (controls, cue A, *r* = 0*.*656; cue B, *r* = 0*.*803; patients, cue A, *r* = 0*.*844, cue B, *r* = 0*.*822).

Consistent with this configuration of surface activity, in controls source analysis localized the main sources for the alpha and beta ERDs during the evaluation period in posterior regions of the brain, in the visual areas, with an emphasis on inferotemporal regions (**Figures 6A,B**). For patients, there was no detectable source at the same location (not shown), though for alpha ERD only there was a weaker source more localized to the occipital pole (**Figure 6C**).

Despite the smaller magnitude of the beta ERD, patients showed similar pre-stimulus absolute amplitude to controls within both the alpha (Pat: 17.708 ± 1.744; Con: 14.386 ± 1.963; *p* = 0*.*217) and beta (Pat: 3.333 ± 0.446; Con: 3.843 ± 0.417; *p* = 0*.*440) frequency ranges. No significant cue-induced modulation was observed within either theta or gamma frequency ranges during the evaluation period in any task version for either group. Furthermore, no difference in absolute amplitudes of theta or gamma activity was observed across groups.

#### **FIGURE 5 | Evaluation window. (A)** Location of the electrodes over the occipito-posterior scalp from which the data was derived. **(B)** Variation of the alpha ERD with the probability ratio for all three task variations, for cue A. There were no significant effects. **(C)** Variation of the beta ERD with probability ratio for all three task variations, for cue A (**Table 2**). There was a significant group difference and an interaction of task by cue. **(D)** Relation of the difference in beta ERD between cue A and B and performance measured by d' context, with linear trendline for all data points.

#### *Preparation window (0.8–1.2 s)*

In the preparation window, a significant reduction in beta activity was also observed in the control group. However, as opposed

to the beta ERD observed in the evaluation window, beta ERD during the preparation window was maximal over frontocentral scalp, with asymmetric distribution toward left motor/hand region. Amplitude and spectral width of the beta ERD increased progressively to the point of target presentation (**Figure 7B**).

Furthermore, the pattern of the beta ERD across task versions was significantly different from the pattern of beta ERD in the evaluation window. Specifically, the magnitude of the beta ERD varied with the probability of a response following the presentation of cue A (**Figure 8A**). Thus, it was least in the AY-70 condition in which target ("X") probability was low even after an "A" cue, intermediate in the BX-70 condition in which target probability was intermediate, and largest in the AX-70 condition, in which target probably was highest, leading to significant main effects of task [*F(*2*,* <sup>26</sup>*)* = 3*.*407, *p* = 0*.*049] and cue [*F(*1*,* <sup>27</sup>*)* = 14*.*22, *p* = 0*.*001] and of the cue by task interaction [*F(*2*,* <sup>26</sup>*)* = 6*.*354, *p* = 0*.*006]. This relationship was not present for cue B, which guaranteed that there was no need to prepare a response, independent of the upcoming probe type. Thus, the beta ERD in the preparation window appeared to track probability of response. In controls, the reaction time across tasks closely followed the pattern of the beta ERD, suggesting that the beta ERD represents response preparation (**Figure 8B**).

In patients, the magnitude of the beta ERD was dramatically reduced leading to a highly significant main effect of group [*F(*1*,* <sup>27</sup>*)* = 11*.*38, *p* = 0*.*002; **Figure 7B**, bottom panel]. Patients, like controls, showed increases in beta ERD magnitude across conditions as a function of target probability (**Figure 8A**). Although the differences were smaller overall than in controls, the group X task interaction was not significant [*F(*1*,* <sup>27</sup>*)* = 0*.*729, *p* = 0*.*492], suggesting a similar level of modulation. RTs for patients were significantly longer than controls across task versions [*F(*1*,* <sup>27</sup>*)* = 27*.*65, *p <* 0*.*0001], but were proportionate to

the left fronto-central region adjusted for power of the different frequencies, for task AX-70. The red dotted square represents the time-frequency window used for the analysis of beta power, and the black dotted square represents the window for the alpha power, a little earlier (0.9–1.1 ms after cue onset). **(C)** Top view of the spatial distribution of the alpha activity for task AX-70.

the smaller beta ERDs (**Figure 8B**). Source analysis of the beta ERD in controls shows high localization to hand motor/premotor areas (**Figure 9A**). This is not observed in the model for the patients (not shown).

In addition, in the preparation interval (0.8–1.2 s) there was an increase in alpha power that was only marginally observed in controls but which was prominent in patients (**Figure 7B**) and preceded the beta ERD, so an exploratory analysis was performed. The activity showed a similar spatial distribution (**Figure 7C**), and, consistent with the visualized data, the increase in alpha power was significantly larger in patients than controls [*F(*1*,* <sup>27</sup>*)* = 4*.*81, *p* = 0*.*04]. A model of the sources of alpha activity in this period indicated that in controls the main source was also localized to the antero-medial region, in a similar in location to the beta activity, but was absent in this region in patients. The main peak of activity in patients had predominant occipital localization (**Figure 9C**).

As opposed to baseline activity over posterior scalp regions, baseline activity over left frontal scalp was modulated both as an effect of task and across group. Activity in this region showed a task effect [*F(*2*,* <sup>26</sup>*)* = 5*.*505, *p* = 0*.*010], and, although there was no main effect of group [*F(*1*,* <sup>27</sup>*)* = 1*.*155, *p* = 0*.*292], an interaction of task by group [*F(*2*,* <sup>26</sup>*)* = 3*.*619, *p* = 0*.*041] (**Figure 8C**). While for patients there was not much variation of the baseline, for controls there was an increase in beta baseline for task AX-70. Tests of within-subject contrasts showed that there were

**FIGURE 8 | Preparation window. (A)** Relationship of preparation beta ERD with the probability of target probe X, given cue A, for the different task variations (**Table 2**). There were main effects of task and group. **(B)** Relationship of preparation beta ERD power and reaction time (in milliseconds), showing that as beta ERD in the preparation window decreases, the reaction time increases. **(C)** Variation of ongoing baseline beta power with the different tasks that create different expectations of responses. In controls, task AX-70, that has a higher probability of response, has a larger power of beta than the other tasks. **(D)** Location on the scalp of the left antero-medial electrodes used for statistical analysis.

significant differences between the baselines for task AX-70 and AY-70 (*F* = 5*.*629, *p* = 0*.*025) and between AX-70 and BX-70 (*F* = 11*.*340, *p* = 0*.*002).

#### **GAMMA**

Analysis of baseline-corrected gamma activity over the occipitoparietal region showed no significant effects of group, cue or task, or any interactions of these factors in any of the periods. The same was true for the antero-medial region.

In the occipito-parietal region only, the baseline value of gamma showed an interaction of task by group [*F(*2*,* <sup>26</sup>*)* = 4*.*560, *p* = 0*.*020], which was driven by a modulation of absolute gamma power by task in controls [*F(*2*,* <sup>11</sup>*)* = 4*.*728, *p* = 0*.*033], but not in patients [*F(*2*,* <sup>14</sup>*)* = 2*.*424, *p* = 0*.*125]. There were no main effects of task [*F(*2*,* <sup>26</sup>*)* = 0*.*364, *p* = 0*.*698] or group [*F(*1*,* <sup>27</sup>*)* = 1*.*406, *p* = 0*.*246].

# **DISCUSSION**

Cognitive deficits represent a core component of schizophrenia and a primary determinant of impaired long-term outcome. While such deficits are distributed throughout cortex, deficits are strongly manifest within early visual regions as well, and thus provide insights into mechanisms by which early visual deficits contribute to the overall cognitive dysfunction associated with schizophrenia. The present study utilizes evolving methods in time frequency analysis of ERP data to isolate neural patterns that may underlie the increasingly well-documented visual disturbances in schizophrenia.

Principal findings are two-fold: first, that significant additional activity can be detected in time-frequency analysis of single-trial ERP data than is apparent from analysis of average ERP data alone and, second, that deficits in stimulus evaluation and task performance in schizophrenia are primarily reflected in impaired modulation of beta activity across sensory, evaluation and motor preparation intervals. In contrast, relatively preserved activity was observed within other frequency bands, suggesting that deficits in beta generation may be a core component of schizophrenia and a tool for translational modeling of cortical dysfunction.

In visual tasks, such as the AX-CPT, primary evoked activity occurs within the theta frequency range. In intracortical recordings, this activity corresponds to a feedforward pattern of activation within primary (V1,V2) and secondary (e.g., V3a) regions, and thus most likely reflects feed-forward, thalamocortical activation mediated by magnocellular- and parvocellulargeniculocortical afferents. Stimuli used in this task are naturalistic (i.e., letters) and so do not isolate either magno- or parvocellular pathways. Nonetheless, the main difference between groups was early after cue presentation, supporting previous reports that the main deficit in patients is in the faster magnocellular input (Butler et al., 2005; Martinez et al., 2008, 2012).

In addition to evoked activity, in cognitive tasks such as AX-CPT, stimuli lead to well-described modulations of ongoing activity that interact with task demands. These modulations occur with relatively large phase variability across trials, and so are not typically observed in traditional time-domain analyses. By contrast, in frequency domain analyses of single trial (induced) power, both event-related reductions in power are observed reflecting ERDs or increases in power reflecting primarily ERSs. Reductions in alpha/beta power, in particular, are thought to reflect local disinhibition of cortical regions, permitting them to engage in detailed stimulus processing. In addition, coherent beta activity across brain regions is thought to reflect interregional information transfer (Siegel et al., 2012).

In AX-CPT, we have previously observed significant beta desynchrony over visual areas during a stimulus evaluation period in healthy volunteers as well as beta desynchrony over motor regions during response preparation (Bickel et al., 2012). We have also observed that patients with schizophrenia are slower, and less accurate than controls in the AX-CPT task (Dias et al., 2011). The present study evaluated integrity of time-frequency modulations in patients with schizophrenia vs. healthy volunteers relative to their performance impairment. As expected, deficits were observed in both the evaluation and motor preparation latency ranges, reflecting disrupted neural activity, as described below.

#### **EVALUATION WINDOW**

In the evaluation latency range, controls showed a prominent, long-lasting ERD that involved both alpha and beta frequency ranges, and that persisted for variable time depending upon task. An unexpected finding in the patient group is that while they had marked reduction in magnitude of the beta ERD, the magnitude of the alpha ERD was unchanged.

In controls, alpha and beta modulations showed very similar source localizations, and thus likely represent operation of similar brain regions. Nevertheless, whereas activity in the beta range was significantly modulated as a function of cue and task, activity in the alpha range was not, suggesting that they may reflect operation of ensembles engaged in extraction of separate types of information. For example, alpha-blocking is observed even during processing of relatively primitive stimuli (e.g., flash, gabor), whereas in this study the stimuli are both complex and contain information related to their configural identity. Thus, the beta activity may be more related to processing of the complex stimuli such as letters, rather than simply as visual events.

Consistent with this hypothesis, in controls the magnitude of the beta ERD varied significantly across tasks, with inverse pattern for A and B cues. Moreover, the magnitude of the beta ERD was greatest when the information content of the stimulus was greatest relative to both global and local prepotencies. Thus, for task AX-70, cue A was more common than cue B and also indicated a higher probability of a response. In this case, the beta ERD was greater for cue B than cue A, consistent with the fact that cue B held more information in that it necessitated a decision to override the prepotent set to respond to a subsequent target stimulus.

By contrast, in task BX-70, cue B was more common and the global prepotency was to not respond even when a target was presented. In this case, an A cue would signal the need to override the global prepotency and thus was most informative with respect to successful task completion. The differential response within the alpha and beta frequency bands thus suggests that these reflect operations of spatially overlapping but functionally distinct neuronal ensembles, with alpha-frequency ensembles being primarily involved in stimulus detection, but beta frequency ensembles being involved in information extraction from the stimuli. The similar magnitude of the alpha modulation suggests that patients are appropriately detecting stimuli, but impaired in their ability to bring "on line" the specific ensembles needed to interpret the cues relative to task demands.

Other studies have also associated beta modulations with cognitive tasks (Tallon-Baudry, 2003; Pesonen et al., 2007; Haenschel et al., 2009), including beta activity in occipito parietal areas as in this study (Pesonen et al., 2007).

#### **MOTOR PREPARATION**

Beta activity in pre-central areas has been traditionally linked to motor control. Studies have shown that beta decreases prior to movement onset, and rebounds after termination, maintaining the activation during steady contraction (Pfurtscheller, 1981; Sanes and Donoghue, 1993; Crone et al., 1998; Farmer, 1998; Donner et al., 2009). It appears that during periods of low beta power the motor cortex is in a receiving state for new information or commands from other cortical areas (Crone et al., 1998; Bickel et al., 2012). Furthermore, since movements require only a small region of motor cortex (e.g., right index finger), the scalp-recorded reduction may represent a sharpening of the tuning curve for motor cortex in general, with suppression of those regions of motor cortex not necessary for the response but facilitation of regions particularly involved. Given the well-known association of beta modulation over motor cortex and motor response, it was possible to monitor task-related modulations of activity over antero-medial regions, and in particular left motor/premotor hand area relative to performance on the task.

In this study, as in our previous one with healthy subjects (Bickel et al., 2012), there was a gradual increase of beta ERD in the epoch preceding the potential target, proportional to the probability that a response might be required, that is, following A cues in the different tasks. This increase in beta ERD was also proportional to the reaction time, supporting the link between preparation beta and motor preparation (Bickel et al., 2012). Patients showed a much reduced beta ERD overall, though all trials examined resulted in a correct response. However, SCZ patients had significantly longer reaction times, suggesting that alternative, and possibly longer, neural pathway were recruited or, alternatively, weaker activity had accumulate before a response could be initiated. The model of the source of beta band activity in the preparation window showed no detectable activity in patients in the areas most activated in controls, i.e., the left antero-medial region, overlaying the motor/premotor cortex. In addition, patients had a stronger alpha ERS in the preparation window, that might represent activation of alternative pathways that contribute to task completion, but this remains to be tested more extensively. An increase in cortical synchronization in alpha frequencies ("high alpha," 11–13 Hz), with a similar topography, has been previously shown in schizophrenia patients relative to controls, during the correct performance of a working memory task (De Vico Fallani et al., 2010). The authors also suggested this increase in synchronization could be related to compensatory mechanisms.

As in our previous study (Bickel et al., 2012), there was no significant gamma modulation during the task in either controls or patients, in the regions and for the range of frequencies we tested (30–50 Hz). The only gamma difference we observed was a variation of baseline activity with task in controls that was not present in the patient data. Recent studies have shown that gamma modulation in schizophrenia patients occurs mainly for higher frequency gamma (*>*50 Hz), for example in working memory (Haenschel et al., 2009) and visual encoding (Grutzner et al., 2013). Since our study was restricted to frequencies below 50 Hz, these reductions in high gamma would not have been detected in our study.

Although this study used relatively simple stimuli, and a relatively simple task, we would like to propose that similar deficits in beta activation may underlie other aspects of visual dysfunction. In addition, the fact that we observed deficits over motor cortex during the motor preparation period suggests that there is a generalized failure to modulate beta activity throughout task performance. The lack of coordination between brain regions, and lack of selective and sequential activation of appropriate brain regions may give rise to the cortical dysfunction that could underlie the profound impairment of schizophrenia patients.

The dissociation between beta modulations, which were severely impaired within both sensory and motor regions, and alpha and gamma modulations, which were relatively intact in schizophrenia, gives rise to a relatively specific pattern of neurophysiological impairment which can be used translationally as a relatively specific marker of cortical dysfunction in schizophrenia.

# **FUNDING**

This work was supported by the National Institute of Health (R37MH49334, P50MH086385 to Daniel C. Javitt).

# **REFERENCES**


Dysfunction of early-stage visual processing in schizophrenia. *Am. J. Psychiatry* 158, 1126–1133. doi: 10.1176/appi.ajp.158.7.1126


# **ACKNOWLEDGMENTS**

The authors would like to thank Marina Ross and Jeanette Piesco for EEG data collection, Joanna DiCostanzo for patient recruitment, Erica Saccente for help with data analysis, Gail Silipo for invaluable help with management of the studies, and also the faculty and staff of the Clinical Research and Evaluation Facility and the Outpatient Research Service at the Nathan S. Kline Institute for Psychiatric Research.

a high density electrical mapping study of cortical control. *Cereb.Cortex* 13, 701–715. doi: 10.1093/cercor/13.7.701


*Sci. U.S.A.* 101, 13050–13055. doi: 10.1073/pnas.0404944101


sustained visuospatial attention. *J. Neurophysiol.* 95, 3844–3851. doi: 10.1152/jn.01234.2005


C. S. (2003). A specific deficit in context processing in the unaffected siblings of patients with schizophrenia. *Arch. Gen. Psychiatry* 60, 57–65. doi: 10.1001/archpsyc.60.1.57


activities in man. *Electroencephalogr. Clin. Neurophysiol.* 51, 253–264. doi: 10.1016/0013-4694(81)90139-5


activity over occipital cortex indexes visuospatial attention bias and predicts visual target detection. *J. Neurosci.* 26, 9494–9502. doi: 10.1523/JNEUROSCI.0875-06.2006


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 16 May 2013; accepted: 31 July 2013; published online: 27 August 2013. Citation: Dias EC, Bickel S, Epstein ML, Sehatpour P and Javitt DC (2013) Abnormal task modulation of oscillatory neural activity in schizophrenia. Front. Psychol. 4:540. doi: 10.3389/fpsyg. 2013.00540*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology. Copyright © 2013 Dias, Bickel, Epstein, Sehatpour and Javitt. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# On disturbed time continuity in schizophrenia: an elementary impairment in visual perception?

#### **Anne Giersch<sup>1</sup>\*, Laurence Lalanne<sup>1</sup> , Mitsouko van Assche<sup>1</sup> and Mark A. Elliott <sup>2</sup>**

1 INSERM U1114, Department of Psychiatry, Fédération de Médecine Translationnelle de Strasbourg (FMTS), University Hospital of Strasbourg, Strasbourg, France <sup>2</sup> Department of Psychology, National University of Ireland Galway, Galway, Ireland

#### **Edited by:**

Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA

#### **Reviewed by:**

Duje Tadin, University of Rochester, USA Junghee Lee, University of California Los Angeles, USA

#### **\*Correspondence:**

Anne Giersch, INSERM U1114, Department of Psychiatry, Fédération de Médecine Translationnelle de Strasbourg, University Hospital of Strasbourg, 1, Place de l'Hôpital, F-67091 Strasbourg Cedex, France. e-mail: giersch@unistra.fr

Schizophrenia is associated with a series of visual perception impairments, which might impact on the patients' every day life and be related to clinical symptoms. However, the heterogeneity of the visual disorders make it a challenge to understand both the mechanisms and the consequences of these impairments, i.e., the way patients experience the outer world. Based on earlier psychiatry literature, we argue that issues regarding time might shed a new light on the disorders observed in patients with schizophrenia. We will briefly review the mechanisms involved in the sense of time continuity and clinical evidence that they are impaired in patients with schizophrenia. We will then summarize a recent experimental approach regarding the coding of time-event structure in time, namely the ability to discriminate between simultaneous and asynchronous events. The use of an original method of analysis allowed us to distinguish between explicit and implicit judgments of synchrony. We showed that for SOAs below 20 ms neither patients nor controls fuse events in time. On the contrary subjects distinguish events at an implicit level even when judging them as synchronous. In addition, the implicit responses of patients and controls differ qualitatively. It is as if controls always put more weight on the last occurred event, whereas patients have a difficulty to follow events in time at an implicit level. In patients, there is a clear dissociation between results at short and large asynchronies, that suggest selective mechanisms for the implicit coding of time-event structure. These results might explain the disruption of the sense of time continuity in patients. We argue that this line of research might also help us to better understand the mechanisms of the visual impairments in patients and how they see their environment.

**Keywords: schizophrenia, time, anticipation, synchrony, attention, implicit processing, Simon effect**

# **INTRODUCTION**

Patients with schizophrenia are known to suffer from cognitive disturbances that affect their every day life and might subtend their clinical symptoms. Abnormal visual perception represents one such impairment. The deficits are diverse, affecting visual organization in space, the processing of low-spatial frequencies, and the pattern of eye movements. Here we offer an attempt to characterize abnormal visual perception in schizophrenia from a different, non-exclusive perspective, i.e., a time perspective. All aspects of mental life involve coding the succession of events in time with disturbance in the processing of events structure of likely influence on any one or more of a number of cognitive functions, in this case the cognitive functions leading to the perception of visual structure. The hypothesis of an elementary time disorder playing a central role in psychosis has been developed earlier by psychiatrists, including Minkowski (1933) as well as Andreasen (1999) who proposed the concept of a cognitive dysmetria. The types of cognitive impairment covered by cognitive dysmetria still remain to be defined. However, in essence those impairments are temporal and here we focus on the effects of a basic disturbance in temporal coding. We will review current evidence for an elementary impairment at coding

the temporal events structure in patients with schizophrenia, and consider its possible impact on the general perception of the world from the point of view of the patients. We are especially interested in understanding the mechanisms subtending the sense of time continuity and its disturbance in patients with schizophrenia. Healthy subjects feel that time is flowing without interruption; they know that what has just happened is past, and expect something to happen next. This ability to follow events in time, and the expectation of the moments to come would provide its dynamism to psychic life and would be at the root of what has been called the "vital dynamism" by phenomenologists (Minkowski, 1933). In contrast, several psychiatrists have suggested that the sense of time continuity is disturbed in patients with schizophrenia. Patients would experience a fragmentation of the normal flow of events (Fuchs, 2007;Vogeley and Kupke, 2007) and a loss of "vital dynamism" (Minkowski, 1933). Approaching the phenomenological experience of time in patients with schizophrenia is complex, particularly given patients' typical difficulties in reporting their inner mental life. Patients nonetheless report disturbances that can be taken as indications as to how they experience their disorder. Interestingly, the patients' descriptions often relate to their experience of visual events. For example, Chapman (1966) reports several patients' descriptions, including the following:

Things go too quick for my mind [. . .]. It's as if you were seeing one picture 1 min and another picture the next.

When I start walking I get a fast series of pictures in front of me. Everything seems to change and revolves around me. Something goes wrong with my eyes and I've got to stop and to stand still.

It is difficult to derive from these descriptions which cognitive mechanisms are really disturbed in patients, but many disturbances related to time have also been described at an experimental level in patients with schizophrenia,including in particular the disturbed perception of event duration (Volz et al.,2001;Elvevåg et al., 2004; Davalos et al., 2005; Allman and Meck, 2012). The studies are based on the theory that there is an internal clock allowing us to count time and the results are interpreted in terms of a possible change in the rhythm of the clock (although see Delevoye-Turrell et al., 2012; Roy et al., 2012; Turgeon et al., 2012). However, it is unclear how a change in the internal clock relates to a disturbed sense of time continuity. Hence to explore the latter, we considered a different theoretical framework aimed at understanding time-event structure and time continuity.

Mechanisms involved in the sense of time continuity may operate at various time scales (reviewed in Wittmann, 2011). The experimental evidence points toward the existence of elementary time windows providing us a lower margin in our ability to distinguish separate events in time. Even though two spatially separate stimuli are presented at different times for very brief intervals they may appear to be presented simultaneously. Exner (1875) suggested minimum time differences in temporal order discrimination for intervals of up to 17 and 44 ms. Considering invariances, a common measure that extends across sensory modalities seems to be the minimum time required for temporal order discrimination following the successive presentation of more than two stimuli. For tactile and visual stimuli, and irrespective to the precise structure of the visual stimuli concerned, simultaneity thresholds have been determined with remarkably little variation: Brecher (1932) showed what he referred to as units of "subjective time" corresponded to average periods of 55.3 ms for tactile stimulation and periods of 56.9 ms for visual stimulation, with standard deviations of no greater than 1.4 ms.

Brecher's as well as subsequent and related empirical demonstrations of simultaneity thresholds in the mid 50 ms region have been interpreted in terms of a window of time, or a perceptual moment within which all events will be processed together leading to them being perceived as occurring simultaneously (von Baer, 1864; Lalanne, 1876; Brecher, 1932; Elliott et al., 2006, 2007; van Wassenhove,2009;Wittmann,2011).Although referred to in terms of perception, time windows of this order of magnitude would represent an elementary quantum that does not necessarily relate directly with experienced duration. Only larger time windows of up to 2–3 s would be associated with the experience of duration and these would represent a second or subsequent interval scale related to the sense of time continuity. Across these longer intervals information would not be temporally fused and perceived in terms of a perceptual unity but would be separate events grouped together within the same moment of experience. A popular example is the experience of present time arising when listening to a

melody: when listening to the present tone, the previous tone is still in mind, while the coming tone is usually anticipated. Because past, present, and future tones are all momentarily present in mind, all of them are part of the subjective present. The past tone is nonetheless known as being past, and is thus both past (in objective time) and present (in subjective present time). Similar reasoning holds for the future tone, and this leads to the concept of specious present (James, 1896). Husserl (1893/1917) has proposed that the integration of past, present, and future represents a key mechanism in our sense of time continuity. It is not known to which amount and how the shorter 30 ms time windows are integrated within the experienced 3 s moments (Pöppel, 1997, 2009; Szelag et al., 2004). This requires an understanding of how windows overlap and integrate with one another (Elliott, 2005; Dainton, 2010). It might be proposed that temporal windows and their overlap are brought about by neuronal synchronization (Varela, 1999). Because neuronal synchronization of action potentials requires time, even the processing of the simplest event, such as the display of a square on a computer screen, is time-consuming and so coded within a temporal window of some duration. Different events would then overlap in time even if their onset is shifted in time. The result would be a sense of continuity rather than the perception of discrete moments.

There are thus several candidate mechanisms bringing about a sense of time continuity: the ability to relate past, present, and future moment, the hierarchical organization of elementary and longer time windows, or the overlap between successive elementary time windows. This means several possibilities to explain the disruption of the sense of time continuity in patients with schizophrenia. However, what we currently know about time continuity might not be enough to understand the mechanisms of its disruption and a number of questions first require clarification: we might ask whether events judged to be simultaneous are really processed as co-temporal? Further, are there other means that can bring about the sense of time continuity? Finally, at which time scale does the sense of time continuity emerge? These questions require answers if we want to understand why patients with schizophrenia suffer from a disrupted sense of time continuity.

We have chosen to focus on very short time scales below 100 ms, with the aim of assessing the elementary mechanisms that subtend processing of successive stimulus events in patients. A focus on short time scales was motivated by several results from the literature: elementary timing mechanisms have been related directly with neuronal synchronization (reviews in Elliott, 2005; Elliott et al., 2006; van Wassenhove, 2009) which seems appropriate given that schizophrenia is very precisely characterized by abnormalities in synchronization at frequencies in the EEG gamma band (i.e., at around 40 Hz), and thus correspond to intervals of around 25 ms (reviewed in Uhlhaas and Singer, 2010). In addition, impairments are also observed at a behavioral level that may be explained by elementary time disorders, such as in motor actions (Delevoye-Turrell et al., 2003; Carroll et al., 2009): for example, when participants lift an object or hit a target with an object, the peak grip force is usually synchronized with the time of impact or maximal load. In patients however, grip force is delayed by around 30–100 ms (Delevoye-Turrell et al., 2003). This means a difficulty at precisely synchronizing grip force in time. Another

example of abnormalities at short time scales can be found in visual perception tasks: it has been proposed that patients have difficulties to efficiently allocate attention in time, i.e., to focus attention at precise moments in time, when two stimuli follow each other at delays of between 50 and 250 ms Stimulus-Onset Asynchrony (SOA) (Granholm et al., 2009; Lalanne et al., 2012a).

We first explored the length of the elementary time window in patients with schizophrenia and found elementary time windows to be altered in patients (Giersch et al., 2009). However these disturbances led us to wonder about the processing of events within the time windows themselves. To address such questions, we devised a new method of analyzing participants' responses. The results question the usual assumption that events are treated as co-temporal within temporal windows, and lead us to reconsider the mechanisms underlying the sense of time continuity and to propose an explanation for its disruption in patients with schizophrenia. We summarize our findings in the following<sup>1</sup> .

#### **MEASURING THE WINDOWS OF VISUAL SIMULTANEITY**

Visual stimuli separated in time by an SOA of less than 20–30 ms are judged as being simultaneous. Our aim was to compare the length of the time window in patients with schizophrenia against that found in healthy participants. The paradigm used to determine this window is relatively simple: two visual stimuli (two bars or two squares, for instance) are shown on a computer screen. They appear either simultaneously or with a short asynchrony and participants decide whether the two stimuli are simultaneous or asynchronous. In our experiments, participants responded by pressing a left response key in case of simultaneity or a right response key in case of an asynchrony. To date, four studies have shown that patients require larger asynchronies than healthy participants to report two stimuli as appearing at different times (Foucher et al., 2007; Giersch et al., 2009; Schmidt et al., 2011; Lalanne et al., 2012b). Several possible confounds have been eliminated, i.e., the role of eye movements, interhemispheric transfer, modality specificity, bias, and access to consciousness. Hereafter, we review shortly how we can rule out these possibilities.

It has been shown that the impairment persists if patients are required to look at a central fixation point until the stimuli appear (Lalanne et al.,2012b) indicating that the impairment is not related to abnormal eye movements (Phillips and Davis, 1994; Holzman, 2000; Trillenberg et al., 2004). In addition, impairments are similar when stimuli are displayed within the same hemifield as across hemifields (Lalanne et al., 2012b), thus eliminating the influence of impaired interhemispheric connections (Schwartz et al., 1984; Mohr et al., 2008; Knöchel et al., 2012; but see David, 1993). A difficulty to discriminate simultaneous from asynchronous events

is also observed in the auditory (Foucher et al., 2007) as well as in cross-modal conditions (i.e., audio-visual, Martin et al., 2013) and finally, a possible decisional bias has been carefully ruled out by both Giersch et al. (2009) as well as Schmidt et al. (2011). A decisional bias is independent of perceptual ability and may occur during response selection: in this case it might be that patients process asynchronies as do the healthy participants but need larger asynchronies to select an"asynchronous"response. Classical methods based upon signal detection theory (Green and Swets, 1974) showed no difference between the bias of patients and healthy participants (Giersch et al., 2009; Schmidt et al., 2011).

We used an additional method, based upon priming, which bypasses the problem of an explicit judgment on the part of the patients. The possibility of impairment due to the explicit nature of the response is distinct from a decisional bias. A deficit at making an explicit judgment is related to the mechanisms allowing information to become available to consciousness (Del Cul et al., 2006), whereas decisional biases are related to variations in the use of information to give a response. Each time a patient is explicitly asked to make a judgment, and shows abnormal performance, it can be asked whether the impairment is related to the difficulty the patient experiences in formulating his/her judgment. Patients have often been shown to be thus impaired although implicit information processing remains unimpaired (Del Cul et al., 2006). In the present context an apparent difficulty to detect an asynchrony between two items might be due to a non-specific difficulty at making a subjective judgment rather than impairment in processing the timing of events. In our task, we used a procedure developed by Elliott et al. (2007) and took an implicit measure of the effects of an asynchrony on a subsequent explicit judgment of simultaneity/asynchrony. In this paradigm two priming bars were displayed on a computer screen, either simultaneously or asynchronously while a series of six distracter bars were rapidly switched on and off around the priming bars, thus making the temporal relation between the priming bars impossible to accurately report. After the distracters switched off the priming bars remained on screen and after a short interval increased in luminance separately and with a variable SOA (which included a simultaneous increase) (**Figure 1**). Participants reported whether this luminance increase was simultaneous or asynchronous across the two bars.

Elliott et al. (2007) showed that in healthy volunteers, the masked asynchronous bars biased participants toward reporting an asynchrony in the subsequent increase in bar luminance. We showed, in addition, that this bias increases with the asynchrony between priming bars: the larger this asynchrony, the larger the bias (Giersch et al., 2009). This was observed even though the asynchrony in the prime was individually adapted so that it was below threshold in all cases,indicating that the effect is due to the implicit priming of the asynchrony. We reasoned that if the difficulties of patients are due to impaired explicit judgments *per se*, there should be dissociation between implicit and explicit processing of the asynchrony: explicit judgments of asynchrony would be impaired while their implicit influence would remain unaffected. If this were the case, we expected priming to be identical between groups for equivalent sub-threshold asynchronies. In our case however, subthreshold asynchronies were derived from explicit judgments and adapted individually. As a consequence they were not equivalent

<sup>1</sup>The patients with schizophrenia participating to our studies (around 20 in each study) were mildly symptomatic outpatients, usually treated with antipsychotics but without benzodiazepines. Their performance was compared with those of control participants individually matched on sex, age, and education level (Giersch et al., 2009; Lalanne et al., 2012a,b,c). We cannot firmly exclude any effect of antipsychotic medication, although dosages do not correlate with performance. In addition, it should be noted that clinical descriptions of disrupted time continuity predate the establishment of antipsychotic medication. What's more, the mechanisms of the sense of continuous time, as well as impairments require clarification even in treated patients. All subjects gave an informed consent, which is archived by the first author.

across groups, i.e., the asynchronies used to test the priming effect were larger in patients. Since the priming effect increases with the amplitude of the sub-threshold asynchronies, difficulty restricted to explicit judgments should have resulted in larger priming effects in patients with schizophrenia. But this was not what was observed. Instead, near identical priming effects were found for each group (**Figure 2**). This indicates that the enlarged temporal window observed in patients presents a true difficulty at discriminating events over time that is not explicable in terms of a non-specific difficulty in formulating the required judgment. It is to be noted that these results did not allow us to compare the implicit processing of asynchronies in patients and controls, since sub-thresholds asynchrony were not equivalent across groups. The results mainly suggest that patients require larger asynchronies than controls to reach the same perceptual level (i.e., a level that yields a priming effect), independent of the need to give an explicit response.

Inasmuch as the subjective experience of the present is built upon elementary time windows, it can be expected that a lengthened interval of subjective simultaneity distorts the sense of time continuity. Before drawing this conclusion however,we questioned whether this lengthened interval entailed the temporo-perceptual fusion of all the events within the associated temporal window, i.e., whether it meant that events are treated as co-temporal even at an implicit level. It was indeed surprising to observe the amplitude of the deficit in some patients with schizophrenia, who were unable to detect an asynchrony of more than 100 ms. Did this really mean that patients fuse events in time over periods longer than 100 ms? The next step was thus to check the integrity of implicit information processing within the window of simultaneity.

#### **IMPLICIT PROCESSING OF INFORMATION WITHIN THE TEMPORAL WINDOW**

In order to evaluate the quality of implicit information processing within the 55 ms integration window, we devised an original method of analysis, by examining the Simon effect. The Simon

effect refers to the speeding of and more accurate responses when a visual stimulus is presented within the same perceptual hemifield as the responding hand (Hommel, 2011a). The precise mechanisms of the effect are a matter of discussion (i.e., Hommel, 2011a,b; van der Lubbe and Abrahamse, 2011), but the important point is that the effect is independent of the explicit instructions given to the participant. In our paradigm, it allowed us to check whether or not the stimuli are fused in time when the asynchrony is not consciously detected. During our task, two stimuli are displayed on the screen, one to the left and one to the right and participants give their response – "simultaneity" or "asynchrony" by pressing the left or right response key, respectively. When the two stimuli are displayed simultaneously, a Simon effect cannot occur: the participants cannot be biased to respond on any particular side since the displayed information is equivalent to both sides. However, when the stimuli are asynchronous there is an asymmetry and under these conditions we observed a Simon effect (Lalanne et al., 2012b,c). Healthy participants were systematically biased to answer to the side of the second stimulus independent of its right or left location and independent of the asynchrony amplitude (Lalanne et al., 2012c). Given that physical information is identical on both sides, it is only the temporal difference between the two stimuli that can lead to such a Simon effect. The important point is that this Simon effect was observed even at short asynchronies when participants reported the majority of stimulus

presentations to be simultaneous (Lalanne et al., 2012b,c). This suggests that the asynchrony is processed even though participants are unable to report it. Interestingly, a Simon effect was also observed in patients with schizophrenia. At large asynchronies patients were biased to the side of the second stimuli to the same extent as were the healthy participants (Lalanne et al., 2012b). However at small asynchronies, patients were biased to the side of the first stimulus although the controls still showed a bias toward the second stimulus (**Figure 3**; Lalanne et al., 2012b,c). This effect has been observed in three different studies, with three different groups of 18–20 patients with schizophrenia. It has been observed for asynchronies of 8–17 ms, i.e., at delays that lead to rates of "simultaneous" responses identical to the rates observed for perfect synchrony (Lalanne et al., 2012c). As with healthy participants, this means that the asynchronies are processed in patients with schizophrenia even for delays of below 20 ms and are not fused in time. Had the stimuli been fused in time, then the two stimuli would have been processed as if identical. In this case, the Simon effect could not have occurred, i.e., the subjects could not have been biased to either stimulus. These results show that patients process very short asynchronies similarly to (but not quite in the same way as) healthy participants even though they need much larger asynchronies to explicitly report them. The Simon effect observed at asynchronies below 20 ms can be considered as implicit not only because the effect is independent of the instructions, but also because asynchronies below 20 ms are not detected and cannot drive a response bias. This contrasts with the Simon effect observed at larger asynchronies: the perception of a temporal order between two stimuli that are clearly asynchronous might drive subjects to switch attention toward the last occurring event. This cannot occur at short asynchronies. All in all this means that the Simon effect differentiates patients from controls only at short asynchronies, when temporal processing is implicit. These data show a clear dissociation between the implicit response – indicated by the Simon effect occurring at 8 and 17 ms asynchronies, and the subjective experience of temporal relations between stimuli occurring at larger asynchronies – indicated by the explicit judgments of those relations. A similar dissociation has also been observed with multisensory stimuli (Martin et al., 2013). Our results do not indicate a preserved implicit processing however: on the contrary, the results in patients with schizophrenia differ qualitatively from those in healthy participants: at short asynchronies, the responses of healthy participants are biased to the side of the second stimulus whereas the responses of patients are biased to the side of the first. Hence and consistent with our previous studies (Giersch et al., 2009) it seems that both the implicit and explicit processing of asynchronies is affected in patients with schizophrenia: the implicit impairment is revealed by the Simon effect at short asynchronies, and the explicit impairment reflects in the difficulty to explicitly report an asynchrony.

The critical question at this stage is the meaning of the bias to the side of the first or second stimulus. Bias to the side of the second stimulus might be related with studies showing that temporal coding is more precise for events offsets than onsets (Bair et al., 2002; Clifford, 2002; Tadin et al., 2010). Although there is no offset in our studies since stimuli stay on the screen until subjects give their response, a stimulus offset and the second stimulus in

our study both represent the end of an event. This bias toward an event's end initially seems surprising given the effect of prior entry. The latter has been demonstrated using similar tasks involving simultaneity/asynchrony or temporal order judgments on successive or simultaneous stimuli (review in Spence and Parise, 2010): it has been frequently shown that cueing to the first stimulus in a sequence of two facilitates subsequent judgments of order or of simultaneity/asynchrony. A cue,for example an indicator or a flash encourages deployment of attention to the cued location. This is believed to expedite processing of the first stimulus and thus facilitates detection of an asynchrony, or of a succession between first and second stimuli (Spence and Parise, 2010). The prior effect thus shows that focusing attention on the first stimulus in a sequence is important for the processing of sequential order. By contrast, we found that healthy participants were biased to the second stimulus. The major differences between the two observations is, first, that our observations take place when subjects are not aware of any asynchrony, and second, that prior entry takes place before the critical stimuli are presented. In contrast the Simon effect is recorded at the time of the participant's response, i.e., after all stimuli are presented. From these observations it is tempting to conclude that healthy participants proceed from processing the first to the second stimulus during the sequence as if biased toward the second. Conversely, patients remain fixed on the first stimulus.

A number of difficulties and questions still remain with this interpretation: first it might be possible that deployment of visuospatial attentional mechanisms rather than any disorder in temporal processing accounts for the difficulties in task performance presented by patients with schizophrenia. Given that the two stimuli differ not only on their time onset but also on their spatial location,it might be asked whether patients have a difficulty to shift attention from the first to the second stimulus as a result of spatial or visual organization impairments in attentional mechanisms. In fact and related to this, patients with schizophrenia are known to be impaired at grouping items when those items are spatially separate (review in Silverstein and Keane, 2011) and this may make it difficult to sequentially process successively presented stimuli. To check for this possibility, we used the fact that patients' difficulties with perceptual grouping are alleviated when clear grouping cues relate the stimuli (Giersch and Rhein, 2008; van Assche and Giersch, 2011). We adapted our paradigm and added line-segments between the stimuli in order to facilitate grouping (**Figure 4**).

We also manipulated the spatial predictability of the second stimulus to further facilitate a shift in processing toward this stimulus in patients: in one experiment there was only two possible locations for the stimuli and the second stimulus location was always predictable, whereas in a second experiment, four locations were used, and the location of the second stimulus was uncertain (**Figure 5**).

Since perceptual grouping promotes the perception of simultaneity (Nicol and Shore, 2007), it was expected that all Simon effects would be reduced by making right and left stimuli more similar in their temporal properties. As expected, the bias shown by healthy participants' to the side of the second stimulus was reduced when stimuli were connected (Lalanne et al., 2012c). On the other hand, the Simon effect to the side of the second stimulus

was increased when the location of the second stimulus was predictable. The results in healthy participants thus confirmed that our experimental manipulations had the expected effect. However in patients there was a very clear bias to the side of the first stimulus when the first and second stimulus were related with a line-segment, and when the second stimulus' location was predictable in 100% of the cases (**Figure 6**).

This suggests that in patients the abnormal bias to the side of the first stimulus persists when grouping and spatial difficulties are alleviated by experimental manipulations. Overall and in summary, these results suggest difficulties related to time rather than to spatial and visual organization impairment.

It should be noted that it is unlikely that the implicit processing of asynchronies involves a coding of succession. This point is critical when considering the sense of time continuity. As emphasized above, it has been proposed that the sense of time continuity arises from the integration of past, present, and future moments within the subjective present (Husserl, 1893/1917). Succession is a way of establishing a link between past, present and future events, and it would represent a mechanism of integration. If this occurs within elementary time windows, it would mean that integration of past, present and future takes place on a shorter time scale than previously believed. Several empirically founded arguments speak against this possibility: first, the physical characteristics of the stimuli, i.e., their spatial and temporal separation make it unlikely that the Simon effect is mediated by a coding of motion between the two successive stimuli (see Lalanne et al., 2012c,for a complete discussion on this point). Besides, it would be surprising that events are coded one relative to another on time scales shorter than 20 ms. Automatically ordering events in time might indeed be costly. In our experiments stimuli were sometimes displayed in two different hemifields and thus processed in different cerebral hemispheres

**FIGURE 5 | Illustration of the successive events in the simultaneity/asynchrony discrimination task in case of an asynchrony (from left to right)**. When four locations are used (upper row) and the first square is filled in (middle figure in the upper row), there are two possible locations for the second one (figure on the right), and there is thus an uncertainty regarding the location of the second stimulus. The spatial location of the second stimulus is always predictable when only two locations are used (lower row). These two experiments have been conducted in two different groups of participants (Lalanne et al., 2012c).

**FIGURE 6 | Amplitude of the bias (in %) to the side of the first or second stimulus for an asynchrony of 17 ms**. A negative bias corresponds to a bias to the side of the first stimulus (in patients), whereas a positive bias corresponds to a bias to the side of the second stimulus (in healthy participants). In the present experiment, there were two possible locations for the squares, one on the left part and one on the right part of the screen. As a consequence, the location of the second square was always predictable (similar results were observed when the location of the second square was more uncertain, Lalanne et al., 2012c). The squares were either connected (the displayed graph) or unconnected (data not shown, but see Lalanne et al., 2012c).

(e.g., **Figure 2**). Comparing their onsets is bound to involve longdistance connectivity (at least to transfer information regarding the stimuli onset) and some specialized comparison mechanisms. If this is to be generalized in the natural environment, it would mean permanent comparisons between unrelated stimuli. The usefulness of such computation might be questioned. In addition the computation cost would increase exponentially in a crowded environment. It might thus be proposed that even if events delayed by short asynchronies are not processed as being co-temporal, their succession is not coded automatically. This possibility is supported by our most recent results in healthy participants (Giersch et al., in press).

Even if the lack of succession coding makes it impossible to tag stimuli as "before" and "after," the second stimulus might still be identified as the last occurring event. In fact, a bias to the side of the second stimulus in a succession of two suggests the existence of a mechanism assigning priority to the last occurring event. This might be analogous to what happens when attention is driven reflexively by an external event. It is known that attention deploys to novel information, meaning that associated brain mechanisms are designed to continually check for novel information, not only in space but also in time. Feedback loops described in visual perception, predictive coding, or forward models might provide neural bases for these mechanisms (Miall et al., 1993; Wolpert et al., 1998; Elliott and Müller, 2000; Kompass and Elliott, 2001; Friston, 2008). Although the results suggest some kind of prioritization and possibly an involvement of attention, the processes making this prioritization possible are not available

to consciousness. Participants do not perceive very short asynchronies, and they put nonetheless more weight on the second and last stimulus. Even if this entails the involvement of attention, it cannot be induced by conscious expectation.

What is the relationship between the ability to distinguish stimuli at very short asynchronies and the sense of time continuity? In so far as succession is coded only at longer delays, the integration of past, present and future moments, proposed as a mechanism of "time continuity" defines only delays of that order of magnitude. However, what is observed at the shortest delays is an automatic priority for the latest events, possibly sub-tended by mechanisms allowing us to permanently look ahead or anticipate future events. It might be asked whether this bias to anticipate future events also participates in the sense of continuity, thereby providing an elementary basis for the expectation of what is coming next. Husserl described the concept of a protention,which allows us to anticipate the future during the present time, an implicit, temporally defined form of which has been demonstrated experimentally by Kompass and Elliott (2001). What Husserl describes might be more easily related to conscious phenomena taking place at larger time scales, but the bias toward the latest stimulus might nonetheless be considered as an elementary mechanism subserving protention.

Related to that, it can be asked whether patients' disrupted sense of time continuity is due to their difficulty at assigning priority to the latest occurring events. This is certainly reminiscent of the observations of Minkowski (1933), p. 259, who gave the following description:"Touché dans son dynamisme vital, le schizophrène non-seulement sent tout s'immobiliser en lui, mais est encore comme privé de l'organe nécessaire pour assimiler ce qui est dynamisme et vit au dehors," i.e., "Not only does the patient with schizophrenia, who is affected in his vital dynamism, feel everything as coming to a halt inside him, but he also seems restricted in the very organ allowing assimilation of what is both dynamic, and exists outside."

It might be asked how patients perceive rapid succession and so how they are able to follow a stream of incoming information over time. The impact on duration perception should also be investigated. Inasmuch duration perception relies on the accumulation of information as time moves forward, an inability to follow information over time can be expected to disturb perception duration. However, duration perception concerns time scales that are much longer than those studied here, and theoretical models relating these different time scales are missing (vanWassenhove, 2009; Grondin, 2010). More generally, the consequences of an elementary impairment at moving attention over time is necessarily speculative at this stage.Yet some possibilities seem likely in light of the known impairments described in patients with schizophrenia, and are briefly evoked in the next section.

# **POSSIBLE IMPACT OF ELEMENTARY TIMING IMPAIRMENTS ON VISUAL PERCEPTION**

The visual environment is organized in both time and space, and several studies have shown that patients have a difficulty at organizing information in space. Our results suggest that their inability to follow and expect events over short time periods is independent of spatial impairments. It is possible that patients have specific impairments at binding information together both

in space and time. However, timing difficulties might aggravate spatial difficulties in different ways.

In every day life, the outer world is usually dynamic rather than static and events succeed each other rapidly. Wind can make a tree leaves move, alternately uncovering, or occluding visual information. We usually experience no difficulty in distinguishing the tree from the objects located behind it. Yet moving objects like the moving leaves create much ambiguity in the organization of information, and some processing is required in order to attribute elements of information to the right objects. In dynamic environments, this requires discrimination in both space and time: an item has to be precisely focused in time in order to avoid confusion with distracter information, especially if it is displayed in the same location but at a different time. As a matter of fact, the use of masking experiments has shown that patients with schizophrenia have difficulties at distinguishing target information when it is closely preceded or followed by a distracter (Saccuzzo and Braff, 1981; Green andWalker,1986;Rund,1993;Cadenhead et al.,1998;Butler et al., 2003; Schechter et al., 2003). Explanations for this impairment are diverse (Breitmeyer, 1984; Schuck and Lee, 1989; Green et al., 1994, 2011a,b; Bedwell et al., 2003; Butler et al., 2003), but the impairment certainly confirms that patients have a difficulty with stimuli shown in close succession. Besides, based on masking experiments, it has been proposed that patients with schizophrenia have a difficulty in the temporal precision of target-directed attention (Granholm et al., 2009; Lalanne et al., 2012a). All in all, temporal difficulties mean heightened difficulties at resolving ambiguities arising in case of dynamic visual information.

Temporal impairments may also contribute to the patients' distractibility. A Simon effect to the side of the first stimulus might indeed be similar to attentional capture, inasmuch it impedes patients to move forward in time. Such an interpretation would be consistent with a series of studies suggesting that patients are abnormally sensitive to sudden information onsets (Schwartz and Winstead, 1982; Schwartz et al., 1988; Schuck and Lee, 1989; Ducato et al., 2008), possibly related with enhanced magnocellular sensitivity (see also Laprévote et al., 2010). This hypothesis had been discarded in the literature because several studies suggested difficult detection of information known to be conveyed by the magnocellular pathway, i.e., information with a high content of low-spatial and high-temporal frequency (Butler et al., 2001,

#### **REFERENCES**


the color red tell us? *Schizophr. Res.* 63, 273–284.


2005; Butler and Javitt, 2005; Calderone et al., 2013, but see Skottun and Skoyles, 2007). But in fact, the results taken as a whole are reminiscent of the dissociation described above comparing explicit and implicit responses: in this case patients were excessively sensitive to the first stimulus-onset, but this effect was apparent only at an implicit level (the Simon effect) and did not help patients to make explicit judgments regarding stimuli asynchronies. This might mean that automatic response of the magnocellular pathway is increased in patients, whereas the conscious experience of the information conveyed by this pathway is impaired. The important point here is that our results would renew the hypothesis that patients are abnormally captured by information onsets, thus loosing pertinent information.

Taken together these two lines of reasoning point toward difficulties in organizing and following the flow of information. This can be expected to destabilize the patients' representations of the outer world, and impair them in distinguishing pertinent from non-pertinent information. Further studies are required to check to which extent the basic impairments reviewed in the present work account for known difficulties in patients.

# **CONCLUSION**

Although many questions remain, our results show that information is not fused in time at very short asynchronies, either in healthy participants or in patients. Our results also indicate that healthy participants move in time very rapidly between succeeding events and that this capability is impaired in patients. These observations may mean new additional and non-conscious mechanisms underlying the sense of time continuity. It remains to be investigated whether and how these mechanisms are really involved in the sense of time continuity, and how the impairments of these mechanisms impact cognitive functions and clinical symptoms in schizophrenia.

#### **ACKNOWLEDGMENTS**

The research reviewed in this manuscript was supported by the French National Institute for Health and Medical Research (INSERM), the Centre Hospitalier Régional Universitaire of Strasbourg (API-HUS n˚3494), the Medicine Faculty of Strasbourg, and the french National Research Agency (ANR SaMenta Causamap n˚ANR-12-SAMA-0016-03).

in schizophrenia. *Curr. Opin. Psychiatry* 18, 151–157.


backward-masking deficits of schizophrenia patients. *Biol. Psychiatry* 43, 132–138.


are biased toward low spatial frequency to decode facial expression at a glance. *Neuropsychologia* 48, 4164–4168.


for further studies. *J. Abnorm. Psychol.* 120, 476–482.


space. *Philos. Trans. R. Soc. Lond. B Biol. Sci.* 364, 1815–1830.


Wolpert, D. M., Miall, R. C., and Kawato, M. (1998). Internal models in the cerebellum. *Trends Cogn. Sci.* 2, 338–347.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 31 January 2013; accepted: 02 May 2013; published online: 28 May 2013.*

*Citation: Giersch A, Lalanne L, Assche Mv and Elliott MA (2013) On disturbed time continuity in schizophrenia: an elementary impairment in visual perception? Front. Psychol. 4:281. doi: 10.3389/fpsyg.2013.00281*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Giersch, Lalanne, Assche and Elliott. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Negative correlation between leftward bias in line bisection and schizotypal features in healthy subjects

# *Michele Ribolsi\*†, Giulia Lisi †, Giorgio Di Lorenzo , Giuseppe Rociola , Cinzia Niolu and Alberto Siracusano*

*Clinica Psichiatrica, Dipartimento di Medicina dei Sistemi, Università degli Studi di Roma Tor Vergata, Rome, Italy*

#### *Edited by:*

*Randolph Blake, Vanderbilt University, USA*

#### *Reviewed by:*

*Christine Mohr, University of Lausanne, Switzerland Bruce K. Christensen, McMaster University, Canada*

#### *\*Correspondence:*

*Michele Ribolsi, Clinica Psichiatrica, Dipartimento di Medicina dei Sistemi, Università degli Studi di Roma Tor Vergata, Via Nomentana 1362, 00137 Rome, Italy e-mail: michele.81@live.it*

*†These authors have contributed equally to this work.*

**Introduction:** Recent studies have found a lack of normal pseudoneglect in schizophrenia patients and in their first degree relatives. Similarly, several contributions have reported that measures of schizotypy in the healthy population may be related to signs of rightsided lateralization, but most of these studies differ greatly in methodology (sample size, choice of schizotypy scales, and laterality tasks) and, consequently, the results cannot be compared and so definitive conclusion cannot be drawn. In this study, our purpose is to investigate whether some tasks of spatial attention may be related to different dimensions of schizotypy not only in a larger sample of healthy subjects (HS), but testing the same people with several supposedly related measures several times.

**Materials and Methods:** In the first part of the study (Part I), the performance on "paper and pencil" line bisection (LB) tasks in 205 HS was investigated. Each task was repeated three times. In the second part of the study (Part II), a subgroup of 80 subjects performed a computerized version of the LB test and of the mental number line bisection (MNL) test. In both parts of the study, every subject completed the 74-item version of the Schizotypal Personality Questionnaire (SPQ) and the Edinburgh Handedness Inventory (EHI).

**Results:** In both parts of the study, high scores on the subscale "magical thinking" of SPQ have resulted in being closely linked to a decreased pseudoneglect as assessed by the LB task. On the contrary, right handedness is related to an increased leftward bias at the same task. No association was found between MNL and the other variables.

**Discussion:** The main finding of this study is that a decreased spatial leftward bias at the LB task correlates with positive schizotypy in the healthy population. This finding supports the hypothesis that a deviation from leftward hemispatial visual preference may be related to the degree of psychosis-like schizotypal signs in non-clinical population and should be investigated as a possible marker of psychosis.

**Keywords: pseudoneglect, line bisection, mental number line, schizotypy, psychosis**

# **INTRODUCTION**

Lateralization of cognitive functions between the right and left hemispheres is known to be a prominent feature of the human brain (Kosslyn, 1987; Hellige, 1990; Gazzaniga, 2000; Crow, 2010). For instance, it is known that the cortical networks of the right hemisphere involving the posterior parietal cortex play a dominant role in visuo-spatial attention, so that right hemisphere lesions often induce visuo-spatial neglect, a severe neurological disorder characterized by failure to acknowledge or explore stimuli presented to the contralesional side of space (Bisiach and Luzzatti, 1978; Heilman et al., 2000; Vallar et al., 2003). The most commonly used technique for detecting the presence of unilateral spatial neglect is the line bisection (LB) test: the patient is asked to place a pencil mark at the center of a series of horizontal lines. Displacement of the bisection mark toward the side of the brain lesion is interpreted as a symptom of neglect (referred to as perceptual neglect). Moreover, the phenomenon known as "pseudoneglect" (Bowers and Heilman, 1980) refers to the systematic leftward misbisection of horizontal lines made by neurologically intact observers. The magnitude of this bisection error is much smaller than in neglect patients but studies have widely shown that neglect and pseudoneglect are closely related (McCourt and Jewell, 1999) and possess similar susceptibilities to a variety of modulating factors. For example, both the magnitude and the direction of the bisection errors in pseudoneglect are modulated by stimulus or task factors (e.g., line length, line location, task instructions) (McCourt and Jewell, 1999; Fink et al., 2002) which may also influence the magnitude and the direction of the visual neglect (Marshall and Halligan, 1989; Seki et al., 1999). There is also evidence supporting the existence of both representational neglect and pseudoneglect, which can be detected through the mental number line bisection (MNL) test, in which numbers are conceived as falling along a mental number line spatially oriented from left to right and the subject is asked to bisect a numerical interval (Heilman et al., 2000; McGeorge et al., 2007). The LB test and the more recent MNL bisection test (Bisiach and Luzzatti, 1978) have different underlying mechanisms, the MNL is in fact mostly related to a purely abstract non-spatial representation of an imaginary number line (Aiello et al., 2013). Different anatomical areas are thought to be involved in these tasks: neuroimaging studies demonstrated that while physical LB depends on the striate, the extrastriate visual cortex and the inferior and superior parietal lobe, comparative judgments of numeric quantities activate prefrontal areas (Doricchi et al., 2005; Tian et al., 2011).

Interestingly, various psychopathological conditions may influence the expression of pseudoneglect (Rao et al., 2010). To our knowledge, in the context of schizophrenia research, six studies (Mather et al., 1990; Barnett, 2006; Michel et al., 2007; Zivotofsky et al., 2007; McCourt et al., 2008; Ribolsi et al., 2013) have investigated perceptual pseudoneglect and only three studies representational pseudoneglect (Cavezian et al., 2007; Tian et al., 2011; Ribolsi et al., 2013).

Concerning representational pseudoneglect, Cavezian and colleagues found an exaggerated leftward bias in the MNL of the schizophrenia patients (SCZ) in comparison to the healthy subjects (HS) (Cavezian et al., 2007), while two more recent studies found no difference between the two groups (Tian et al., 2011; Ribolsi et al., 2013).

However, concerning the relationship that occurs between schizophrenia and perceptual pseudoneglect, two studies reported a leftward bias in the SCZ sample (Mather et al., 1990; Michel et al., 2007) while all the others have provided evidence of a significant lack of leftward bisection error in the LB test (Barnett, 2006; Zivotofsky et al., 2007; McCourt et al., 2008; Ribolsi et al., 2013). This result has been hypothesized as being linked to a reduced or reversed brain asymmetry with a deficit of right hemisphere functions (Michel et al., 2007; Zivotofsky et al., 2007; McCourt et al., 2008; Rao et al., 2010; Ribolsi et al., 2013) and in particular of the right parietal cortex in SCZ (Petty, 1999; Malhotra et al., 2009; Venkatasubramanian et al., 2011). Intriguingly a dysfunction of this area has been linked to schizophrenia in a structural, neurophysiological and functional way (Zhou et al., 2007; Kato et al., 2011; Venkatasubramanian et al., 2011). Interestingly, in a recent study conducted by our research group, selective transcranial direct current stimulation (tDCS) of right posterior parietal cortex was able to determine a partial correction of the lack of leftward bias in a group of medicated SCZ, confirming the hypothesis of the involvement of this area in the onset of this phenomenon (Ribolsi et al., 2013).

In this study, our purpose is to investigate whether the LB and MNL performances may be related to different dimensions of schizotypy in a large sample of HS. In particular, our purpose is to investigate the hypothesis of a continuum between schizophrenia and schizotypy not only on a phenomenological and genetic level (Nelson et al., 2013) but also in measures of visuo-spatial attention and lateralization.

Schizotypy is a psychological construct that describes temporally stable personality characteristics and specific perceptions, cognition, beliefs, and experiences that are phenomenologically similar to, but less severe than, the symptoms of schizotypal personality disorder and schizophrenia (Meehl, 1962; Shaw et al., 2001). This condition resembles schizophrenia not only in terms of observable symptoms but also in the underlying multidimensional structure (Gruzelier, 1996). There is a wide variety of questionnaires designed for the assessment of schizotypy: the most widely used are the Wisconsin-Madison scales (Chapman and Chapman, 1980), the Schizotypal Personality Questionnaire (SPQ) (Raine, 1991) and the Oxford-Liverpool Feelings and Experiences Questionnaire (O-LIFE) (Mason and Claridge, 2006).

Unfortunately, till now the studies on the link between schizotypy and functional hemispheric asymmetry have been inconclusive, with some of them reporting a right-over-left hemisphere shift (Kravetz et al., 1998; Suzuki and Usher, 2009), others a leftover-right hemisphere shift (Mason and Claridge, 1999; Liouta et al., 2008), and yet others finding no relation between schizotypy and laterality (Najt et al., 2012). All these studies used different measures of laterality (language and visuo-spatial attention tasks).

In the debate concerning the relation between schizotypy and measures of laterality (Schofield and Mohr, 2013) only a few studies have examined whether pseudoneglect is related or not to a specific dimension of schizotypy. Results are controversial. Using a tactile rod bisection task, Brugger and Graves (1997) highlighted a right-sided inattention and reported a significant association between the size of this right-sided inattention and high magical ideation scores in male participants. Nalcaci et al. (2000), evaluating performances on Corsi's Block-Tapping Test, also reported an association between right hemispatial inattention and high magical ideation scores. These findings have been interpreted as supporting the notion that a deficit in left hemispheric functioning underlies schizotypy. Contradicting results are reported by Liouta and colleagues, who reported an association between schizotypy and rightward hemispatial bias using two different spatial behavior tasks (Liouta et al., 2008). Interestingly, Mohr and colleagues, using three different kinds of spatial behavior tasks (LB "paper and pencil," whole-body turns, and veering behavior when attempting to walk in a straight line while blindfolded), reported that high scores in magical ideation scales were linked, on the one hand, to right-sided inattention at lateralized whole-body movement tasks (turning and veering) and on the other hand to a lack of pseudoneglect in conventional "paper and pencil" LB tasks (Mohr et al., 2003).

In conclusion, as shown in **Table 1**, what emerges from today's literature is a multifaceted research panorama in which the studies conducted are very different in methods (choice of schizotypy scales and laterality tasks) and, consequently, the results and conclusions are not fully comparable. In this context, the aim of our study is to investigate the supposed correlation between different dimensions of schizotypy, LB and MNL performances not only in a larger sample of HS, but testing the same people with several supposedly related measures several times. Given the current literature, we would expect to find a relation between a decreased pseudoneglect and the degree of schizotypal traits in HS. Furthermore, in our study we hypothesize that, unlike the LB test, the performance in the MNL test is not influenced by schizotypy, probably because of its non-spatial origin (Van Dijck et al., 2011).

**Table 1 | Summary of the studies that investigated schizotypy and spatial tasks.**


#### **MATERIALS AND METHODS**

#### **SUBJECTS**

In Part I, 205 HS were recruited [136 women; age (*SD*): 34.78 (12.74)], while in Part II 80 HS were recruited [46 women; age (*SD*): 32.23 (11.05)].

The subjects were recruited by personal contact and flyers posted at the University Hospital of Tor Vergata, Rome. All the subjects had a university education.

In both parts, each subject underwent a specialistic neurological and psychiatric consultation before recruitment. Subjects were excluded from participating if they exhibited any neurological or ophthalmological disorders (as assessed by a careful neurological examination), a history of head trauma (as reported by the subjects), or if they met criteria for substance dependence within the previous 6 months, or substance abuse within the month preceding testing. Moreover, a diagnosis of any psychiatric disorder was excluded by means of consultations with physicians and the Structured Clinical Interview for *Diagnostic and Statistical Manual of Mental Disorder, Fourth Edition.* All subjects gave written informed consent for the study. The experimental procedures used were approved by the local Ethics Committee and were carried out in accordance with the Declaration of Helsinki. Finally, for each subject handedness was ascertained by the Edinburgh Handedness Inventory (EHI) (Oldfield, 1971). The EHI is a valid and reliable quantitative measurement tool that assesses a participant's hand, eye and foot preference for 12 tasks (Ransil and Schachter, 1994). Finally, each subject underwent the 74-item version of the SPQ to evaluate the presence of schizotypal traits. The SPQ is a 74-item, forced choice, self-report questionnaire, which yields nine subscales designed to give a dimensional assessment of the Schizotypal Personality Disorder features listed in the DSM-III-R (APA, 1987): ideas of reference, excessive social anxiety, odd beliefs, or magical thinking, unusual perceptual experiences, odd, or eccentric behavior, no close friends, odd speech, constricted affect, suspiciousness (Raine, 1991). In our study, we used the Italian version validated by Fossati et al. (2003).

Although several factor analytic studies of this measure have been conducted (Linscott, 2013), we chose to use the SPQ subscales in order to investigate every specific feature of schizotypy according to the DSM-III-R rather than grouping the different dimensions. Similarly, various previous studies have selectively investigated the relation between a single specific dimension of schizotypy and measures of pseudoneglect (Brugger and Graves, 1997; Kalaycioglu et al., 2000; Taylor et al., 2002; Mohr et al., 2003; Brugger et al., 2010). In this regard, recent studies have reported that the single SPQ subscales may be useful in screening for schizotypal traits in the general population (Bedwell et al., 2013; Salokangas et al., 2013).

#### **PART I**

Subjects were seated comfortably at a writing table and the pages (size A4) containing the tasks were placed in front of them. They were told to follow the instructions written in Italian at the top and then to fixate the line in the paper. All tasks were to be performed using the dominant hand. The lines to be divided in the bisection task were centered on each page and located below the midline. The lines were 125 mm-long black lines on an otherwise blank white page. Every page instructed the subject to "divide the line in half as accurately as possible." Participants were asked to bisect each line into two equal lengths using a pencil. In order to compute scores, each line was measured to the nearest millimeter. The deviation from the center of the line was calculated as the absolute error in mm. Negative values indicate leftward bias and positive values a rightward bias. The task was repeated three times on three different days (a total of nine lines for each subject) and the mean deviation was calculated.

# **PART II**

In this part of the study, the subjects enrolled underwent a computerized version of the LB test and of the MNL test. E-Prime® (version 2.0, Psychological Software Tools, Inc.; http://www*.* pstnet*.*com) software was used to create computerized versions of the two pseudoneglect protocols. Stimulus presentations and data collection were performed on a 15.4-laptop computer screen.

#### *Line bisection*

Visual stimuli consisted of black 1 mm-thick horizontal lines transected by a 1 mm-thick and 1 cm-long vertical bar, presented on a white background with the transector positioned exactly in the center of the screen. A modified version of LB created by Fierro et al. (2000) was performed to detect LB in medicated psychiatric patients. Stimuli were presented for 750 ms. Three lines of 15 cm were presented, differing in the position of the transector (at midpoint, rightward, or leftward). Subjects were given 30 trials in random order, 10 with the transector at the exact center (7.5 cm), 10 with a rightward transector at 8.0 cm and 10 with a leftward transector at 7.00 cm. Interstimulus intervals were 3750 ms.

Subjects were seated at a distance of 45 cm from the laptop screen and were asked to focus on a centrally positioned fixation cross that disappeared as soon as the three numbers were presented.

Participants judged the position of the transector in prebisected lines by pressing one of three buttons with the right index, middle, or ring finger for "left," "equal," or "right" responses. The performance of the subject on each trial was scored as follows: 0, correct response; 1, if the subject judged the transector to be right of its real position; –1, if the subject judged the transector to be left of its real position.

#### *Mental number line*

Stimuli (integers from 1 to 99) consisted of 30 different one- and two-digit number triplets, constituted by a middle number and two outer numbers defining a number interval for each side. The three number stimuli were spaced 25 mm apart. The numerical distance between the middle number and the outer numbers was equal (bisectable triplets: e.g., 2\_9\_16), bigger on the right side (e.g., 9\_15\_6), or on the left side (e.g., 9\_19\_11) in an equal number of trials. Triplets that are part of a multiplication table were not included. Stimuli were presented for 750 ms, with the middle number exactly in the center of the 15.4- laptop screen. The intertrial interval was 3750 ms. Subjects were seated at a distance of 45 cm from the screen and were asked to focus on a central fixation cross that disappeared as soon as the three numbers were presented.

Participants judged the magnitude of the middle number in relationship to the outer ones by pressing one of three buttons with the right index, middle, or ring finger for "left," "equal," or "right" responses. The performance of the subject on each trial was scored as follows: 0, for a correct response; 1, if subjects judged the middle number nearer to the right number of the triplet;

−1, if subjects judged the middle number nearer to the left number of the triplet.

#### **STATISTICS**

The normal distribution of the pseudoneglect indices was evaluated with Shapiro-Wilk *W*-test. In the case of a non-normal distribution, a logarithmic transformation was performed prior

Stepwise multiple linear regression analysis was used in order to investigate the predictors of LB index (Part I and II) and MNL index (Part II). We built regression models with LB (or MNL) index as the dependent variables (DVs); as possible independent predictors we entered in the regression models only the variables that were significant in the univariate analyses (*t*-test for independent sample and Pearson's product-moment correlation). As the presence of outliers can increase type I error rate in regression analysis, Mahalonobis' Distance (MD) was used to identify potential multivariate outliers. As effect size measures, the Beta standardized coefficient (β) and *R*<sup>2</sup> change were also reported to evaluate the degree of association between the significant independent predictor(s) and the DV. Statistical significance was set at *p <* 0*.*05.

# **RESULTS**

All pseudoneglect indices in the two parts of the study had a normal distribution (Experiment 1: LB index, *W* = 0*.*994, *p* = 0*.*813; Experiment 2: LB index, *W* = 0*.*989, *p* = 0*.*282; MNL index, *W* = 0*.*985, *p* = 0*.*120).

For the analysis of the Part I, we entered gender, age, EHI and "Odd beliefs or magical thinking" SPQ scale in the regression model as possible predictors of LB index (see Tables 1, 2 in Supplementary Materials). MD critical value of chi-square distribution, for degrees of freedom = 3 and *p <* 0*.*001, was 16.27. The regression model was significant [*F(*3*,*201*)* = 12*.*791, *p <* 0*.*0001] and explained 12% (adjusted *<sup>R</sup>*<sup>2</sup> <sup>=</sup> <sup>0</sup>*.*119) of LB index variance. Observed statistical power of the regression model was 0.997. No multivariate outliers were detected in the model (highest MD value: 8.739). The LB index was independently predicted by "Odd beliefs or magical thinking" SPQ scale (*<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*0001, <sup>β</sup> <sup>=</sup> <sup>0</sup>*.*273, *<sup>R</sup>*<sup>2</sup> change = 0*.*076) (**Figure 1**) and EHI (*p <* 0*.*0001, β = −0*.*231, *<sup>R</sup>*<sup>2</sup> change <sup>=</sup> <sup>0</sup>*.*052) (**Table 2A**). A decreased pseudoneglect was related to an increase of odd beliefs or magical thinking scores and a decrease of handedness score (indicating a leftward/mixed hand attitude).

For the analysis of the Part II, we build two regression models, respectively, in the first one LB index was the DV and MNL index in the second one.

We entered EHI and "Odd beliefs or magical thinking" SPQ scale in the first regression model as possible predictors of LB index (see Tables 1, 2 in Supplementary Materials). MD critical value of chi-square distribution, for degrees of freedom = 2 and *p <* 0*.*001, was 13.82. The first regression model was significant [*F(*2*,*77*)* = 11*.*801, *p <* 0*.*0001] and explained 14% (adjusted *<sup>R</sup>*<sup>2</sup> <sup>=</sup> <sup>0</sup>*.*141) of LB variance. No multivariate outliers were detected in the model (highest MD value: 9.572). The LB values was independently predicted by "Odd beliefs or magical thinking" SPQ scale (*<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*0001, <sup>β</sup> <sup>=</sup> <sup>0</sup>*.*281, *<sup>R</sup>*<sup>2</sup> change <sup>=</sup> <sup>0</sup>*.*097) and EHI (*<sup>p</sup> <sup>&</sup>lt;* <sup>0</sup>*.*0001, <sup>β</sup> = −0*.*242, *<sup>R</sup>*<sup>2</sup> change <sup>=</sup> <sup>0</sup>*.*066) (**Table 2B**).

As no association was found between MNL index and the other variables at the univariate level, no variable entered in the second regression model (MNL index as DV) (see Tables 1, 2 in Supplementary Materials).

# **DISCUSSION**

In this study, we investigated whether some tasks of spatial attention may be related to the degree of schizotypal traits in the healthy population. In particular, our purpose was to investigate the relation between the single dimensions of the SPQ and some

measures of both perceptual pseudoneglect (the LB task) and representational pseudoneglect (the MNL task).

#### **PART I**

The main finding of Part I is that a deviation from the leftward bias on the LB task correlates with schizotypy in the healthy population, particularly with the dimension of "magical thinking." To date, the current study is the largest one ever published on HS (as the observed statistical power of the regression model was 0.997).

Similarly, Liouta and colleagues found in a sample of forty right-handed HS a rightward bisection as a function of positive schizotypy (Liouta et al., 2008). These results contradict several previous findings which have shown a correlation between magical ideation and a leftward shift in spatial attention (Brugger and Graves, 1997; Kalaycioglu et al., 2000; Nalcaci et al., 2000; Taylor et al., 2002; Mohr et al., 2003; Brugger et al., 2010).

It is possible to hypothesize that different schizotypy questionnaires are potential contributors to the contradictory findings in the literature. This hypothesis is certainly undesirable and raises questions about the validity of the basic psychometric tool for assessing schizotypy: self-report questionnaires (Liouta et al., 2008). The majority of the previous studies on this topic used the Magical Ideation Scale (Brugger and Graves, 1997; Kalaycioglu et al., 2000; Taylor et al., 2002; Mohr et al., 2003; Brugger et al., 2010) while in the study by Liouta et al. (2008) schizotypy was assessed through the Oxford-Liverpool Inventory of Feelings and Experience (O-LIFE) (Mason et al., 1997; Mason and Claridge, 2006). This questionnaire produces scores for three main factors of schizotypy: positive, negative and cognitive disorganization. To assess schizotypy in our study, we used the Schizotypal Questionnaire (SPQ). Differently from the Magical Ideation and Perceptual Aberration scales, for example, which represent only single features of schizotypal personality, the SPQ contains all the subscales for all nine schizotypal traits according to DSM-III-R. Because of its high internal reliability and test-retest reliability the SPQ may be useful in screening for schizotypal personality

**Table 2 | Results of regression models in Experiment 1 (A) and Experiment 2 (B). (A) Experiment 1. Dependent variable: LB index.**


*Regression model: adjusted R<sup>2</sup>* <sup>=</sup> *0.119; F(3, 201)* <sup>=</sup> *12.791; p <sup>&</sup>lt; 0.0001.*

#### **(B) Experiment 2. Dependent variable: LB index.**


*Regression model: adjusted R<sup>2</sup>* <sup>=</sup> *0.141; F(2, 77)* <sup>=</sup> *11.801; p <sup>&</sup>lt; 0.0001.*

disorder in the general population and also in researching the correlates of individual schizotypal traits (Raine, 1991).

In the present study, only the Magical Thinking subscale is related to LB performance. Very similarly, in the previous studies only the positive psychopathological domains of schizotypy correlated with the degree of pseudoneglect (Brugger and Graves, 1997; Kalaycioglu et al., 2000; Nalcaci et al., 2000; Taylor et al., 2002; Mohr et al., 2003; Liouta et al., 2008; Brugger et al., 2010). This is partially because the majority of these studies used the magical ideation scale in isolation (Brugger and Graves, 1997; Kalaycioglu et al., 2000; Taylor et al., 2002; Mohr et al., 2003; Brugger et al., 2010) and in another study none of the Chapman scales (magical ideation, perceptual aberration, social anhedonia, physical anhedonia) were related to hemispatial attention (Gooding and Braun, 2004). Conversely, Liouta et al. (2008) found that only positive schizotypy, and not cognitive disorganization and negative schizotypy, correlated with LB performance. Positive schizotypy (Ruhrmann et al., 2010; Barrantes-Vidal et al., 2013) and mild subthreshold psychotic symptoms (Yung et al., 2004) have a strong predictive value of proneness to psychosis and this may provide some explanation for the correlation between positive schizotypy (and the magical thinking dimension) and the well-observed rightward bias at the LB performance in schizophrenia patients (Barnett, 2006; Zivotofsky et al., 2007; McCourt et al., 2008; Rao et al., 2010).

Another important aspect to be considered is that spatial pseudoneglect is not a discrete measure, as it depends on several variables, such as the segment length and spatial dislocation of the stimulus (Balconi et al., 2012), the hand (Leonards et al., 2013), and the direction of performance of the endpoint task, i.e., left-toright or right-to-left (Urbanski and Bartolomeo, 2008). Following this line of research, some studies have investigated the influence of different psychiatric disorders in LB performance. In particular, these previous studies have shown that whereas patients with affective disorders seem to have a leftward bias (He et al., 2010; Rao et al., 2010), patients with psychotic disorders show the opposite (Barnett, 2006; Zivotofsky et al., 2007; McCourt et al., 2008; Rao et al., 2010; Ozel-Kizil et al., 2012; Ribolsi et al., 2013). In this study, we suggest the hypothesis that, among the other variables, schizotypy, as a measure of proneness to psychosis, may influence the degree of pseudoneglect.

In this regard, the result of this part of the study is in line with our previous study which has shown that SCZ have an abnormal rightward bias in comparison with HS in the LB test (Ribolsi et al., 2013). Given the data from both of our studies, it is possible to hypothesize that besides a continuum of psychosis-like and schizotypal traits across both clinical and non-clinical populations, there is a corresponding continuum of deviation from the leftward bias in subjects with schizotypal traits to an abnormal clear rightward bias on the LB task in SCZ. Therefore, it is interesting to remember that schizotypy is presumed to reflect a genetically-determined disposition to schizophrenia (Meehl, 1962; Cadenhead and Braff, 2002). On the basis of a dimensional model of psychosis that assumes that pathological symptoms of schizophrenia lie on a continuum with psychosis-like schizotypal signs in non-clinical populations (Meehl, 1962; Eysenck and Barrett, 1993; Claridge et al., 1998; Van Os et al., 2000; Verdoux and Van Os, 2002), there is thought to be a link between schizotypy and schizophrenia not only on a phenomenological and genetic level but also in measures of visuo-spatial attention and lateralization.

Furthermore, in our previous study, we found that right parietal tDCS altered the performance of SCZ in the LB test with a partial correction of the rightward bias. We hypothesized that this correction could be due to an increase in the neural activity of the right PPC induced by tDCS (Ribolsi et al., 2013). In this regard, several studies reported the involvement of the right PPC in the phenomenon of visual neglect (Heilman et al., 2000; Vallar et al., 2003; Koch et al., 2008) and other authors have hypothesized that a right parietal dysfunction is responsible for the rightward bias in the LB test in schizophrenia (McCourt et al., 2008).

Similarly, abnormalities of hemispheric asymmetry assessed by right hemisphere tasks have been related to right hemisphere dysfunction in positive schizotypy (Jutai, 1989; Overby et al., 1989; Claridge and Beech, 1996; Mason et al., 1997; Nunn and Peters, 2001). On the basis of the hypothesis of the continuum between schizotypal traits and schizophrenia, we can hypothesize the involvement of the right parietal cortex dysfunction to explain the relationship between performance on the LB task and measures of schizotypy in the healthy population, but further studies are needed.

However, some recent studies have supported the hypothesis that schizotypy, and in particular magical ideation, may be related to reduced cerebral asymmetry for language (Crow, 1997; Barnett and Corballis, 2002), and that magical ideation and creativity are related to enhanced right hemisphere processing (Taylor et al., 2002; Weinstein and Graves, 2002). Recently, however, it has been shown that both magical ideation and creativity are negatively correlated with absolute hand preference but not with hand performance or with other signs of cerebral asymmetries, strongly contradicting the hypothesis of a neuropsychological explanation based on reduced single hemisphere dominance (Badzakova-Trajkov et al., 2011). Finally, other recent studies have focused their attention on brain measures other than those based on hemisphericity. In particular, subjects with high positive schizotypy show morphologic abnormalities in brain areas which have been studied also in high-risk mental state subjects and in schizophrenia, confirming that psychotic or psychotic-like experiences may have common neuroanatomical correlates across schizophrenia spectrum disorders (Modinos et al., 2010).

#### **PART II**

In this part of the study, the subjects underwent a computerized version of the LB test and the MNL test.

MNL is a common technique for assessing the so-called "representational pseudoneglect." In Western cultures, the mental representation of numbers takes the form of a number line along which magnitude is positioned in ascending order from left to right. Patients with right brain damage usually neglect smaller numbers while mentally setting the midpoint of number intervals (Vuilleumier et al., 2004; Umilta et al., 2009).

In our study, as for the paper and pencil version, the computerized LB values were also independently predicted by "odd beliefs or magical thinking," and no significant correlation was found with MNL as a DV. The MNL result contradicts previous findings by Brugger and colleagues, who found in a sample of HS that leftward bias in number space is modulated by magical ideation: higher Magical Ideation scores were associated to a stronger leftward bias. According to the authors, this correlation may be explained in terms of an overreliance on a right hemisphere semantic system, which may lead to the association between magical thinking and lateral spatial attention (Brugger et al., 2010). Conversely, in a recent study we showed that SCZ expressed the same leftward bias in the visuo-spatial representation of numbers as HS (Ribolsi et al., 2013), confirming the previous findings of Tian and colleagues of a dissociation in performance between visual line and number bisection in schizophrenia (Tian et al., 2011).

Such dissociation between the two types of visuo-spatial bisection tasks (perceptual *vs*. representative) in SCZ may imply that the neural mechanisms underlying these different forms of pseudoneglect are not identical. Indeed, neuroimaging studies revealed that the visual LB task is related to the activity of the striate and the extrastriate visual cortex and of the parietal lobe (Husain and Nachev, 2007); in contrast, the mental number bisection task is mostly related to the prefrontal cortex beside the right parietal lobe (Rusconi et al., 2011; Tian et al., 2011). Furthermore, other authors have reported that representational forms of neglect only occasionally coexist with neglect in physical space (Loetscher et al., 2010). Moreover, neuropsychological examination revealed that the apparent leftsided neglect in the bisection of number intervals has a purely non-spatial origin (Van Dijck et al., 2011). Interestingly, it has recently been shown that bias toward higher numbers in the mental bisection of number intervals in right brain-damaged patients depends on disruption of a purely abstract non-spatial representation of small numerical magnitude (Aiello et al., 2013), confirming the hypothesis that perceptual and representational pseudoneglect have different underlying neurobiological substrates.

Therefore, in our study we hypothesize that, unlike the LB test, performance in the MNL test is not influenced by schizotypy, probably because of its non-spatial origin and because it involves different neural circuits from those of perceptual pseudoneglect (Van Dijck et al., 2011). This conclusion may be in line with the suggestion that schizophrenia spectrum disorders should be seen as the consequence of basic perceptual anomalies (Silverstein et al., 2000; Herzog et al., 2004; Uhlhaas and Mishara, 2007; Silverstein and Keane, 2011) rather than as "representational disorders."

# **CONCLUSION**

The main finding of this study is that a decreased pseudoneglect as assessed by the LB task correlates with positive schizotypy in the healthy population. This result is in line with our previous study (Ribolsi et al., 2013), where we found a lack of normal leftward bias in a sample of SCZ. Taking the two studies together, we can hypothesize the existence of a correlation between the deviation from the leftward bias in the LB task and the degree of psychotic traits across the population. Deeper analysis is required, however, of a number of areas. First, LB performance should be investigated in a sample of patients with a diagnosis of schizotypal personality disorder. One possibility is that such patients may display intermediate behavior between HS with schizotypal traits and schizophrenia patients, but this hypothesis should be demonstrated in a specific study. Second, further research may be needed to relate measures of pseudoneglect with factors of schizotypy rather than the single dimensions of SPQ, as we did in this paper. Third, the neurobiological underpinning of the correlation between the degree of pseudoneglect and the severity of schizotypal traits in the healthy population should be studied. In the case of schizophrenia patients, more substantial data may suggest a pivotal role of the right parietal cortex dysfunction (McCourt et al., 2008; Ribolsi et al., 2013) to explain the lack of normal leftward bias in the LB test, but in a healthy population with schizotypal traits it is more difficult to draw any definitive conclusions.

#### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fpsyg.2013. 00846/abstract

# **REFERENCES**


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 13 April 2013; accepted: 23 October 2013; published online: 14 November 2013.*

*Citation: Ribolsi M, Lisi G, Di Lorenzo G, Rociola G, Niolu C and Siracusano A (2013) Negative correlation between leftward bias in line bisection and schizotypal features in healthy subjects. Front. Psychol. 4:846. doi: 10.3389/fpsyg.2013.00846*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Ribolsi, Lisi, Di Lorenzo, Rociola, Niolu and Siracusano. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The relation between cognitive-perceptual schizotypal traits and the Ebbinghaus size-illusion is mediated by judgment time

# *Paola Bressan\* and Peter Kramer*

*Department of General Psychology, University of Padua, Padua, Italy*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Antoine Bechara, University of Southern California, USA Xavier Noel, F.R.S.-F.N.R.S., Belgium*

#### *\*Correspondence:*

*Paola Bressan, Dipartimento di Psicologia Generale, Università di Padova, Via Venezia 8, 35131 Padova, Italy e-mail: paola.bressan@unipd.it*

In the Ebbinghaus illusion, a circle surrounded by smaller circles is perceived as larger than an identical one surrounded by larger circles. The illusion is reportedly weaker in individuals with (disorganized) schizophrenia or schizotypy than in controls, a finding that has been interpreted as evidence that both schizophrenia and schizotypy involve reduced contextual integration. In support of this view, we show that the Ebbinghaus illusion also decreases, in the general population, with cognitive-perceptual schizotypal traits (measured with both the cognitive-perceptual subscale of the *Schizotypal Personality Questionnaire-Brief* and the *Magical Ideation* scale). Our results were strong and separately replicable in different within-subjects and between-subjects conditions. However, a mediation analysis revealed that the reduction of the Ebbinghaus illusion was (statistically, hence without implying a causal relationship) entirely due to increased judgment time, i.e., the time subjects took to complete size comparisons. Judgment time increased with the strength of cognitive-perceptual schizotypal traits, but subjects with longer judgment times had smaller illusions regardless of these traits. We argue that there are at least two possible accounts of our results. Reduced contextual integration might be due to a reduced *ability* to integrate context, as previously suggested; alternatively, it could be due to a reduced *tendency* to integrate context—that is, to a detail-oriented processing style. We offer predictions for future research, testable with a deadline experiment that pits these two accounts against one another. Regardless of which account proves to be best, our results show that contextual integration decreases with cognitive-perceptual schizotypal traits, and that this relationship is mediated by judgment time. Future studies should thus consider either manipulating or measuring this time.

**Keywords: schizotypy, schizophrenia, magical ideation, Ebbinghaus illusion, contextual integration**

# **INTRODUCTION**

Schizophrenia is a heterogeneous mental disorder with various degrees of overlap with other diseases, such as bipolar disorder (Peralta and Cuesta, 2001; Dutta et al., 2007). It comprises several genetic, neurophysiological, and behavioral aspects (Lisman et al., 2008; Howes and Kapur, 2009; Javitt, 2009) whose co-occurrence, and relative dominance, vary between patients and over the course of the illness. Together, the behavioral aspects form a syndrome characterized by delusions (including paranoia), hallucinations, catatonia, psychomotor problems, and social dysfunction (Peralta and Cuesta, 2001). Schizotypy is a cognitive or biological vulnerability to schizophrenia that may or may not express itself clinically, and that affects people in the general population to various extents. To some, individuals with schizophrenia or schizotypy are categorically different from healthy ones (e.g., Meehl, 1990). To others, the three groups lie on one continuum (e.g., Claridge, 1994; McCreery and Claridge, 2002). In the current study, we will not commit to either view, but analyze quantitative differences between individuals as well as categorical ones.

There is ample evidence that schizophrenia-spectrum disorders are associated with a reduced consideration of context in both cognition and perception. Effects of reduced context processing have been found on sustained and selective attention, lexical disambiguation, latent inhibition, and size perception (for reviews, see Green et al., 2005; Silverstein and Keane, 2011), orientation perception (Yoon et al., 2009, 2010; Yang et al., 2013; but cf. Tibber et al., 2013), and early visual organization (Green et al., 2005)—more specifically on contour integration, perception of fragmented drawings, pattern recognition, grouping of dot patterns by proximity and color similarity, and coherent integration of different moving elements, such as in biological motion perception (Green et al., 2005; Kim et al., 2011; Silverstein and Keane, 2011).

With regard to contour integration, for example, some studies have investigated the detection of a low-contrast oriented grating (Gabor patch) flanked by two high-contrast ones that are either collinear or orthogonal to it (Must et al., 2004; Kéri et al., 2009, 2005; for related studies, see Keane et al., 2012; Halász et al., 2013). In non-patient controls, the collinear Gabor patches facilitate detection relative to the orthogonal ones. In patients with schizophrenia, instead, this effect was not found. The lack of facilitation in schizophrenia patients depends neither on medication, nor on insufficient attention to the task; these patients perform normally on an attentional control task with the same two flanking Gabors. In another study, a large set of small Gabor patches was presented in which some were oriented along a closed path and others were oriented randomly. Individuals with disorganized schizotypy (Uhlhaas et al., 2004) or disorganized schizophrenia (Uhlhaas et al., 2006a) detected the closed path less well than controls without schizotypy or schizophrenia. In an event-related-potentials study that used similar stimuli, both early and late processing components were implicated (Butler et al., 2013).

Related research investigated perceptual grouping by proximity and similarity. Schizophrenia patients were found to require a longer stimulus-duration than controls to perceive the grouping of little squares (Kurylo et al., 2007). A mask limited early perceptual processing, but not late decision making and response preparation. The observed effects thus appear to be bottom– up rather than top–down. In fact, at least the poorer grouping by proximity may be due to reduced sensitivity in low-spatialfrequency channels (the channels that pick up a blurred image of the stimulus) relative to high-spatial-frequency ones (O'Donnell et al., 2002).

Other studies have found superior performance among schizophrenia patients on tasks designed so that normal visual context processing would interfere with accuracy. In one such study, for example, the apparent contrast of a textured target surrounded by a high-contrast textured ring diminished, but only in patients without schizophrenia and in non-patient controls; patients with schizophrenia, instead, were not susceptible to this contrast-contrast effect (Dakin et al., 2005; see also Tibber et al., 2013; Yang et al., 2013; but cf. Barch et al., 2012). In two other studies, reports of the numerosity of a set of bars deteriorated with the orientation heterogeneity of the bars, but again only in patients without schizophrenia and in non-patient controls, and not in patients with schizophrenia (Schwartz Place and Gilmore, 1980; Wells and Leventhal, 1984). These findings of superior, rather than inferior, performance in schizophrenia patients more definitively exclude potential confounds related to motivation or a general difficulty in performing the task.

The reduced processing of context in schizophrenia has been argued to have a large impact and to lead to a cascade of other abnormalities. Indeed, impairments have been found in face perception, which arguably depends on visual organization (Green et al., 2005; Silverstein and Keane, 2011), and in theory-of-mind skills, which arguably depend both on the ability to interpret facial expressions and on memory for the particular circumstances (context) in which people operate (Green et al., 2005).

Still, there have also been failures to find reduced contextual effects in schizophrenia. For example, though some have found weaker contextual-integration effects on perceived motion (Tadin et al., 2006; Yang et al., 2013), others have found an abnormally strong one (Chen et al., 2008; see also Chen, 2011). The perception of shades of gray (lightness) is another case in point. It heavily depends on context (for reviews, see Gilchrist

**FIGURE 1 | Ebbinghaus illusion.** The central disk on the left appears smaller than the physically identical one on the right. (Similar effects are obtained with circles instead of disks.)

et al., 1999; Bressan, 2006; Kingdom, 2011) and typically a target region appears lighter if surrounded by a dim region than if surrounded by a bright one (although exceptions exist: e.g., Kramer and Bressan, 2010). This lightness-contrast effect, however, has not been found to differ between patients with schizophrenia and healthy individuals (Tibber et al., 2013; Yang et al., 2013), although—as mentioned earlier—the contrast-contrast effect does differ between the two groups.

In the Ebbinghaus illusion, a circle surrounded by smaller circles is perceived as larger than an identical one surrounded by larger circles (**Figure 1**). The illusion is reportedly weaker in individuals with disorganized schizophrenia (Uhlhaas et al., 2006a) or disorganized schizotypy (Uhlhaas et al., 2004) than in controls without schizophrenia or schizotypy, a finding that, again, has been interpreted as evidence that both schizophrenia and schizotypy involve reduced contextual integration (see also Green et al., 2005; Silverstein and Keane, 2011). The reduction in the illusion has not been found in individuals with schizotypy (Uhlhaas et al., 2004) or schizophrenia (Uhlhaas et al., 2006a; Yang et al., 2013) lacking disorganization.

Because a smaller Ebbinghaus illusion amounts to a higher accuracy in size estimation, confounds related to insufficient motivation or reduced ability to perform the task are unlikely. In principle, however, it is possible that the strength of the Ebbinghaus illusion could vary with the time subjects take to compare the central circles (judgment time). Subjects relying on a quick glance at the stimulus, for example, may perceive it differently than those who inspect it more in detail (van Zoest and Hunt, 2011). Some studies showed that size judgments of attended compared to unattended objects were more accurate (Epstein and Broota, 1986; Prinzmetal and Wilson, 1997), whereas another study found that attended lines appeared slightly longer than unattended ones (Masin, 2008). Indeed, attention has also been found to modulate the Ebbinghaus illusion itself (Shulman, 1992). Due to the potentially confounding effect of either method- or self-imposed limitations on judgment time stemming, for example, from fixed presentation times or large numbers of trials—the effects on the illusion of schizophrenic or schizotypal traits other than disorganization may have escaped detection and prove moderated or mediated by this variable.

In the current study, we investigate this possibility. More specifically, taking judgment time into account, we test whether the Ebbinghaus illusion decreases with cognitive-perceptual schizotypal traits. To accomplish this, we use a relatively large sample of individually-tested ordinary subjects (mostly psychology students), who are expected to express varying degrees of schizotypal characteristics. First, we investigate whether the Ebbinghaus illusion diminishes with increasing scores on the cognitive-perceptual subscale of Raine and Benishay's (1995) *Schizotypal Personality Questionnaire-Brief* and, as a separate control, with increasing scores on Eckblad and Chapman's (1983) *Magical Ideation* scale, that measures similar aspects of cognitive-perceptual schizotypy with different items. And second, although—as in related previous studies—we do not urge subjects to be fast, for the first time we measure how long it takes them to complete their task.

# **MATERIALS AND METHODS**

#### **PARTICIPANTS**

A total of 123 naïve (mostly psychology) students of the University of Padua volunteered to participate in the study (83 women and 40 men; median age 23, age range 20–30 years). They were requited and tested individually. The experimental procedures were approved by the Institutional Review Board at the University of Padua, and were in accordance with the Declaration of Helsinki (Sixth Revision, 2008). All participants gave their informed written consent to participate in the study.

#### **APPARATUS, STIMULI, AND PROCEDURE**

Stimuli were presented with the help of a personal computer and a custom E-Prime (Psychology Software Tools, Inc.). program on a calibrated Quato Perfect Color 22-inch CRT monitor, shielded by a custom black hood and placed in a totally dark laboratory. Response times were recorded with millisecond precision. Viewing distance was about 57 cm (measures reported here in centimeters were therefore roughly equivalent to measures in degrees of visual angle). Two white disks were presented in the centers of two abutting black backgrounds (15*.*5 × 15*.*5 cm). On half of the trials, eight small disks (1.1 cm in diameter) surrounded the left disk and six large ones (4.8 cm in diameter) the right disk (**Figure 1**). On the other half of the trials, the left and right stimuli were reversed. With a 1 × 1 cm cursor, placed within a 26*.*5 × 1 cm slide centered 3 cm below each stimulus pair, subjects modified the size of the right disk until it matched that of the left disk (2.3 cm in diameter). Subjects finalized their response with a space-bar press. The adjustable disk started very small, so that it had to be increased, on half the trials, and very large, so that it had to be decreased, on the other half. The combination of adjustable disk's context (small, large) and initial size (very small, very large) gave a total of four conditions, presented once to each subject and in the same order to all. (For a purpose related to a separate study, concurrently run with the present one, the stimuli were preceded by others in which the same method of adjustment was used, and the rest of the screen was either blue or red).

The left and right halves of the Ebbinghaus stimuli each contribute to the Ebbinghaus illusion (Franz et al., 2000). However, their relative contributions are not comparable. In our stimuli, for example, the distance between the inner edges of the contextual and target disks is smaller for the small contextual disks than for the large ones. In a comparison between the contributions of the left and right sides, this difference in edge-to-edge distance would be a confound. It is impossible to eliminate this confound without introducing another (Rose and Bressan, 2002). Thus, here we only consider the size of the entire Ebbinghaus illusion and not of its parts.

# **MATERIALS**

After the experiment, participants filled out our Italian translations of (1) Eckblad and Chapman's (1983) *Magical Ideation scale* (*MI*) and (2) the cognitive-perceptual subscale of Raine and Benishay's (1995) brief version of the *Schizotypal Personality Questionnaire* (*SPQB*) (Raine, 1991). The first author translated the scales from English into Italian, the second author—blind to the originals—translated them back into English. An independent reviewer checked the back-translations' accuracy; no noteworthy differences were found. The authors and the reviewer are all fluent in both English and Italian and have extensive translation experience.

The MI scale measures cognitive-perceptual aspects of schizotypy and consists of 30 items that require a true or false response. Example items are: "I have occasionally had the silly feeling that a TV or radio broadcaster knew I was listening to him"; "Numbers like 13 and 7 have no special powers" (scored negatively); "I have felt that I might cause something to happen just by thinking too much about it." The SPQB's cognitive-perceptual (SPQB-CP) subscale measures the same aspects and consists of 8 items that require a yes or no response. Example items are: "Do you often pick up hidden threats or put-downs from what people say or do?"; "Have you had experiences with astrology, seeing the future, UFOs, ESP, or a sixth sense?"; "Have you ever had the sense that some person or force is around you, even though you cannot see anyone?". The SPQB-CP subscale and the MI scale measure similar aspects of schizotypy, allowing us to cross-validate them within our sample.

# **STATISTICS**

One of the analyses we performed was a mediation analysis (Baron and Kenny, 1986), an analysis that establishes whether or not the correlation between one variable and another is statistically due to their correlation with a third variable. Note that the third variable need not necessarily be the cause of the correlation between the first two variables, because one or more variables that are not considered could play that role instead.

In order to confirm that a third variable (here judgment time) mediates the relationship between an independent variable (here either SPQB-CP or MI scores) and a dependent one (here Ebbinghaus-illusion magnitude), it has to be established that (1) the independent variable correlates significantly with the dependent variable, (2) the independent variable correlates significantly with the third variable, and (3) the third variable correlates significantly with the dependent variable, even after controlling for the independent variable. That is, in a multiple regression, the contribution of the independent variable should Bressan and Kramer Schizotypy and the Ebbinghaus illusion

be greatly reduced in the presence of the third variable and ideally become non-significant. (If the contribution of the independent variable remains significant and considerable, but does depend on the third variable, then the latter is a moderator rather than a mediator.)

## **RESULTS**

One data point corresponded to an Ebbinghaus illusion that was more than three standard deviations below the mean, and one data point to a response time that was more than three standard deviations above it. Although their inclusion led to virtually identical results, these two data points were excluded from analysis.

In the Ebbinghaus illusion, the circle surrounded by smaller circles is perceived as larger than the identical one surrounded by larger circles (**Figure 1**). Thus, if subjects are to adjust the former's size so that it matches the latter's, they should have a tendency to adjust downward. In the converse case, they should have a tendency to adjust upward. Indeed, when the fixed disk in our experiment was surrounded by larger ones and the adjustable disk by smaller ones, subjects reduced the adjustable disk's size by 16% (SD 8%). Conversely, when the fixed disk was surrounded by smaller ones and the adjustable disk by larger ones, subjects enlarged the adjustable disk's size by 24% (SD 11%). The mean Ebbinghaus-illusion magnitude was thus 20% (SD 8.4%). The average time taken to complete the adjustment (judgment time) was 7.7 s (SD 6.1 s). To eliminate skew in judgment times, we log-transformed them, but the transformed and untransformed data produced almost identical results.

The SPQB-CP scores ranged from 0 to 8 and had a mean of 1.6 (SD 2.0); nearly half the subjects (47.2%) scored 0. Because the distribution was so heavily skewed, we performed all analyses both on the raw SPQB-CP scores and on a dichotomized variable on which scores were categorized as either "zero SPQB-CP" or "positive SPQB-CP" (above 0). Still, results in the two cases were quite similar, and hence, we only report the former. For better visualization, in **Figure 2** we show the dichotomized data.

We found that the Ebbinghaus-illusion magnitude decreased with SPQB-CP scores, but the effect was entirely mediated (Baron and Kenny, 1986) by judgment time (**Figure 2**): (1) Ebbinghaus-illusion magnitude decreased with SPQB-CP scores: *r* = −0*.*41, *p <* 0*.*0001, *N* = 122; (2) SPQB-CP scores increased with judgment time: *r* = 0*.*64, *p <* 0*.*0001, *N* = 122; (3) in a multiple regression that explained 38% of the Ebbinghausillusion variance: *R* = 0*.*62, *F(*2*,* <sup>118</sup>*)* = 36*.*5, *p <* 0*.*0001, only the judgment-time coefficient was significant: β = −0*.*60, *t* = −6*.*28, *p <* 0*.*0001, whereas the SPQB-CP coefficient did not even reach marginal significance: β = −0*.*03, *t* = −0*.*36, *p* = 0*.*716. Age correlated with SPQB-CP and judgment time, but not with Ebbinghaus-illusion magnitude: respectively, *r* = −0*.*20, *p* = 0*.*030, *N* = 123, *r* = −0*.*20, *p* = 0*.*027, *N* = 122, *r* = 0*.*02, *N* = 122; adding age to the multiple regression model left the results virtually unchanged.

Taking SPQB-CP out of the multiple regression model, the Ebbinghaus illusion still decreased with judgment time: *r* = −0*.*62, *p <* 0*.*0001, *N* = 121. Moreover, it did so not only in

represented in black. The regression line is a fit to all data points regardless of SPQB-CP score (i.e., regardless of symbol color). Note that individuals with a positive SPQB-CP score tend to be less susceptible to the Ebbinghaus illusion, but to have a longer judgment time, and that regardless of SPQB-CP score, individuals with a longer judgment time tend to be less susceptible to the Ebbinghaus illusion.

subjects whose SPQB-CP score was positive: *r* = −0*.*50, *p <* 0*.*0001, *N* = 64, but also in those whose SPQB-CP score was 0: *r* = −0*.*52, *p <* 0*.*0001, *N* = 57.

The MI scores ranged from 0 to 23 and had a mean of 6.0 (SD 4.0). MI and SPQB-CP scores were highly correlated, *r* = 0*.*74, *N* = 123, *p <* 0*.*0001, and the MI results followed the same pattern as the SPQB-CP results. We found that (1) Ebbinghausillusion magnitude decreased with MI: *r* = −0*.*19, *p* = 0*.*037, *N* = 122, (2) MI increased with judgment time: *r* = 0*.*38, *p <* 0*.*0001, *N* = 122, and (3) in a multiple regression that explained 38% of the Ebbinghaus-illusion variance: *R* = 0*.*62, *F(*2*,* <sup>118</sup>*)* = 36*.*9, *p <* 0*.*0001, only the judgment-time coefficient was significant: β = −0*.*64, *t* = −7*.*78, *p <* 0*.*0001, whereas the MI coefficient did not even reach marginal significance: β = 0*.*07, |*t*| *<* 1. Age did not correlate with MI.

Demonstrating within-subjects replicability, the Ebbinghaus illusion decreased with judgment time, and separately also with SPQB-CP scores, both when the adjustable disk was surrounded by smaller disks and when it was surrounded by larger ones (all *p*s *<* 0.0001). Demonstrating between-subjects replicability, the Ebbinghaus illusion decreased with judgment time in both men and women (both *p*s *<* 0.0001). The illusion also separately decreased with SPQB-CP scores, significantly in the relatively large sample of women (*p <* 0*.*0001; *N* = 82) and marginally so in the smaller sample of men (*p* = 0*.*09; *N* = 40).

# **DISCUSSION**

In a relatively large sample taken from the general population, we found that the Ebbinghaus illusion decreases with cognitiveperceptual schizotypal traits. Unlike in earlier related studies, however, we measured the time subjects took for their size judgment and found that the reduction in the illusion was (statistically, hence without implying a causal relationship) entirely mediated by this judgment time. Before examining possible explanations of our results, let us first discuss how our study compares to previous ones.

#### **COMPARISON WITH PREVIOUS STUDIES**

First, the Ebbinghaus illusion has previously (Uhlhaas et al., 2004) been reported to decrease with disorganization traits in schizotypy, but not with other schizotypal traits. In our study, we presented a fairly strong version of the Ebbinghaus illusion, similar to the classic one—with small inducing disks around one target and large inducing disks around the other (**Figure 1**). Instead, Uhlhaas et al. (2004)—and also Yang et al. (2013), who failed to find an effect of schizophrenia on the illusion—used a version of the Ebbinghaus illusion in which only one half of the stimulus was presented along with a neutral condition, a version known to produce an illusion less than half as large as the classic one (Franz et al., 2000). This choice of stimuli may thus have reduced power. Indeed, in Uhlhaas et al. (2004) the results of individuals with disorganized schizotypy were significantly different from those of individuals with non-disorganized schizotypy, and from controls without schizotypy, only when the illusion-inducing context consisted of large disks and not when it consisted of small ones.

Second, Uhlhaas et al. (2006a) reported that the Ebbinghaus illusion decreased in patients with disorganized schizophrenic traits but not in patients with other schizophrenic traits; however, no significant difference was found *between* these two conditions. Hence, the results do not allow the conclusion that the Ebbinghaus illusion is affected differently by disorganized schizophrenic traits and by cognitive-perceptual ones. Horton and Silverstein (2011) did find a significant difference between these conditions, but only for deaf patients, and they were not compared to individuals without schizophrenia. For normally hearing patients—who were not compared to individuals without schizophrenia either—the difference was significant only when the illusion-inducing context consisted of small disks and not when it consisted of large ones; a result opposite to that found for schizotypy by Uhlhaas et al. (2004).

Third, in our main analyses, unlike in the earlier study on schizotypy and the Ebbinghaus illusion by Uhlhaas et al. (2004), we did not cluster subjects into an experimental and a control group, but maintained the raw questionnaire scores. Thus, we preserved information about, for example, differences between subjects scoring low and subjects scoring 0 on the SPQB-CP. With 47.2% of our sample scoring indeed 0, this preservation of information may have been important.

Fourth, we used a different method from previous related studies: the method of adjustment rather than a forced choice. One possible consequence of our method is that our subjects might have taken longer to perform their task than the subjects of earlier studies. Because previous studies of the Ebbinghaus illusion did not measure response time, it is difficult to verify this. Still, Uhlhaas et al. (2006a)—who did not investigate response time either—reported that their Ebbinghaus stimuli were presented for 4 s. Assuming that subjects responded before or immediately after stimulus offset, they were likely to have responded faster than our subjects (whose average response time was 7.7 s). It therefore remains unclear whether effects of cognitive-perceptual schizotypal traits could have emerged if presentation, and response, times had been unlimited. Notably, Uhlhaas et al. (2006b), who did investigate response time and also perceptual organization, but not the Ebbinghaus illusion, found that patients with schizophrenia had longer response times than healthy controls.

# **POSSIBLE EXPLANATIONS OF OUR RESULTS**

The differences between the current and previous studies were thus large enough to explain how we could have obtained different results. The question now is their explanation. Various studies use tasks that require contextual integration. A reduced ability, or reduced tendency, to perform this integration is then likely to increase response time, decrease accuracy, or both (e.g., Uhlhaas et al., 2006b). In studies like ours that use a size-judgment task, contextual integration is not required, but our data suggest that it is nevertheless, to various extents, performed by almost all subjects (**Figure 2**). It is thus possible that subjects with a reduced ability to integrate context lose time doing it. This conjecture is consistent with our finding that the Ebbinghaus illusion, which depends on contextual integration, decreases with judgment time. We also found that the Ebbinghaus illusion decreases with cognitive-perceptual schizotypal traits and this effect could be due to a negative relationship between these traits and the ability to integrate context. However, because we found that the Ebbinghaus illusion decreases with judgment time *regardless* of these traits, our results suggest that the link between reduced contextual integration and these traits cannot be an exclusive one.

A reduced Ebbinghaus illusion amounts to increased sizejudgment accuracy. As discussed in our introduction, this fact has been used to argue that a reduced Ebbinghaus illusion is unlikely to be due to inattention or weaker motivation. Yet, it is still possible that some subjects base their judgment on a quick global impression of the stimulus, whereas others go through the trouble of inspecting it in detail. The processing of a stimulus's details generally takes longer than the processing of its global properties (for a review, see Kimchi, 1992). In addition, subjects performing their size judgments for a longer time focus, necessarily, more on the task-relevant targets and less on the task-irrelevant context. As unattended context affects the Ebbinghaus illusion less than attended context (Shulman, 1992), one would thus expect the illusion to decrease with judgment time. Indeed, this is what we found. The remaining question is whether the tendency to inspect the stimulus in detail, or to inspect *any* stimulus in detail (i.e., the tendency to rely on a detail-oriented processing style), may somehow be related to schizotypal traits (see also Phillips et al., 2004; Doherty et al., 2010).

One possible relation comes to mind when one considers that schizotypy and schizophrenia—and especially their positive symptoms that we have measured here—are often comorbid with anxiety and negative mood (e.g., Lewandowski et al., 2006). It has been argued that these moods induce a detail-oriented processing style, possibly via a reduction in the scope of attention (the socalled Easterbrook hypothesis: Easterbrook, 1959; for a review, see Friedman and Förster, 2010). Although a few studies have failed to find evidence for effects of mood on perceptual organization, many have corroborated it.

Among the former are some studies by Silverstein and colleagues. Silverstein et al. (1992), for example, found no effect on perceptual organization of either depression or anhedonia. Likewise, Silverstein et al. (1996) found no effect on perceptual organization of belonging to a control group of individuals with either depression accompanied by psychotic features, bipolar disorder accompanied by psychotic features, schizoaffective disorder, or delusional disorder. Uhlhaas et al. (2006a), finally, found no effect on the Ebbinghaus illusion of belonging to a control group that included depressed individuals; in this group, however, the depressed individuals were a minority.

A large number of studies by different research groups have found evidence that both anxiety and negative mood do promote detail-oriented processing. The tendency, for example, to classify large shapes consisting of small shapes (Navon, 1977; Kimchi, 1992) on the basis of the small shapes, rather than the large ones, was found to increase with both trait anxiety (Tyler and Tucker, 1982) and negative mood (Basso et al., 1996; Gasper and Clore, 2002; Gasper, 2004; Fredrickson and Branigan, 2005). Likewise, responses to central targets were hampered less by flanking irrelevant context (Eriksen and Eriksen, 1974) when mood was negative than when it was neutral or positive (Fenske and Eastwood, 2003; Rowe et al., 2007; Schmitz et al., 2009; Moriya and Nittono, 2011). Consistent evidence has also been found with various other techniques, ranging from the detection of peripheral targets (Weltman et al., 1971) to the holistic recognition of faces (Curby et al., 2012). [For further evidence, discussion, and the latest theoretical refinements, see Friedman and Förster (2010), and Huntsinger (2012).]

#### **PREDICTIONS FOR FUTURE RESEARCH**

In our view, there are thus at least two possible accounts of our results: one in which reduced contextual integration is due to a reduced *ability* to integrate context and one in which it is due to a reduced *tendency* to integrate context—that is, to a detail-oriented processing style. Given that a processing tendency should be more flexible than a processing disability, with a deadline experiment (an experiment in which subjects are forced to respond before a variable deadline; e.g., Kramer et al., 2013) these two accounts can be pitted against each other. At a long deadline, some may have a large Ebbinghaus illusion and some may have a small one. As the deadline decreases, however, the reduced-tendency account predicts that the two groups should become more similar, whereas the reduced-ability account predicts they should not. That is, according to the reduced-tendency account, time pressure should redirect subjects away from a timeconsuming detailed analysis of the stimulus and toward a quick judgment based on a short glance at it. Stated differently, time pressure should redirect subjects' orientation from the trees (i.e., details) to the forest (i.e., the big picture). (For similar ideas see, e.g., Tyler and Tucker, 1982.) According to the reduced-ability account, this redirection should not occur, because those who integrate little of the context at long deadlines should not be able to integrate more of it at short ones.

# **CONCLUSION**

Regardless of which account turns out to be best, our present results challenge previous claims that contextual integration does not decrease with cognitive-perceptual schizotypal traits, but also show that this relationship (again statistically, and hence without implying a causal relationship) is entirely mediated by judgment time. Future studies should thus consider either manipulating or measuring this time.

# **ACKNOWLEDGMENTS**

Supported by a grant from the University of Padova (Progetto di Ricerca di Ateneo CPDA084849) to Paola Bressan. We thank Francesca Loria and Lucia Lazzarini for data collection, Luca Semenzato for writing the E-Prime program, and Nino Trainito for checking the translations of the questionnaires.

# **REFERENCES**


*J. Int. Neuropsych. Soc.* 2, 249–255. doi: 10.1017/S1355617700001193


*Schizophr. Bull.* 37, 709–715. doi: 10.1093/schbul/sbr020


Bressan and Kramer Schizotypy and the Ebbinghaus illusion

the classification of psychosis. *Schizophr. Bull.* 33, 868–876. doi: 10.1093/schbul/sbm059


*Psychiatry Rev.* 1, 11–22. doi: 10.2174/1573400052953529


*Vision Res.* 51, 652–673. doi: 10.1016/j.visres.2010.09.012


(1996). Stimulus configuration and context effects in perceptual organization in schizophrenia. *J. Abnorm. Psychol.* 105, 410–420. doi: 10.1037/0021-843X. 105.3.410


Place and Gilmore. *J. Abnorm. Psychol.* 93, 231–234. doi: 10.1037//0021-843X.93.2.231


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

#### *Received: 01 February 2013; accepted: 27 May 2013; published online: 12 June 2013.*

*Citation: Bressan P and Kramer P (2013) The relation between cognitive-perceptual schizotypal traits and the Ebbinghaus size-illusion is mediated by judgment time. Front. Psychol. 4:343. doi: 10.3389/fpsyg. 2013.00343*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Bressan and Kramer. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Cognitive and neuroplasticity mechanisms by which congenital or early blindness may confer a protective effect against schizophrenia

# **Steven M. Silverstein1,2\*,YushiWang<sup>1</sup> and Brian P. Keane1,2,3**

<sup>1</sup> University Behavioral HealthCare, University of Medicine and Dentistry of New Jersey, Piscataway, NJ, USA

<sup>2</sup> Department of Psychiatry, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Piscataway, NJ, USA

<sup>3</sup> Rutgers University Center for Cognitive Science, Piscataway, NJ, USA

#### **Edited by:**

Michael Green, University of California Los Angeles, USA

#### **Reviewed by:**

Yue Chen, McLean Hospital, USA Jonathan K. Wynn, University of California Los Angeles, USA

#### **\*Correspondence:**

Steven M. Silverstein, University Behavioral HealthCare, University of Medicine and Dentistry of New Jersey, 151 Centennial Avenue, Piscataway, NJ 08854, USA. e-mail: silvers1@umdnj.edu

Several authors have noted that there are no reported cases of people with schizophrenia who were born blind or who developed blindness shortly after birth, suggesting that congenital or early (C/E) blindness may serve as a protective factor against schizophrenia. By what mechanisms might this effect operate? Here, we hypothesize that C/E blindness offers protection by strengthening cognitive functions whose impairment characterizes schizophrenia, and by constraining cognitive processes that exhibit excessive flexibility in schizophrenia. After briefly summarizing evidence that schizophrenia is fundamentally a cognitive disorder, we review areas of perceptual and cognitive function that are both impaired in the illness and augmented in C/E blindness, as compared to healthy sighted individuals. We next discuss: (1) the role of neuroplasticity in driving these cognitive changes in C/E blindness; (2) evidence that C/E blindness does not confer protective effects against other mental disorders; and (3) evidence that other forms of C/E sensory loss (e.g., deafness) do not reduce the risk of schizophrenia. We conclude by discussing implications of these data for designing cognitive training interventions to reduce schizophrenia-related cognitive impairment, and perhaps to reduce the likelihood of the development of the disorder itself.

**Keywords: schizophrenia, blindness, perception, cognition, vision, vision disorders, plasticity**

# **INTRODUCTION**

Over the past 60 years, several authors (Chevigny and Braverman, 1950; Abely and Carton, 1967; Horrobin, 1979; Riscalla, 1980; Feierman, 1982; Sanders et al., 2003) have postulated that congenital or early (C/E) blindness may serve as a protective factor against the development of schizophrenia. Supporting data derive from several sources, including literature reviews, examination of cohorts of blind patients in psychiatric hospitals, and surveys of agencies that treat large numbers of blind people. The most recent of these Sanders et al. (2003) noted that across all past papers, there has not been even one reported case of a congenitally blind person who developed schizophrenia. While we identified one reported case of a blind "schizophrenic" child (Stewart and Sardo, 1965), the 6-year-old described in that paper had many symptoms of autism and no symptoms of psychosis, and would almost certainly be diagnosed with an autism-spectrum disorder using DSM-III (American Psychiatric Association, 1980) or later criteria. In contrast to the lack of cases of people with C/E blindness who meet modern criteria for schizophrenia, reports do exist of people who develop both blindness and schizophrenia later in life (Checkley and Slade, 1979). We therefore hypothesize that it is the brain changes that occur secondary to C/E blindness – rather than blindness *per se* – that protect against schizophrenia. However, the mechanisms that confer this apparent protection have not yet been identified. In this paper, we advance the hypothesis

that protection occurs by (1) strengthening the perceptual and cognitive functions whose impairment forms the essential nature of schizophrenia; and (2) constraining cognitive processes that are excessively neuroplastic in the illness.

In the discussion below, after briefly summarizing the position that schizophrenia is fundamentally a cognitive (i.e., informationprocessing) disorder, we review evidence that a cluster of perceptual and cognitive functions is at once markedly impaired in schizophrenia and significantly augmented in C/E blindness, compared to healthy sighted individuals (see **Table 1**). In subsequent sections, we explicate the role of neuroplasticity in driving these perceptual and cognitive differences, we provide evidence that C/E blindness does not confer protection against other mental disorders (e.g., depression, anorexia nervosa) and finally we argue that other forms of C/E sensory loss (e.g., deafness) do not reduce the risk of schizophrenia. We conclude by noting the implications of these data for designing cognitive training interventions for patients and people at high risk for the disorder, and for developing interventions to prevent schizophrenia.

# **SCHIZOPHRENIA AS A COGNITIVE DISORDER: KRAEPELIN AND DEMENTIA PRAECOX REVISITED**

Although the most obvious clinical features of schizophrenia include psychotic symptoms such as hallucinations, delusions, and bizarre behavior, it has long been thought that such symptoms **Table 1 | Summary of cognitive and brain function enhancements (**+**) and impairments (**−**), compared to healthy sighted individuals, in C/E blindness and schizophrenia.**


\*Faster than normal.

\*\*Slower than normal.

are secondary, compensatory, features of the disorder. In contrast, much evidence suggests that the core features of schizophrenia are cognitive in nature, involving disturbances in perception, attention, memory, learning, language, and context-based modulation of these processes as a function of expectations and action plans (Kraepelin, 1903; Bleuler, 1950; Weiss, 1989; Phillips and Silverstein, 2003; Nuechterlein et al., 2012; Tamminga, 2013; Khan, in press). In this way, schizophrenia can be viewed as similar to brain disorders such as Parkinson's disease dementia and dementia with Lewy bodies,in which psychotic symptoms can occur, but are often late-developing phenomena, and are not considered the defining features of the disorder (Ballard et al., 2001; Williams-Gray et al., 2006; Ffytche, 2007). A large body of research evidence now indicates that perceptual and cognitive impairments are found in nearly all people with schizophrenia, and occur with greater frequency than psychotic symptoms (Palmer et al., 2009). Significant cognitive impairment is also easily detectable in most schizophrenia patients even when psychotic symptoms are in remission (Altshuler et al., 2004). In addition: (1) low intelligence is a risk factor for schizophrenia; (2) cognitive decline and intellectual

underperformance precede the onset of psychotic symptoms (and therefore the diagnosis) by many years; (3) decline in cognitive functioning continues after psychosis onset; and (4) this pattern of changes is not found in other major mental illnesses, even those that can include psychotic symptoms, such as bipolar disorder (Neumann et al., 1995; Schenkel and Silverstein, 2004; Khan, in press). Cognitive impairments in schizophrenia are also significantly correlated with level of functional disability (Green, 1996; Silverstein et al., 1998), and are therefore important treatment targets.

# **DOMAINS OF PERCEPTUAL AND COGNITIVE FUNCTIONING THAT ARE STRENGTHENED BY C/E BLINDNESS AND IMPAIRED IN SCHIZOPHRENIA**

# **AUDITORY PERCEPTION**

Compared to sighted people (and in some cases to late-blind people), C/E blind individuals are characterized by enhanced auditory acuity for pitch, better pitch discrimination, better pitch-timbre categorization (Wan et al., 2010), better speech discrimination (Hertrich et al., 2009), better sound localization, better temporal auditory resolution, better noise-embedded speech discrimination (Niemeyer and Starlinger, 1981; Muchnik et al., 1991), lower sound threshold for identification of pseudo-words (Rokem and Ahissar, 2009), shorter latencies for auditory event-related potentials (ERPs;Roder et al.,1996,1999), and greater tolerancefor short stimulus-onset asynchronies in an auditory backward masking task (Stevens and Weaver, 2005). C/E blindness is also associated with comprehension of ultra-fast synthetic speech at a rate of ∼25 syllables per second, which is close to three times the rate at which non-blind individuals can comprehend such speech (Hertrich et al., 2009). In addition, C/E blind subjects have shown larger mismatch negativity (MMN) components of the ERP, suggesting a compensatory improvement in pre-attentional processes (Kujala et al., 1995). It has been demonstrated that these sensory and perceptual processing differences are due to enhanced basic perceptual skills, not to "higher order"functions such as attention, memory, language, or executive functions (reviewed in Cattaneo and Vecchi, 2011).

The above findings are in contrast to what has been observed in schizophrenia, in which, for example, patients are typically characterized by poor auditory acuity, precision, and discrimination (including in tone matching) (Javitt et al., 1997; Rabinowicz et al., 2000; Li et al., 2002; Rojas et al., 2007; Turetsky et al., 2009; Perrin et al., 2010; Kantrowitz et al., 2013), impairments in sound localization (Perrin et al., 2010), temporal resolution of sound (Foucher et al., 2007), auditory backward masking (Kallstrand et al., 2002), pre-attentional processing as indicated by the (smaller) MMN ERP component (Umbricht et al., 2003), and longer latency of post-attentional auditory ERPs (Ward et al., 1991; Sevik et al., 2011; Iyer et al., 2012). Individuals with schizophrenia also have abnormal processing of speech sounds (Ngan et al., 2003; Hirano et al., 2008), are worse than other groups at detecting speech embedded in noise, and are more likely than controls to identify non-speech sounds as speech (Vercammen et al., 2008). Importantly, it has been shown that these impairments in pitch processing in schizophrenia are not secondary to symptoms, medication, or poor attentional functioning (Rabinowicz et al., 2000).

The impaired auditory functions noted above are not mere laboratory curiosities; they have important clinical and functional implications. For example, poor sound localization is thought to be associated with an externalization bias in which there is a reduced ability to recognize that the sound of one's own voice is coming from the self, as opposed to an external cause, which may contribute to delusional thinking (Johns et al., 2001; Ilankovic et al., 2011). At a more general level, perception has been conceptualized as an active, hypothesis-generating function that involves generating a model of the world, and then updating this model via coding of prediction error (i.e., the degree to which continued incoming sensory input does not conform to the current perceptual model of the world; Corlett et al., 2009; Clark, in press). It has been postulated that symptoms of schizophrenia, such as hallucinations and delusions may result from an impaired ability to detect inconsistencies between what would normally be expected (based on past experience) and incoming sensory information (Corlett et al., 2007, 2009; Clark, in press). Both faulty bottom-up processing and poor quality top-down feedback, as have been demonstrated in schizophrenia (Silverstein and Keane, 2009, 2011a,b), would degrade the quality of perceptual representations and the extent to which they are modified by context to most accurately reflect the nature of reality (Corlett et al., 2007, 2009;Clark, in press; Silverstein, in press). The result can be abnormal perceptions, hypersalience of normally ignored stimuli, and new and unusual beliefs to account for these changes in subjective experience – in short, an internal representation of the world that can deviate markedly from reality, and increasingly so over time (Clark, in press). In contrast, as C/E blindness is associated with enhanced sensory and perceptual processing, it can be seen how this would reduce the likelihood of the sort of prediction coding failures, and their sequelae, associated with schizophrenia. The auditory impairments in schizophrenia described above have also been shown to have other important consequences, which would not occur in people with C/E blindness, such as problems decoding cues to emotion in speech (Leitman et al., 2011; Gold et al., 2012) and impaired semantic analysis (Leitman et al., 2007).

#### **AUDITORY ATTENTION**

Blindness has been associated with enhancements in several components of attention that are impaired in schizophrenia. For example, on reaction time tasks, blind subjects are superior to sighted individuals in responding to auditory and tactile cues (Collignon and De Volder, 2009). On tactile and auditory *selective* attention tasks involving spatial discrimination, blind people are more efficient than sighted subjects (Roder et al., 1996, 1999; Hotting and Roder, 2004; Collignon et al., 2006). Since these studies demonstrated that blind and sighted participants did not differ in sensory sensitivity on these tasks, these findings appear to reflect superior attentional capabilities (Cattaneo and Vecchi, 2011), although teasing apart attentional and perceptual deficits is often difficult. Blindness is also associated with superior performance on tasks of *divided* attention, such as dichotic listening (Hugdahl et al., 2004), and tasks requiring dividing attention across more than one sensory modality (Kujala et al., 1995). Again, such a pattern of performance contrasts sharply with what has been long been observed in schizophrenia, namely, severely impaired selective and divided

#### **WORKING AND LONG-TERM MEMORY**

Compared to healthy, sighted individuals, C/E blind individuals have superior working memory capacity, in terms of both the amount that can be processed, and the ability to recall the sequence in which information was presented (Hull and Mason, 1995;Amedi et al., 2003; Roder and Neville, 2003; Raz et al., 2007). The authors of a recent study that matched perceptual threshold of blind and sighted subjects concluded that superior short-term memory in C/E blind people was due to enhancement of stimulus encoding (related to the processes reviewed above), rather than to differences in storage or recall processes (Rokem and Ahissar, 2009). In addition, it is thought that, in C/E blindness, primary reliance on controlled, sequential processes (i.e., via haptic and auditory perception) during navigation – as opposed to the more automatic and parallel processing normally afforded by vision – leads to superior working memory abilities (Raz et al.,2007; Salillas et al., 2009;Cattaneo and Vecchi, 2011). C/E blind people have also demonstrated superior long-term memory compared to sighted controls (Bull et al., 1983; Roder et al., 2001). All of this is in stark contrast to schizophrenia, in which working memory (Silver et al., 2003; Conklin et al., 2005; Brahmbhatt et al., 2006; Horan et al., 2008; Silverstein et al., 2010) and long-term memory (Van Snellenberg, 2009) impairments are typically observed, along with problems in stimulus organization during encoding (Harvey et al., 1986; Brebion et al., 1997, 2004; Landgraf et al., 2011).

These data have important clinical implications. A critical function of WM is to integrate perceptual information with information stored in long-term memory, to allow for efficient learning, memory, and reasoning (Cattaneo and Vecchi, 2011). Enhanced WM abilities would thus facilitate these other processes, as is the case in C/E blindness (Cattaneo andVecchi,2011),whereas impairment would have the opposite effect, as is the case in schizophrenia (Tek et al., 2002). Thus, it can again be seen how C/E blindness could protect against development of the cognitive impairments that comprise the syndrome of schizophrenia.

#### **LANGUAGE**

Congenital or early blind children often experience language delays and abnormalities related to semantic, syntactic, and phonological processing (Perez-Pereira and Conti-Ramsden, 1999). One possible consequence of these is a reduction in the risk of developing disordered thought,which is commonly seen in schizophrenia. For example, many thought-disordered patients with schizophrenia have a tendency toward overabstraction and overinclusion in their thinking, as well as overly elaborated (and more easily primed) semantic networks (Siekmeier and Hoffman, 2002; Lerner et al., 2012). In addition, compared to healthy controls, individuals with schizophrenia more often use language in odd ways, including

generating neologisms (i.e., novel words; Solovay et al., 1987; Kreher et al., 2008). In contrast, it has been shown that compared to sighted children, blind children are characterized by a lack of overgeneralization of concepts and categories, and a reduced number of word inventions (Andersen et al., 1984, 1993). It has also been shown that semantic networks in blind children are more dependent on language, and less dependent on sensory experience, than sighted children (Pring, 1988). This could be a protective factor against delusion formation secondary to altered sensory experience, as has been hypothesized to exist in schizophrenia (Maher, 1974; Uhlhaas and Mishara, 2007).

#### **OLFACTION**

Cuevas et al. (2009) demonstrated that congenitally blind males were better at discriminating odors, and at naming familiar odors, compared to sighted subjects. Other studies have reported mixed results, for example, demonstrating superior odor identification but not odor sensitivity (Wakefield et al., 2004). However, studies with mixed findings have grouped early blind and late-blind subjects together, and did not match groups on gender (women are typically better than men at olfaction tasks; Cattaneo and Vecchi, 2011). While more research is needed on this issue, evidence does suggest that people born blind may develop increased olfactory abilities (Kupers et al., 2011). This is in contrast to people with schizophrenia, in which olfactory impairments have been reported (Nguyen et al., 2010; Kamath et al., 2011). Although the functional consequences of olfactory impairment in schizophrenia are not well understood, such deficits could impair self-regulation, learning, attachment, and interpersonal functioning (see Sullivan, 2003). For example, the detection of maternal odor can promote attachment to the mother, reduce stress, and help normalize the sleep-wake cycle in infants (Goodin-Jones et al., 1997; Sullivan and Toubas, 1998). In addition, odor perception is important in forming bonds with peers in childhood (Sullivan, 2000). It is therefore interesting that risk for schizophrenia is associated with problems in maternal attachment (demonstrated at 3 days, 1, and 6 years of age; Naslund et al., 1984; Persson-Blennow et al., 1984; McNeil and Kaij, 1987),increased stress reactivity (Myin-Germeys and van Os, 2007), and fewer peer relationships (Schenkel and Silverstein, 2004), as well as olfactory processing deficits (Brewer et al., 2003).

#### **CONSTRUCTION OF SUBJECTIVE EXPERIENCE**

In this section, we discuss: (1) differences between people who are C/E blind and people with schizophrenia with regard to the relative importance of parallel versus serial data acquisition in the generation and quality of subjective experience; and (2) the role that preservation of a form of visual imagery in C/E blindness may play in the integration of experience. As noted above, C/E blind people rely to a much greater extent than sighted people on serial processing due to primary reliance on haptic and auditory processing, and lack of visual input. That is, due to the temporalsequential nature of data acquisition about objects in auditory and haptic perception, the relatively smaller amounts of information available at any one time in haptic perception compared to vision (Barber and Lederman, 1988; Amadeo and Speicher, 1995; Herssens, 2010), and the more rapidly changing nature of information flow in audition relative to vision (Shamma et al., 2011), formation of coherent mental representations in haptics and audition requires that information must be "built up" over time to a greater extent than in vision (Geenens, 1999). One result of this imposed learning style is that C/E blind individuals develop abilities in sequential and controlled processing that may be superior to those of sighted people (Vecchi et al., 2004; Raz et al., 2007). In addition, as we argue below, compensatory processes involving working memory capacity, perceptual organization via controlled processing, multisensory integration, learning, and a preserved form of visual imagery lead to a rich, seemingly automatic, and non-fragmented flow of experience (seeCattaneo et al., 2008; Salillas et al., 2009 and above; Cattaneo and Vecchi, 2011). In contrast, in schizophrenia, impairments in working memory capacity, perceptual organization, and multisensory integration,which develop long after birth and for which there appear to be no compensation, contribute to an over-reliance on serial processing that is often inefficient, and that can lead to fragmented subjective experience (Chapman, 1966; Knight, 1984; Carr and Wale, 1986; Knight and Silverstein, 1998; Silverstein and Keane, 2011a). This is illustrated in the following quote from a person with schizophrenia: "Everything I see is split up. It's like a photograph that's torn in bits and put together again. If somebody moves or speaks, everything I see disappears quickly and I have to put it together again" (Chapman, 1966, p. 229).

Interesting examples of how blind individuals construct mental representations incrementally via haptic perception come from Helen Keller's (1908) book *The World I Live In*; her descriptions also illustrate how this process can be facilitated by prior knowledge and augmented with imagery. For example, she noted: "My fingers cannot, of course, get the impression of a large whole at a glance; but I feel the parts, and my mind puts them together" (p. 7). Similarly: "My hand has its share in this multiple knowledge, but it must never be forgotten that with the fingers I see only a very small portion of a surface, and that I must pass my hand continually over it before my touch grasps the whole. It is still more important, however, to remember that my imagination is not tethered to certain points, locations, and distances. It puts all the parts together simultaneously as if it saw or knew instead of feeling them." (pp. 28–29). It is true of course that vision also incrementally builds distal representations, viz, through saccades or shifts in attention (Ullman, 1984), but because this process is not limited by the speed of grasping and reaching, it plausibly occurs over a much shorter time scale, lessening the load on working memory. In short, while C/E blindness is characterized by a greater than normal reliance on serial processing, compensatory processes lead to superior serial processing abilities and to integrated subjective experience.

In addition, we speculate that a preserved form of visual imagery is involved in the spatial integration abilities of C/E blind people. A reasonable amount of evidence supports the presence of imagery in C/E blindness, including: (1) visual cortex areas are activated by input from other senses in blind and non-blind people (Amedi et al., 2001, 2002; Ortiz et al., 2011); (2) the occipital lobe helps generate mental representations even when there has been no prior visual input (Amedi et al., 2008); (3) after training with tactile stimulation, blind people reported visual qualia, and the extent of these reports were related to ERP activity in the lateral occipital complex (Ortiz et al., 2011); (4) C/E blind people have visual content in their dreams, and can draw it, and the extent of their dream imagery is negatively correlated with EEG alpha band power, just as it is in sighted individuals (Bertolo et al., 2003); (5) C/E blind people report visual imagery (Cornoldi et al., 1979; Zimler and Keenan, 1983), although this is reduced, and more verbally mediated, compared to sighted people (Cornoldi et al., 1979); (6) however, spatial representation is not completely verbally mediated in blind people; during the encoding and working memory maintenance phases of a haptic mental rotation task, both sighted and blind subjects demonstrated increased parietal activation (as demonstrated in ERP data; Roder et al., 1997); (7) spatial images are supramodal, in that they are not simply the sum of modality-specific inputs; (8) there are several characteristics of visual experience, such as size, contour, and edge information, shape and texture that can be perceived through touch, and blind people can generate visual representations of haptically explored objects (reviewed in Cattaneo et al., 2008; Cattaneo and Vecchi, 2011); (9) vision and haptics may share common representations (Aleman et al., 2001); and (10) both the dorsal and ventral pathways are thought to contain supramodal regions to code shape and location of objects, respectively, regardless of the nature of the sensory input (Pietrini et al., 2009). We suggest, based on the quotes cited above, and others in the literature, that these "visual" experiences and supramodal representations aid in integrating subjective experience in C/E blindness. However, direct evidence for this claim has not yet been reported.

#### **C/E BLINDNESS IS PROTECTED FROM NEGATIVE DEVELOPMENTAL EFFECTS OF VISUAL PROCESSING IMPAIRMENTS**

Vision allows for the simultaneous perception of a great deal of information, and therefore for attention to be spread over many stimuli, or to be switched between stimuli. In schizophrenia, with its related sensory gating deficits, this can lead to the experience of being flooded with stimuli, and of stimuli being more intense than usual (Carr and Wale, 1986). In contrast, blindness is associated with touch as the primary method of exploring the environment. The nature of haptic perception is such that the person can only attend to the object currently being touched, and in this way attention cannot be involuntarily drawn to other objects (unless there are auditory or olfactory cues to other objects; Cattaneo and Vecchi, 2011).

Of course, we are not saying that vision is, by itself, a risk factor for schizophrenia. However, in combination with sensory gating, perceptual organization, and working memory impairments, visual processing, with its parallel inputs, may contribute to sensory flooding and subsequent disorganized cognition and behavior in schizophrenia. That is, visual *impairment* may be a risk factor for schizophrenia, not only in being reflective of abnormal neural development but also by being a cause of it. Conversely, C/E blindness may exert its protective effects against schizophrenia, in part, by preventing the possibility of abnormal visual input. Evidence consistent with this comes from several sources, including: (1) visual dysfunction in children of mothers with schizophrenia predicts the later development of schizophrenia (Schubert et al., 2005); (2) children (regardless of parental history) who later developed schizophrenia-spectrum disorders had significantly higher eye exam scale and strabismus scale scores compared to children who developed other non-psychotic psychopathology and children who did not develop a mental illness, while functioning in other sensory domains was far less impaired (Schiffman et al., 2006); and (3) the high specificity of visual abnormalities for predicting transition to schizophrenia among at-risk youth (Klosterkotter, 1992). In all of these cases, even if the presence of visual dysfunction is a reflection of the diathesis for schizophrenia rather than a primary cause of it, the continued state of abnormal visual input may lead to further perceptual and cognitive problems throughout development that increase the risk of developing the full-blown disorder.

Impaired visual functioning may also compromise multisensory integration,which,in turn,may lead to other cognitive deficits in schizophrenia. Efficient integration of different sensory inputs requires remapping these into an external spatial reference frame (Roder et al., 2007), which requires normal vision to properly develop (Collignon et al., 2009). However, in schizophrenia, there are multiple visual processing impairments (e.g., in gain control, perceptual organization, motion perception, etc.; Butler et al., 2008). Multisensory integration is also impaired in schizophrenia (Williams et al., 2010), and it is thought that deficits in visual processing contribute to this (Landgraf et al., 2012). In contrast, blind people appear to rely on somatotopic spatial coordinates as opposed to external, visual, coordinates (Roder et al., 2004). Therefore, in contrast to schizophrenia, in C/E blindness, the primary sense used to create a reference frame for multisensory integration (i.e., touch) is associated with superior perceptual abilities compared to sighted people.

Blind people rely primarily on vestibular and somatosensory feedback for motor control. This leads to enhanced voluntary motor control and motor-kinesthetic processing (Deutschlander et al., 2009). In contrast, schizophrenia is associated with impaired motor control (Cortese et al., 2005; Putzhammer and Klein, 2006; Velasques et al., 2011), and it has been hypothesized that this could be related to passivity phenomena and the development of delusions of control (e.g., the belief that one is not the agent of one's own actions, and/or that an external entity is controlling one's thoughts and/or actions; Danckert et al., 2004). Abnormalities in movement have also been observed in people at high risk for schizophrenia (Mittal et al., 2008). Therefore, the enhanced development of motor control and kinesthetic processing in C/E blindness can be viewed as protective against these schizophrenia-related symptom formation processes.

Finally, it has been recently proposed that visual processing disturbances in individuals at-risk for schizophrenia may contribute to the later development of the disorder by causing impairments in body perception, involving the senses of "ego boundaries," body ownership, body agency, and sense of self in space. These phenomena may then lead to problems in sense of identity, and in awareness of symptoms and insight into illness (Landgraf et al., 2012). Body ownership and agency can be measured experimentally with the "rubber hand illusion" (Ehrsson et al., 2004). In this paradigm, subjects have the experience that they are touching their own right hand with their left index finger, when in reality they are touching a rubber hand with their left index finger while the

experimenter touches their right hand in a synchronized manner. This effect is enhanced in schizophrenia and is thought to reflect abnormal body perception including a reduction in sense of body ownership (Thakkar et al., 2011). In contrast, in people with C/E blindness, the illusion is not experienced (Petkova et al., 2012). These results have been interpreted as indicating that C/E blind people have a more veridical perception of self-touch, along with a *less* flexible and dynamic representation of their own body in space compared to sighted individuals (Petkova et al., 2012). We believe that this reduced flexibility in dynamic representation of the body, secondary to loss of vision, is a protective factor against the body perception abnormalities associated with schizophrenia that were noted above.

# **NEUROPLASTICITY**

In this section we review evidence on how and where in the brain changes occur in response to C/E blindness, and how these subserve the changes in cognitive functioning discussed above. There is a large literature on brain plasticity in blind people, including evidence for developmental crossmodal plasticity (see below) – where areas of the occipital lobe are recruited for auditory and haptic perception (see Cattaneo and Vecchi, 2011 for a review), and also for olfaction (Kupers et al., 2011). Beyond this, however, other changes in developmental *intra*modal plasticity can occur which appear to result in perceptual and cognitive abilities that are opposite to those found in schizophrenia. For example, a magnetic source localization study (Elbert et al., 2002) indicated that dipoles associated with processing low- and high-frequency tones were more distant in blind people than in sighted subjects, suggesting an expansion of regions in the auditory cortex, supporting the superior pitch discrimination abilities noted above. Further evidence for intramodal plasticity in blind individuals is *reduced* hemodynamic responses to auditory stimulation in comparison to sighted subjects, which is thought to reflect *increased* efficiency in auditory signal processing within the temporal lobe, in light of the generally superior auditory processing abilities in this population (Stevens and Weaver, 2009). Evidence for changes regarding haptic perception comes from findings of *reduced* accuracy in identifying which finger is touched during a sensory threshold task (Sterr et al., 1998, 2003), within the context of overall superior tactile perception (Wong et al., 2011). These data suggested altered cortical representations for touch in the sense that the cortical representations of individual fingers are "fused" together due to their simultaneous use during Braille reading. Moreover, these data are strongest for the hand used to read Braille, and do not generalize to other body parts that are not used for reading, such as lips. Of note, the "holistic" processing of Braille letter patterns and reduced precision of individual finger-level input is in contrast to what is often observed in the visual processing of people with schizophrenia, where processing of holistic visual patterns is impaired, sometimes with enhanced processing of individual visual details (reviewed in Uhlhaas and Silverstein, 2005; Silverstein and Keane, 2011a).

Developmental crossmodal plasticity can lead to superior performance compared to sighted subjects, and there are several interesting instances of this relevant to schizophrenia. For example, somatosensory and auditory ERP data from oddball and MMN

tasks (which involve detection of a deviation from a sequential pattern in an attended or unattended sequence of stimuli, respectively),indicate normal ERP amplitude,with more activity over the occipital cortex in blind subjects, and enhanced pre-attentional processing in blind compared to sighted subjects (Kujala et al., 1995; Roder et al., 1996). This is in marked contrast to abnormal ERP amplitudes during oddball and MMN tasks in people with schizophrenia (e.g., Kayser et al., 2001) as noted above. Aside from developmental neuroplasticity, other ERP data from oddball tasks in people with schizophrenia suggested reduced short-term neuroplasticity. For example, when people with schizophrenia are presented with a high-frequency sequence of individual auditory stimulations, their ERP signatures reveal abnormally reduced long-term potentiation (Mears and Spencer, 2012). In contrast, short-term neuroplasticity, at least involving the occipital cortex, may be greater in people with C/E blindness (De Volder et al., 1999).

A major focus of research on visual processing in schizophrenia is impaired functioning in the dorsal and ventral visual streams. Therefore, we briefly summarize evidence that, in contrast to what is observed in schizophrenia, functioning in these pathways is intact in C/E blindness, *despite the absence of visual input*. Much evidence suggests that the dorsal stream is specialized for determining where an object is and for guiding action, whereas the ventral stream is specialized for analyzing form and color for determining what an object is (Goodale, 1993; Goodale et al., 2005). Many studies indicate impaired functioning of both of these streams in schizophrenia (e.g., King et al., 2008; Silverstein et al., 2009; Sehatpour et al., 2010; Lalor et al., 2012; Plomp et al., 2012). Surprisingly, these pathways appear to subserve similar functions in blind people as they do in psychiatrically healthy sighted people, suggesting that they can develop even in the absence of visual experience, and therefore that their representations are supramodal (Cattaneo and Vecchi, 2011; Ptito et al., 2012). For example, during spatial imagery tasks using auditory or tactile stimuli, early blind and sighted individuals demonstrated similar dorsal stream and frontal eye field activation, as assessed via PET or fMRI (Vanlierde et al., 2003; Garg et al., 2007;Bonino et al., 2008). In addition, dorsal areas devoted to processing motion information in sighted people (e.g., V5/MT) are employed in blind people in response to tactile and auditory motion cues (e.g., Ptito et al., 2009; Bedny et al., 2010; Matteau et al., 2010). These areas are notably underactivated during perception of visual motion in schizophrenia, in the context of impaired motion perception (Chen, 2011). In the ventral stream, blind subjects demonstrated similar activation during tactile object perception to that shown by sighted subjects during visual and haptic exploration of the same objects (Pietrini et al., 2004; Ptito et al., 2012).

It should be noted that while occipital lobe activity can occur in sighted subjects during tactile and auditory processing tasks, this has been found to be due to visual imagery (Sathian, 2005), which activates early visual cortex areas in a retinotopic fashion (Slotnick et al., 2005), and to calibration of head-centered sound localization (Zimmer et al., 2004). In contrast, occipital cortex activity in C/E blind subjects appears to support actual non-visual stimulus processing, and to represent a true crossmodal reorganization (see Roder et al., 1997; Cattaneo and Vecchi, 2011 for a

review). Consistent with this, transcranial magnetic stimulation (TMS), applied to the occipital cortex, disrupts Braille reading ability in blind individuals, but not in sighted individuals, whereas the reverse is true when TMS is applied over the somatosensory cortex in the parietal lobe (Cohen et al., 1997). Similarly, TMS over the occipital cortex induces tactile sensations in blind readers' fingers, but only phosphene perception for sighted subjects (Ptito et al., 2008).

As noted above, schizophrenia patients have demonstrated olfactory deficits in several studies, while C/E blind individuals have demonstrated superior olfactory abilities. A recent study, comparing congenitally blind subjects to blindfolded sighted controls, demonstrated that this is due to enhanced processing in areas normally dedicated to olfaction (e.g., amygdala, hippocampus) *and* in multiple regions of the occipital cortex (Kupers et al., 2011). This is further evidence of both the supramodal nature of representations in the C/E visually deprived occipital lobe, and also of crossmodal plasticity in C/E blindness.

In addition to subserving enhanced non-visual sensory processing, fMRI data indicate that the occipital lobe in C/E blind individuals also supports syntactic and semantic (more so than phonological) aspects of language processing (Roder et al., 2002; Burton et al., 2003), as well as verbal memory, episodic memory, and verbal fluency (Amedi et al., 2003; Raz et al., 2005), all of these being processes that are impaired in schizophrenia (Tendolkar et al., 2002; Fisher et al., 2009; Costafreda et al., 2011). For example, in addition to occipital lobe activation during memory retrieval, and a correlation between occipital activation and episodic memory in blind but not sighted subjects (Raz et al., 2005), repetitive transcranial magnetic stimulation (rTMS) over the occipital lobe impaired performance in a verb generation task for blind but not sighted subjects (Amedi et al., 2004).

Another consequence of blindness may be increased communication between brain regions. Several studies have demonstrated increased effective connectivity between the occipital lobe and temporal and prefrontal regions in early blind individuals (e.g., Noppeney et al., 2003; Leclerc et al., 2005). This is in contrast to schizophrenia, where reduced connectivity is typically found (e.g., Kim et al., 2008; Dima et al., 2009, 2010), including between occipital and temporal and frontal regions (Rigucci et al., 2012).

Finally, Sanders et al. (2003), based on evidence of visual cortex plasticity in dark-reared animals (Chen and Bear, 2007; Chen et al., 2007), proposed that an effect of early blindness is increased NMDA-receptor activity (although accompanied by reduced numbers of receptors) in visual cortex, and in regions with which this has many connections, primarily the anterior cingulate cortex (ACC), which is involved in resolving response competition and context sensitivity, two cognitive domains that are impaired in schizophrenia (Cohen and Servan-Schreiber, 1992; Dolan et al., 1995; Phillips and Silverstein, 2003). Moreover, schizophrenia has been characterized by NMDA-receptor hypofunction (Phillips and Silverstein, 2003), and, in numerous studies, by ACC hypoactivation (Dolan et al., 1995). The consequences of NMDAreceptor hypofunction in schizophrenia have been hypothesized to include impairments in sensory gating, perception, working memory, learning, and the full range of positive and negative symptoms (Phillips and Silverstein, 2003; Coyle, 2012). Therefore, an increase in NMDA-receptor activity in C/E blindness could have widespread positive effects on cognition, and protective effects against symptoms.

In short, there are several ways in which the neuroplasticity that results from C/E blindness results in changes that are opposite to those observed in schizophrenia, and that may represent protective effects. Specifically, the evidence reviewed above suggests that, in C/E blindness: (1) despite the loss of visual input, cortical reorganization during early development leads to accurate form and space perception (via auditory and tactile processing) which is supported by brain areas that function abnormally in schizophrenia (e.g., dorsal and ventral visual streams); (2) this developmental plasticity also leads to use of occipital areas for functions such as preattentive processing (e.g., as reflected in the MMN), language, memory, and olfaction – all of which are impaired in schizophrenia – which may underlie the superior functioning in those areas in C/E blindness reviewed above; and (3) in addition to these sequelae of developmental plasticity in C/E blindness, there is also evidence of normal later-onset plasticity, which may in part reflect prolonged plasticity of the occipital cortex in C/E blindness (De Volder et al., 1999), and which is the basis for the success of sensory substitution devices (Ward and Meijer, 2010). In contrast, later-onset plasticity has been shown to be reduced in schizophrenia in several domains (Fitzgerald et al., 2004; Oxley et al., 2004; Daskalakis et al., 2008;Hasan et al., 2011, 2012; Mears and Spencer, 2012).

# **CONGENITAL OR EARLY BLINDNESS IS NOT PROTECTIVE AGAINST MENTAL DISORDERS IN GENERAL**

While C/E blindness may protect against schizophrenia, they do not protect against mental illness in general. A range of mental and behavioral disorders has been reported in blind people. For example, anxiety levels have been reported to be higher in congenitally blind than sighted adolescents (Bolat et al., 2011), and autistic symptoms and autism are common in blind children (Keeler, 1958;Wing, 1969;Chess, 1971;Chase, 1972; Fraiberg, 1977; Rogers and Newhart-Larson, 1989; Brown et al., 1997; Ek et al., 1998; Carvill, 2001). Sharp (1993) reported a case of anorexia nervosa and depression in a woman blind since the age of 9 months. A recent case study (Kocourkova et al., 2011) described a case of an adolescent patient, blind from early childhood,who showed symptoms of anorexia, depression, suicidal behavior, and self-harming. A conclusion in this report was that body image is a mental construct that is not dependent on sensory perception. Finally, Musial et al. (2007) reported a case of arachnophobia in a congenitally blind person, suggesting that visual cues are not necessary for the development of phobic disorders.

# **CONGENITAL DEAFBLINDNESS IS NOT PROTECTIVE AGAINST SCHIZOPHRENIA**

Protective effects of blindness may not exist if other forms of severe sensory loss are present. For example, congenital deafblindness is associated with elevated rates of psychosis, as well as mental retardation (Dammeyer, 2011). Also, children with Usher syndrome, in which people are typically deaf from birth and then develop retinitis pigmentosa leading to blindness in childhood, has been associated with psychotic symptoms and schizophrenia

(Dammeyer, 2012). The reason why adding deafness to blindness removes the apparent effect of the latter against schizophrenia is not clear at present. One possibility is that blindness by itself presents a surmountable challenge to cope with the environment and thereby fosters compensatory sensory, perceptual, and cognitive changes that lead to a surprisingly high level of functioning. Deafblindness, on the other hand, may so seriously restrict the opportunity for environmental interaction that it also stunts the development of cognitively based coping strategies. For example, congenital deafblindness typically leads to significant delays in achieving several cognitive milestones, and to profound impairments in speed of processing, the ability to integrate disparate items of information, attention, communication, and the capacity for symbolic understanding and expression (Geenens, 1999; Bruce, 2005; Shamma et al., 2011). It is worth noting here that although C/E blindness and C/E deafblindness are both relatively rare, there are no published cases of the former being comorbid with schizophrenia, whereas there are many reported cases of this comorbidity with the latter.

# **SENSORY LOSS IN GENERAL DOES NOT PROTECT AGAINST SCHIZOPHRENIA**

Another important consideration is that sensory loss *per se* does not protect against schizophrenia. For example, deafness occurs in schizophrenia at as high a rate as in the general population (Kitson and Fry, 1990), and congenital deafness is a risk factor for the development of psychotic symptoms (Thewissen et al., 2005; Atkinson, 2006), including, interestingly, higher rates of visual and somatic hallucinations than are found in the non-deaf people with schizophrenia (Cutting, 1985), which often co-occur with reports of hearing voices (Du Feu and McKenna, 1999). And, while deafness may be associated with compensatory cognitive changes (see Bross, 1979; Dye and Bavelier, 2010; e.g., heightened attention to visual peripheral cues in people who use sign language; Proksch and Bavelier, 2002), these appear to not protect against schizophrenia, although they may lead to subtle differences in profiles of cognitive impairment between deaf and hearing people with schizophrenia (Horton and Silverstein, 2007, 2011). Moreover, in some cases, compensatory changes may not develop until adulthood (Rettenbach et al., 1999).

# **MEDIATING BIOLOGICAL VARIABLES**

The hypothesized protective effects of blindness on schizophrenia may be mediated by other factors. For example, Horrobin (1979) proposed that abnormal pineal gland function and resulting prostaglandin activity are etiological factors in schizophrenia. This is supported by evidence of increased inflammation and reduced melatonin levels in schizophrenia patients (Anderson and Maes, 2012; Muller et al., 2012). Conversely, blind individuals have elevated melatonin levels (Bellastella et al., 1995), which has been shown to attenuate cognitive impairments (Peck et al., 2004) and reduce inflammation (Cuzzocrea et al., 1999; Sharma et al., 2012). Another potential mechanism involves cortical thickness and pruning. Schizophrenia is typically characterized by cortical thinning (Cobia et al., 2012), and it has been hypothesized that excessive neuronal pruning takes place during adolescence (Boksa, 2012). In contrast, the visual cortex of C/E blind people – which

as noted above is involved in multiple perceptual and cognitive processes – is thicker than in sighted people, and is thought to be characterized by a less than normal amount of pruning, due to deprivation of visual experience (Jiang et al., 2009). A consequence of this thickening is that, even if genes related to schizophrenia cause excessive pruning in someone with C/E blindness, the remaining number of neurons may still be greater than normal in some areas. This, in combination with the cortical reorganization noted above, leading to occipital regions being used for non-visual cognitive processes, may protect blind people against crossing the threshold needed for psychotic symptoms to emerge (i.e., an adequate number of neurons devoted to specific types of processing may be present even if pruning is excessive).

# **CONCLUSION**

This paper builds on previous reports of a lack of schizophrenia in people with C/E blindness by advancing the hypothesis that the latter alters cognition and fosters neuroplasticity in ways that confer protective effects against schizophrenia. The reviewed evidence indicates that: (1) schizophrenia is primarily a cognitive disorder; (2) many of the cognitive functions that are impaired in schizophrenia are enhanced among the C/E blind; (3) C/E blindness involves *reduced* flexibility in language and in dynamic representation of the body, and these reductions may protect against thought disorder and alterations in experience of the self, respectively; (4) the mechanisms noted in 2 and 3 above are significantly less affected if the onset of blindness is after the first few months of life; (5) other forms of C/E sensory loss do not protect against schizophrenia; and (6) C/E blindness does not protect against other disorders, suggesting that there is a special link between schizophrenia and visual processing.

Of course, schizophrenia is characterized by cognitive impairments in addition to those described here, such as in executive functioning, that are thought to rely heavily on functioning of the dorsolateral prefrontal cortex (DLPFC). There appears to be almost no research comparing blind (congenital or acquired) and sighted people on these functions, however, and so whether blindness confers any protective effects on them remains an open question. Note, however, that the DLPFC does not fully develop until late adolescence or early adulthood. In contrast, cognitive and academic decline, and interpersonal dysfunction, typically begin in late childhood or early adolescence in people who go on to develop schizophrenia, and this age range corresponds to the time frame in which the visual system in humans is most plastic (i.e., until ∼14– 16 years of age;Cohen et al.,1999). This suggests that the protective sequelae of C/E blindness involve changes in more basic sensory, perceptual, and cognitive (i.e., non-executive) functions.

While the idea of one abnormal condition having a protective effect against another condition may seem unusual, it is not unknown. For example, several studies have now replicated the finding that schizophrenia appears to have an inverse relationship with rheumatoid arthritis (RA), with some estimates putting the risk for RA among schizophrenia patients at 30% that of control counterparts (Eaton et al., 1992; Mors et al., 1999; Oken and Schulzer, 1999). Conversely, there may also be a reduced risk for developing schizophrenia among individuals with RA (Gorwood et al., 2004). This is especially low considering smoking is a risk factor for developing RA (Stolt et al., 2003) and that a disproportionally large percentage (60–90%) of individuals with schizophrenia smoke (Dervaux and Laqueille, 2008). As noted, we hypothesize that the cognitive benefits conferred by C/E blindness act as a prophylactic for schizophrenia but not other mental disorders. This hypothesis, if confirmed, has implications for determining when cognitive training should occur for people with, or at-risk for, schizophrenia, and for the nature of such training. It also has implications for whether insights from C/E blindness can be used to design interventions to prevent schizophrenia. Each of these issues will be addressed in turn below.

Evidence from animal studies indicates that prevention of schizophrenia-related cognitive impairment can be achieved by early cognitive training. For example, in one study, rats received a ventral hippocampus lesion that induces a spatial working memory deficit similar to that found among schizophrenia patients. It was found that among animals that had been provided with pre-lesioning cognitive training, the typical post-lesion working memory deficit did not arise (Lee et al., 2012). While similar preventive data do not exist in humans, a recent study of cognitive remediation in first-episode schizophrenia demonstrated that it significantly slowed the typical gray matter loss associated with the first years after an initial psychotic episode (Eack et al., 2010). Therefore, it is reasonable to study whether the long-term effects of cognitive training in people at high risk for the disorder could be at least as strong, if not stronger, than effects of cognitive training introduced after a diagnosis of schizophrenia is made, given that the effects are occurring in the context of a healthier and younger brain.

But when should training occur? Evidence indicates that intervening when people first begin to show psychotic symptoms or functional decline (i.e., in the "ultra high-risk" state) can delay, but does not prevent, schizophrenia (Yung and Nelson, 2011). Therefore, it has recently been proposed that intervening at a "pluripotent risk stage" – the time of the earliest emergence of behavioral and cognitive difficulties (typically 9–13 years of age) that are non-specific as to final diagnosis – would be more effective (Agius et al., 2010; McGorry, 2010; McGorry et al., 2010; Morgan et al., in press). Whether or not such a strategy can prevent schizophrenia, it may be an ideal time for intensive cognitive training, in terms of affording protection against later and severe schizophrenia-related cognitive impairment. Of course, even earlier training could be effective, but the feasibility of identifying people at high risk for mental illness at younger than 9-years old has not been well-established, except in cases where an identical twin or first-degree relative has the disorder.

What sort of training would be effective? Based on the evidence reviewed above, we suggest that cognitive training in young people at-risk for schizophrenia should be *comprehensive in nature*, and should focus to a greater degree than it has in the past on sensory and perceptual functioning. For example, in addition to a focus on improving aspects of attention, it should focus on improving sensory tuning and the accuracy of sensory representations, perceptual organization, multisensory integration, and controlled processing including integration of bottom-up sensory data with stored information (see Heller quotes above, and Hemsley, 1996). Some interventions have already demonstrated

success in improving functioning in these specific areas in humans (Temple et al., 2003; Mahncke et al., 2006; Genevsky et al., 2010) and animals (Polley et al., 2006; Zhou et al., 2011). In addition, although schizophrenia patients often demonstrate reduced plasticity, it is not absent, and capacity for perceptual and cognitive change has been demonstrated (Fisher et al., 2009, 2010; Silverstein and Keane, 2009) both in and outside the context of specific interventions. Therefore, it is possible that even greater plasticity can be harnessed earlier in the developmental course of the illness, before psychotic symptoms emerge and near the onset of, or prior to, cognitive and functional decline. In addition, it has already been demonstrated in schizophrenia that improving auditory precision by repeated practicing of tone discrimination leads to improvements in auditory and (more interestingly) higher level cognitive functions (e.g., verbal learning and memory; Adcock et al., 2009; Fisher et al., 2009). Therefore, broader improvements (beyond the trained task) may occur in at-risk youth with similar practice.

An additional strategy for people at-risk for schizophrenia involves increasing reliance on non-visual sensory modalities to enhance their functioning. Already it has been shown in people with established schizophrenia, for example, that steady-state auditory responses, which are typically abnormally reduced and associated with an increase in broadband noise, can be significantly improved, and to a greater degree than occurs in healthy people, by closing the eyes during task performance (Griskova-Bulanova et al., 2012). Fostering cognitive reserve, and trying to prevent schizophrenia, or at least minimize later illness-related cognitive impairment, by strengthening capacity and efficiency in non-visual domains is an as yet untested strategy. However, the goal of creating a schizophrenia-specific, *acquired* cognitive reserve, in a manner guided by compensatory changes observed in C/E blindness, would appear to be a reasonable and promising direction to pursue, given that premorbid cognitive reserve predicts later cognitive functioning in normal aging (Steffener and Stern, 2012), neurodegenerative disorders (Koerts et al., 2012; Stern, 2012), and schizophrenia (de La Serna et al., 2013).

Can an understanding of cognition and neuroplasticity in C/E blindness be used to reduce the likelihood of schizophrenia? At first glance, the answer seems to be "no" since: (1) only C/E blindness (but not blindness acquired after the first few months of life) is associated with a protective effect; (2) it is difficult to identify most people who will develop schizophrenia during the first year of life; (3) it is not clear what type of training could be given to people less than 1-year old, given limited attention span and undeveloped verbal abilities; and (4) the presence of vision is likely to reduce the extent of developmental neuroplasticity. However, the ultimate answer to this question may not be this simple. For example, much remains to be learned about biological and computational changes associated with C/E, and later-developing, blindness, and so further investigation into these issues may reveal clues for intervention development. Also, it is not known beyond what age fostering neuroplasticity and cognitive enhancement might no longer be able to prevent schizophrenia, but only reduce schizophrenia-related cognitive impairment. Third, since it is not C/E blindness *per se*, but rather, the corresponding brain and cognitive changes that are protective against schizophrenia, it is possible that methods other than naturally occurring C/E blindness can confer protective effects. For example, an open question is the extent to which particularly intensive sensory-perceptual-cognitive interventions delivered to at-risk youth past the first year of life can create some of the protective changes associated with naturally occurring C/E blindness.

#### **REFERENCES**


correlates of drawing in an early blind painter: a case study. *Brain Res.* 1242, 252–262.


#### **ACKNOWLEDGMENTS**

We thank Thomas Papathomas, Matt Roche, Keith Feigenson, James Gold, and Judy Thompson for their helpful comments on earlier drafts of this manuscript.We also thank Ms. Arwen Lockley for her helpful insights on blindness and schizophrenia, and the two reviewers for their helpful suggestions. This work was supported by a National Research Service Award (F32MH094102) to Brian P. Keane, and R01MH093439 to Steven M. Silverstein.


detection analysis of a visual discrimination task. *Percept. Mot. Skills* 48, 187–194.


parietal cortex. *Schizophr. Res.* 70, 241–261.


McInnes (Toronto: University of Toronto Press), 150–174.


plasticity in schizophrenia revealed by transcranial direct current stimulation. *Behav. Brain Res.* 224, 15–22.


attentional sensitivity for identification of speech sounds. *Brain Res. Cogn. Brain Res.* 19, 28–32.


*in Blind Children with Retrolental Fibroplasia*. New York: Grune & Stratton.


in an auditory detection and discrimination task: exploring the 'selfmonitoring' model of hallucination. *Schizophr. Res.* 55, 115–128.


*Schizophrenia: Evolution and Synthesis*, eds M. B. Silverstein and T.Wykes (Cambridge: MIT Press).


cognitive deficits. *Dev. Psychopathol.* 24, 529–536.


*Proc. Natl. Acad. Sci. U.S.A.* 101, 5658–5663.


search compensate in adult but not in juvenile subjects. *J. Cogn. Neurosci.* 11, 560–583.


individuals. *Neuropsychologia* 47, 843–848.


by phosphamidon. *Fundam. Clin. Pharmacol.* doi:10.1111/j.1472- 8206.2011.00977.x


working memory impairments in schizophrenia. *Arch. Gen. Psychiatry* 59, 146–153.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 03 November 2012; accepted: 31 December 2012; published online: 21 January 2013.*

*Citation: Silverstein SM, Wang Y and Keane BP (2013) Cognitive and neuroplasticity mechanisms by which congenital or early blindness may confer a protective effect against schizophrenia. Front. Psychology 3:624. doi: 10.3389/fpsyg.2012.00624*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Silverstein, Wang and Keane. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, providedthe original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Base rates, blindness, and schizophrenia

# *Steven M. Silverstein1,2\*, Yushi Wang1 and Matthew W. Roché1*

*<sup>1</sup> University Behavioral HealthCare, University of Medicine and Dentistry of New Jersey, Piscataway, NJ, USA*

*<sup>2</sup> Department of Psychiatry, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, Piscataway, NJ, USA*

*\*Correspondence: silvers1@umdnj.edu*

#### *Edited by:*

*Michael Green, University of California, Los Angeles, USA*

*Reviewed by:*

*Michael Green, University of California, Los Angeles, USA*

#### **A commentary on**

**Cognitive and neuroplasticity mechanisms by which congenital or early blindness may confer a protective effect against schizophrenia**

*by Silverstein, S. M., Wang, Y., and Keane, B. P. (2012). Front. Psychol. 3:624. doi: 10.3389/fpsyg.2012.00624*

Our paper on congenital and early (C/E) blindness and schizophrenia was an effort to account for the previously reported negative relationship between these conditions, in light of recent work indicating what the study of C/E blind people reveals about brain function and organization (Cattaneo and Vecchi, 2011; Kupers et al., 2011; Ricciardi and Pietrini, 2011). In particular, we hypothesized that the neuroplastic changes inherent to C/E blindness protect against schizophrenia by promoting strengths in those cognitive domains where schizophrenia is characterized by deficits.

Since our paper was published, we have received several comments asking a version of this question: can the lack of reports of people with both conditions be explained simply by the rarity of each condition, so that the chances of a person having both are so low that even if such people existed, it is unlikely their condition will have been reported? This is an important question, as when studying low base rate phenomena, distorted conclusions can easily be reached (Meehl and Rosen, 1955). We addressed this issue in the original version of our manuscript, but then removed it from the final version due to length considerations. Here, we note our earlier comments.

The conclusion that there are no C/E blind people with schizophrenia is based on a small number of studies that involved relatively small samples. Clearly, this argument would be strengthened by larger, population-based studies. This is because, as a simple calculation demonstrates, a case of congenital blindness and schizophrenia would be extremely rare even if there was no protective effect of blindness: if schizophrenia occurs at a rate of 0.72% in the population (McGrath et al., 2008), and congenital blindness occurs at an estimated rate of 0.03% in people born in the 1970s and 1980s (based on Robinson et al., 1987), then the joint probability of a person having both conditions, if the two are independent, would be 0.02% or 2 out of every 10,000. Although this is a low prevalence rate, it is higher than the rates for childhood-onset schizophrenia (Remschmidt and Theisen, 2005), and many other well-known medical conditions (e.g., Hodgkin's lymphoma, Prader Willi syndrome, Rett's Syndome). Based on this estimated prevalence rate, in the United States alone (with a population of 311, 591, 917, as of July 2011, according the US census), there should be approximately 620 congenitally blind people with schizophrenia. When cases of blindness with an onset in the first year of life (i.e., early blindness) are taken into account, the percentage would be larger. Therefore, it is remarkable that in over 60 years, and with several investigations [including several before DSM-III (1980) when criteria for schizophrenia were broader than at present], not a single case of a C/E blind schizophrenia patient has been reported. Moreover, several published studies, and our experience as well, included surveying Directors of agencies that serve large numbers of blind people, and none of them could recall ever seeing a person who

had both conditions. It is also interesting that rates of C/E blindness are significantly higher in developing, compared to industrialized, countries. Therefore, if C/E blindness did *not* protect against the development of schizophrenia, comorbidity would be more likely to be reported in such countries. However, this has not occurred. In short, available evidence, probabilistic estimates, and the striking contrasts, within the same domains of cognition, between superior functioning in C/E blindness and impaired functioning in schizophrenia, combine to suggest a protective relationship. If the conditions did co-occur at chance levels, reports of such cases should appear at least somewhat as often as those of many other rare medical conditions, especially since reports of an absence of schizophrenia in C/E blind people have appeared since 1950 (Chevigny and Braverman, 1950). The absence of such reports is noteworthy.

One research strategy that could generate further evidence on this issue is the study of schizotypal symptoms in the C/E blind. These subsyndromal psychotic symptoms have a higher base rate than schizophrenia itself (i.e., 10–15% vs. ∼1%) (Fossati and Lenzenweger, 2009), and people with these symptoms share some of the same cognitive and biological impairments as people with the full disorder (Aichert et al., 2012; Cochrane et al., 2012). Studies of schizotypal symptoms would allow for the determination of whether C/E blindness protects against the full spectrum of schizophrenia-related illness or just schizophrenia itself. If no schizotypal symptoms were observed, this would be evidence for the existence of protective mechanisms against all schizophrenia-related psychopathology. If, however, schizotypal symptoms were found, samples of C/E blind people could be studied to gain insight into the neurobiological changes that protect against the development of clinical psychosis. As such, expanding the schizophrenia phenotype studied in the C/E blind holds potential for increasing our understanding of the mechanisms involved in schizophrenia.

In addition to these points, we noted in our original paper that other congenital sensory impairments have not been associated with absence of schizophrenia, that congenital deafness and later blindness, such as found in Usher Syndrome, whose prevalence has been estimated at 0.005% (Rosenberg et al., 1997) has been associated with psychotic disorders at rates between 4–24% (Waldeck et al., 2001), and that non-psychotic psychiatric disorders are observed in C/E blindness. These data suggest a unique relationship between C/E blindness and schizophrenia. However, we acknowledge that the absence of evidence (of people with both conditions) is not evidence of absence. That said, given the potentially important lessons for understanding, preventing, and treating schizophrenia that brain reorganization in C/E blindness provides, we believe it is useful to pursue this line of thought. At the very least, if C/E blindness did *not* prevent schizophrenia, then in light of the cognitive changes secondary to C/E blindness, it would be informative to determine whether it mitigates the associated cognitive impairment. Finally, we hope to raise awareness in the psychiatric community of past literature relating to C/E blindness and schizophrenia, so that if people with both conditions exist, their cases will be reported.

#### **ACKNOWLEDGMENTS**

We thank Michael Green for his constructive criticism on an earlier draft of this commentary.

# **REFERENCES**


incidence, prevalence, and mortality. *Epidemiol. Rev.* 30, 67–76.


*Received: 10 March 2013; accepted: 13 March 2013; published online: 03 April 2013.*


# "To see or not to see: that is the question." The "Protection-Against-Schizophrenia" (PaSZ) model: evidence from congenital blindness and visuo-cognitive aberrations

# *Steffen Landgraf 1,2\* and Michael Osterheider <sup>1</sup>*

*<sup>1</sup> Department for Forensic Psychiatry and Psychotherapy, District Hospital, University Regensburg, Regensburg, Germany <sup>2</sup> Berlin School of Mind and Brain, Humboldt Universität zu Berlin, Berlin, Germany*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Mahesh Menon, University of Toronto, Canada José Alberto González-Hernández, Hermanos Ameijeiras Hospital, Cuba*

#### *\*Correspondence:*

*Steffen Landgraf, Department of Forensic Psychiatry and Psychotherapy, District Hospital, University Regensburg, Haus 45, Raum 001, Universitätsstraße 84, 93053 Regensburg, Germany e-mail: steffen.landgraf@ukr.de*

The causes of schizophrenia are still unknown. For the last 100 years, though, both "absent" and "perfect" vision have been associated with a lower risk for schizophrenia. Hence, vision itself and aberrations in visual functioning may be fundamental to the development and etiological explanations of the disorder.In this paper, we present the "**P**rotection-**A**gainst-**S**chi**z**ophrenia" (PaSZ) model, which grades the risk for developing schizophrenia as a function of an individual's visual capacity. We review two vision perspectives: (1) "Absent" vision or how congenital blindness contributes to PaSZ and (2) "perfect" vision or how aberrations in visual functioning are associated with psychosis. First, we illustrate that, although congenitally blind and sighted individuals acquire similar world representations, blind individuals compensate for behavioral shortcomings through neurofunctional and multisensory reorganization. These reorganizations may indicate etiological explanations for their PaSZ. Second, we demonstrate that visuo-cognitive impairments are fundamental for the development of schizophrenia. Deteriorated visual information acquisition and processing contribute to higher-order cognitive dysfunctions and subsequently to schizophrenic symptoms. Finally, we provide different specific therapeutic recommendations for individuals who suffer from visual impairments (who never developed "normal" vision) and individuals who suffer from visual deterioration (who previously had "normal" visual skills). Rather than categorizing individuals as "normal" and "mentally disordered," the PaSZ model uses a continuous scale to represent psychiatrically relevant human behavior. This not only provides a scientific basis for more fine-grained diagnostic assessments, earlier detection, and more appropriate therapeutic assignments, but it also outlines a trajectory for unraveling the causes of abnormal psychotic human self- and world-perception.

**Keywords: schizophrenia, blindness, visual aberrations, Protection-Against-Schizophrenia (PaSZ), vision therapy, continuous diagnostic criteria, early detection, cognition**

# **INTRODUCTION**

#### **THE PROTECTION-AGAINST-SCHIZOPHRENIA MODEL**

"To see or not to see" may be the fundamental question by which we can elucidate the still unknown causes (Insel, 2010) of one of the most devastating human experiences – schizophrenia. For the last 100 years, both "absent" and "perfect" vision have been associated with a lower risk for the disorder (Landgraf et al., 2012; Silverstein et al., 2013). Therefore, we argue that vision itself and aberrations in visual functioning may be fundamental to the development of the disorder and, thus, may provide important information about its causes. In this article, we present the "**P**rotection-**A**gainst-**S**chi**z**ophrenia" (PaSZ) model by reviewing human visual functioning from two perspectives: (1) "Absent" vision or how congenital blindness contributes to PaSZ and (2) "perfect" vision or how aberrations in visual functioning are associated with an increased risk for psychotic symptomatology. Grading the risk for developing schizophrenia as a function of an individual's visual capacity, we argue that (early) diagnostic and interventional approaches need to be specific to the patient's visual capacity. On the one hand, individuals who suffer from visual deterioration (who previously had "normal" visual skills) may reduce their risk for developing schizophrenia through an *improvement* in visual capacity. On the other hand, individuals who suffer from visual impairment (who have never developed "normal" vision) may, in fact, reduce their risk for schizophrenia through a *decline*in visual capacity (**Figure 1**). Hence, rather than categorizing human behavior into, e.g., "normal" and "mentally disordered," as has traditionally been the practice of diagnostic manuals (APA, 2000; WHO, 2007), our approach may be the first to use a continuous scale to represent psychiatrically relevant human behavior. This not only provides a scientific basis for more fine-grained diagnostic assessments, earlier detection, and more appropriate therapeutic assignments, but this model also outlines a trajectory for unraveling the causes of the schizophrenia disorder.

Current models of developmental trajectories of schizophrenia emphasize the prognostic value of subclinical disease expressions.

**FIGURE 1 | The "Protection-Against-Schizophrenia" (PaSZ) Model.** The continuous PaSZ model depicts the relative risk for schizophrenia as a function of the continuous variable visual capacity. Whereas both "absent" vision (congenital blindness) and "perfect" vision ("supernormal" vision) may be associated with a decreased risk for schizophrenia, the model suggests that the risk for developing schizophrenia increases from both ends of the visual capacity continuum toward a "peak risk" (Landgraf et al., 2012; Silverstein et al., 2013). The location of this peak risk has yet to be determined experimentally. However, the peak has important implications for the understanding of the etiology, development, and therapy of the disorder: individuals suffering from visual impairment (located to the left of the peak), who never developed "normal" vision, may reduce their risk for developing schizophrenia through a *decline* in visual capacity. Individuals suffering from visual deterioration (located to the right of the peak), who previously had "normal" visual skills, may reduce their risk for developing schizophrenia through an *improvement* in visual capacity. Note that the

In his inspirational *Nature* article, National Institute of Mental Health Director Insel (2010) describes

*"psychosis as a late, potentially preventable stage of the illness" (p. 187).*

Thus, current diagnostic criteria and etiological models of the disease may be indicative of only ultimate disease stages. Even more problematic, the *prognostic* or prodromal1 criteria of schizophrenia model does not make a concrete assumption on the association between vision capacity and risk for schizophrenia (linear, exponential, etc.). Instead, we suggest that extensive longitudinal and epidemiological investigations, also controlling for age-related visual capacity decline (Ofan and Zohary, 2007; Cattaneo et al., 2008), are necessary to elaborate this issue. In this context, visual capacity may comprise but is not limited to measures of visual acuity (near/far), sensitivity to light, motion, and color, visual field size, and stereoscopic vision. To the best of our knowledge, this is the first model that uses a continuous (more vs. less psychotic) rather than a categorical ("normal" vs. "mentally disordered") approach to represent psychiatrically relevant human behavior. Abbreviations: PaSZ = Protection-Against-Schizophrenia; SZ = schizophrenia; Prodromals = individuals identified at ultra-high risk for developing schizophrenia; 1st episode patients = patients with schizophrenia that have had one (identified) psychotic episode; Chronic patients = patients with schizophrenia that have had at least three (identified) psychotic episodes.

are still being defined in terms of diagnostic symptomatology, i.e., sub-threshold phenotypic expressions of actual full-blown psychotic episodes (Klosterkotter et al., 2001; McGorry et al., 2003; Yung et al., 2005). Currently, to be identified as prodromal for an impending psychosis, individuals need to present one of the following three criteria: (1) attenuated (sub-threshold) psychotic symptoms, (2) BLIPS or brief limited intermittent psychotic symptoms (lasting less than 1 week and disappearing spontaneously), and (3) functional deterioration in the presence of vulnerability (they must suffer from schizotypal personality disorder or have a first-degree relative who has psychosis) (McGorry et al., 2003; Miller et al., 2003; Yung et al., 2005). Other descriptions

<sup>1</sup>For intelligibility, we use the term "prodromal" to refer to prodromal, ultra-high risk (UHR), or other patients assumed to experience signs of an impending psychotic episode.

of prodromal individuals, such as basic symptoms (Moyer and Landauer, 1967; Gross, 1969, 1989; Klosterkotter et al., 1997, 2001) by which early and late prodromal phases are distinguished (Niendam et al., 2007; Schultze-Lutter et al., 2007; Simon et al., 2007),<sup>2</sup> also presume symptom continuity along with the progression of the disease. The (scientifically) recognized stages of schizophrenia include, in fact, the prodrome, the first-episode, and the stabilized chronic syndrome. Problematically, current early detection and intervention are based on the theoretical yet unproven assumption that symptoms of schizophrenia extend continuously from the prodrome to the stabilized syndrome (Gross, 1969, 1989; Klosterkotter et al., 1997, 2001; Simon et al., 2006; Niendam et al., 2007; Schultze-Lutter et al., 2007; McGorry, 2010). Consequently, prodromal approaches have not yielded high predictive power in their ability to identify transitions to full-blown psychosis (Gottesman and Erlenmeyer-Kimling, 2001; Miller et al., 2002; McGorry et al., 2003; Yung et al., 2003, 2004, 2005; Haroun et al., 2006; Olsen and Rosenbaum, 2006a,b). In their recent review, Gee and Cannon (2011) found that only about one third of prodromal individuals eventually convert to psychosis. Two thirds remain symptomatic at a sub-threshold level or recover completely. Hence, we and other researchers argue that current diagnostic criteria (APA, 2000) systematically underestimate the individuality of symptoms (Andreasen, 2007; Nelson et al., 2008), neglecting, for example, the fact that psychoses are often experienced in phases that include self-disturbances (Parnas et al., 1998; Sass and Parnas, 2001, 2003) and visuocognitive impairments (Elvevag and Goldberg, 2000; Andreasen and Black, 2006; Keefe and Harvey, 2012; Landgraf et al., 2012). This means, first, that prodromal extrapolations of current diagnostic criteria become less valid or less useful the earlier the prognostic verdict is acquired. Second, there may be underlying factors that are not included in the diagnostic criteria of schizophrenia and treatment options; nevertheless, these factors may be crucial for understanding the ontogenetic development and etiology of the disorder. Thus, there is still an imperative need for stage-specific disease progression markers to precisely predict deterioration and the functional outcome. One promising factor for this kind of progression marker is visually mediated cognition.

#### **GOALS OF THE REVIEW**

Since the introduction of the term *"Schizophrenia"* by Bleuler (1908), a plethora of scientific articles have stressed the importance of vision aberrations in schizophrenia. By contrast, "absent" and "perfect" vision have, to the best of our knowledge, been associated with a lower risk for developing schizophrenia. Hence, visual information acquisition and/or processing at some point in life appears to be necessary but not sufficient for the human brain to develop psychosis. Only disturbed visual processing may be sufficient. We therefore present the idea of the continuous PaSZ model, for which we review two vision-related indicators of the etiology and development of the disease.

In the first part of this paper – the blindness perspective – we review cognitive alterations with regard to real-world mental representations, neurofunctional reorganization, and multisensory integration in congenitally blind individuals. We assess how these altered functions provide insight into the developmental causes of schizophrenia and the protection against it.

In the second part – the vision perspective – we provide evidence for how abnormalities in the visual system can lead to schizophrenia. Specifically, we show that visual information acquisition and processing deficits progress along with the progression of the disease. We describe how perceptual abnormalities contribute to higher-order cognitive dysfunctions and subsequently lead to diagnostic phenotypic expressions (symptoms).

Finally, in the third part – the therapeutic perspective – we review interventional implications of the PaSZ model. Specifically, we show how both an *increase in visual impairment* and a *decrease in visual deterioration* may contribute to lowering a person's risk for schizophrenia.

# **THE BLINDNESS PERSPECTIVE ON SCHIZOPHRENIA A WHOLE WORLD WITHOUT VISION**

Understanding blindness may shed light on the nature and etiological causes of schizophrenia. Specifically, the ways in which blind individuals perceive and mentally represent the world may hold the key to identifying vision-specific mediations of schizophrenia. Some information can be perceived by only one modality; for example, vision is needed to perceive stars in the sky or hue, and audition is needed to perceive pitch. More importantly, amodal world representations have been proposed to have fundamental importance for the development and expression of schizophrenia (Andreasen, 2007; Fletcher and Frith, 2009; Insel, 2010). Interestingly, since the famous example of William Molyneux's letter to John Locke in 1688, some people have questioned whether or not blindness affects amodal world representations:

*"Suppose a man born blind, and now adult, and then taught by his touch to distinguish between a cube and a sphere of the same metal, and the same bigness, so as to tell, when he felt one and the other, which is the cube, which is the sphere. Suppose then, the cube and the sphere placed on a table, and the blind man to be made to see. Query, whether by sight, before he touched them, he could distinguish, and tell, which is the globe, which is the cube?"* (Degenaar, 1996)

Almost two and a half centuries later, Von Senden (1932) studied patients in whom blindness was surgically cured. These patients could localize but not discriminate between a cube and a sphere immediately after the operation. However, because other researchers have refuted Von Senden's results (Gregory and Wallace, 1963; Morgan, 1977; Hollins, 1989; Cattaneo and Vecchi, 2011) and rapid cross-modal transfer indicates that the underlying mental representations may be rather similar in nature (Held et al., 2011), an ultimate conclusion to the question is still missing.

According to the International Classification of Diseases (ICD-10) (WHO, 2007), visual impairment is defined as 6/18–3/60 of the visual acuity of an unimpaired sighted individual. Being blind entails less than 3/60 of a sighted person's visual acuity or a central

<sup>2</sup>Substantial overlap exists between the different approaches (Simon et al., 2006).

visual field of less than 10◦. Whereas 161–259 million individuals suffer from visual impairment worldwide, 37–42 million individuals of the world's population fulfill ICD-10's criteria for blindness (Resnikoff et al., 2004; Dandona and Dandona, 2006). About 1.4 million of them are younger than 15 years; 30 million are older than 50. In contrast to individuals with schizophrenia, only 12% of all blind individuals live in developed countries. The most common causes of blindness are cataracts, glaucoma, and age-related macular degeneration (Resnikoff et al., 2004).

Schizophrenia affects approximately 70–80 million individuals worldwide, independent of cultural background, ethnicity, or social status (Andreasen and Black, 2006). Men and women are similarly affected (Markowitch, 1997). Given the prevalence estimates of both diseases, at least 0.00605% of the world's population, or approximately 450,000 individuals should suffer from both blindness *and* schizophrenia. Whereas it is important to keep in mind that the absence of proof (of individuals suffering from both conditions simultaneously) is not proof of absence, the protection mechanism appears specific to blindness (e.g., there are congenitally deaf individuals who become psychotic) and to schizophrenia (e.g., individuals are not protected against other psychiatric diseases) (Silverstein et al., 2013). To account for the lack of individuals who meet the diagnostic criteria for both congenital blindness and schizophrenia, Sanders et al. (2002, 2003) have suggested that dynamic adaptations of NMDA (N-Methyl-D-Aspartic) receptor channels in the visual cortex may account for cognitive functioning that is insusceptible to psychosis in the blind. However, this point of view may be incomplete because functional reorganization in visually deprived individuals also occurs in brain areas other than the visual information processing sites of sighted individuals (Weeks et al., 2000; Burton et al., 2006; Silverstein and Keane, 2011b). Reorganization, thus, depends on brain maturation (Kujala et al., 2000) and the functional interaction of subsystems in the *developing* brain (Striem-Amit et al., 2012a).

Silverstein and colleagues have a different point of view. According to these authors, cognitive coordination is impaired in patients with schizophrenia due to impaired NMDA ion receptor flow (Phillips and Silverstein, 2003; Silverstein and Keane, 2011a). Because occipital (perceptual) and non-occipital (cognitive) activity is required for cognitive organization to occur, early AND late visual processing deficits are associated with the progression of the disease and altered cognitive performance. Psychotic symptoms are, therefore, closely related to disturbed visual information acquisition and processing, which may, in turn, result in brain networks and functionality that are susceptible to psychosis. Stated differently, experiencing the world without vision may be qualitatively similar to but functionally different from experiencing the world as a sighted individual. We consequently consider the following questions in this first part of the review: (i) What evidence is there that blind individuals cognitively represent and experience the world similarly to sighted individuals? (ii) What changes in neurofunctional organization and multisensory integration are necessary for blind individuals to experience the world similarly to sighted individuals? (iii) And most decisively for the present review, how do these changes contribute to PaSZ?

# **"BLIND" PERCEPTION AND COGNITION**

On the one hand, there is no doubt that non-visual perceptual advantages are present in blind individuals (Cattaneo and Vecchi, 2011). Congenitally blind individuals have shown superior performance compared to sighted controls, for example, in auditory sound localization and speech discrimination tasks (Muchnik et al., 1991; Lessard et al., 1998; Roder et al., 1999; Kujala et al., 2000) as well as in haptic two-point discrimination tasks (Roder and Neville, 2003). Interestingly, cognitive and behavioral development can be delayed to up to 2 years in congenitally blind children compared to sighted ones (Warren, 1994). This may be due to the fact that visual information is memorized, discriminated, and explored more quickly than auditory or haptic information. In fact, auditory and haptic *memory* is limited due to *serial* information channeling; auditory *discrimination* and tactile *exploration* is limited due to *simultaneous* information channeling. This means that specific attention-orienting and stimulus-awareness mechanisms are necessary for visual but not other perceptual domains (Posner et al., 1976; Phillips and Silverstein, 2003). These mechanisms, in turn, have been described as relevant for the development of schizophrenia (Butler et al., 2005, 2007; Johnson et al., 2005; Kim et al., 2005; Haenschel et al., 2007).

On the other hand, amodal mental representations, e.g., how the idea of a "tree" is represented in the brain, are independent of their perceptual source (Avraamides et al., 2004; Barsalou, 2008). In other words, no matter which modality this information is perceived through (e.g., audition, touch, vision), the ultimate mental representations are similar. And this is true for real-world mental representations of blind and sighted individuals as Cattaneo and Vecchi (2011) put it so eloquently:

#### *"Our brain, indeed, doesn't need our eyes to 'see'..."* (p. 3).

In line with this argument, Pascual-Leone and Hamilton (2001) proposed that specialized neural functionality is formed as a consequence of receptive field input to brain structures. Hence, there may be no *a priori* functional segregation of the human brain. Instead, the existence of small receptive perceptual fields, which are required for quick categorization, and large receptive fields, which are necessary to coordinate thought processes, result in the specialization of cortical regions for processing "visual" or "auditory" information. The authors refer to the *metamodal* brain as a "mixture of expert architecture" (p. 15) (Pascual-Leone and Hamilton, 2001) and argue that sighted and congenitally blind individuals have an equivalence of amodal representations.

A plethora of investigations have, in fact, provided evidence of this equivalence. First, higher-order cognitive abilities, and spatial abilities in particular, have shown no differences between congenitally blind and sighted individuals, for example, in mental scanning and rotation (Craig, 1973; Marmor and Zaback, 1976; Carpenter and Eisenberg, 1978; Kerr, 1983; Zimler and Keenan, 1983; Vecchi et al., 2004) or allocentric referencing tasks (Tinti et al., 2006). Investigations of the mental number line (Moyer and Landauer, 1967; Dehaene et al., 1999; Krueger et al., 2008, 2011; Landgraf et al., 2010b) have shown similar spatial representations in congenitally blind and healthy sighted individuals (Castronovo and Seron, 2007). Hence, in congenitally blind individuals, equivalent amodal (spatial) representations can apparently be formed

from perceptual cues other than vision. Interestingly, in this same experiment, congenitally blind individuals were able to classify the numbers one and two more quickly than sighted individuals. The authors attributed this effect to the serial processing of tactile and auditory information. Blind individuals may be more accustomed to the idea of distinguishing perceptual phenomena with regard to number counting, especially one and two (e.g., counting steps).

Second, whereas a lack of visual input is detrimental to some tasks, the superior performance of sighted individuals has been shown to disappear when visual task demands are increased. For example, in a 3-D working memory task, congenitally blind individuals performed as well as sighted individuals when information input exceeded visual information processing capacities (Cornoldi et al., 1991). Furthermore, in matching standard figures and line drawings, congenitally blind and late blind individuals did not differ in their reaction times and error rates (Heller and Kennedy, 1990). The authors interpreted their results as indicating that visual imagery and visual experience appear unnecessary for tactile perspective taking. Interestingly, Vecchi et al. (2004) reported that the memory of locations decreased with an increasing number of visuo-spatial (VS) representations that had to be held in working memory. This shows that higher task demands, such as speed, interactive images, and movement, may account for performance decreases in blind individuals. Remarkably, the performance of schizophrenia patients also declines when task demands are increased, indicating a coping mechanism for imprecise visual information acquisition (Landgraf et al., 2011a,b).

Third, it has been proposed that visual input deprivation may lead not only to perceptual enhancement in congenitally blind individuals (Rauschecker, 1995; Lessard et al., 1998; Roder et al., 1999) but also to improved attentional capacities (Cattaneo et al., 2008). Better performance in higher-order cognitive abilities has, in fact, been observed in various memory span (Tillman and Bashaw, 1968; Smits and Mommers, 1976; Pozar, 1982; Hull and Mason, 1995; Roder et al., 2001; Amedi et al., 2003; Roder and Rosler, 2003; Raz et al., 2007) and auditory attention tasks (Roder et al., 1996, 1999, 2001; Roder and Rosler, 2003). Van Velzen et al. (2006) investigated early and late attention indicators in congenitally blind and sighted individuals with electroencephalographic (EEG) recordings. The authors found that early but not late attention modulation could be elicited in blind individuals, indicating that late cognitive coordination mechanisms may be important for the development of psychosis. Late attention indicators also depended heavily on having the individual focus on task-relevant external spatial reference frames, an ability that has been compromised in patients with schizophrenia (Dreben et al., 1995; Parnas et al., 2001; Johnson et al., 2005; Cavezian et al., 2007; Coleman et al., 2009; Landgraf et al., 2011b). In fact, patients with chronic schizophrenia have been found to display a deficit in the disengagement and reorientation of attention (Posner et al., 1988; Daban et al., 2004; Gouzoulis-Mayfrank et al., 2007; Kebir et al., 2008, 2010) even without medication (Amado et al., 2009). Hence, the interaction between bottom-up and top-down processes may provide insight into the immunity against psychosis that blind individuals appear to have (see the Protective Mechanism "Cognition" in **Table 1**).

#### **NEUROFUNCTIONAL REORGANIZATION AND COMPENSATION**

The ability to behaviorally compensate for visual deprivation is another indication that there are equivalent amodal representations in sighted and congenitally blind individuals (Pascual-Leone and Hamilton, 2001; Knauff and May, 2006; Cattaneo et al., 2008). However, because up to 35% of neocortical functioning in sighted humans is devoted to visual information processing (Gilbert and Walsh, 2004), blind and sighted individuals may differ considerably in their neurofunctional processing. In sighted individuals, cortical activity has traditionally been ascribed to region-specific functionality (Kanwisher, 2010), such as visual information processing being ascribed to occipital activity. However, blindfolding sighted individuals for a 5-days period has been found to lead to behaviorally relevant neurofunctional changes (Kauffman et al., 2002) that mimick the supranormal auditory performance of blind individuals (Rauschecker, 1995; Lessard et al., 1998; Roder et al., 1999). This not only implies that neurofunctional reorganization occurs in the adult human brain (Kujala et al., 2000), but also indicates that cortical functionality may be determined by information processing necessities (Cohen et al., 1997) and innate pathways (Striem-Amit et al., 2012c) rather than brain regions.

Blind individuals have been found to employ compensatory neurofunctional strategies to overcome visual information deprivation. Using positron emission tomography (PET), Sadato et al. (1996, 1998) were the first to find a relation between activity in the occipital cortex and non-visual perception and cognition. Specifically, the authors found that during braille reading and nonbraille haptic tasks, the primary, and medial occipital lobes were activated in individuals who became blind both early and late in life. In sighted individuals, on the other hand, these haptic tasks were associated with activity in non-occipital regions, such as the bilateral inferior parietal lobes, as well as the left primary sensorimotor area, insula, and prefrontal regions. In subsequent studies, congenitally blind individuals have shown recruitment of the occipital cortex during higher linguistic and (Cohen et al., 1997; Burton et al., 2003, 2006; Gilbert and Walsh, 2004; Raz et al., 2005; Amedi et al., 2007) auditory motion processing (Poirier et al., 2006), as well as during the localization of auditory signals (Weeks et al., 2000), tactile processing (Sadato et al., 2002), and tongue stimulation (Kupers et al., 2006). Finally, even sighted individuals have shown non-visual information processing in visual areas. In a sophisticated transcranial magnet stimulation (TMS) paradigm, Lewald et al. (2004) demonstrated that the temporary disruption of occipital activity can deteriorate auditory localization in sighted individuals.

These results led us to postulate the following two assumptions: first, neurofunctional plasticity in congenitally blind individuals includes reorganization in non-visual cortical areas, which has been confirmed, for example, in lingual and posterior fusiform gyri (Smith and Gasser, 2005; Striem-Amit et al., 2012a). Therefore, the reorganization of non-visual cortical areas in patients with schizophrenia may be important for the development of the disorder. Second, non-visual information processing in the occipital lobe of blind individuals resembles visual information processing in the occipital lobes of sighted individuals (Burton


et al., 2010). Thus, whereas functional and most likely also structural processing mechanisms are similar between blind and sighted individuals, the contents of the information (visual vs. non-visual) are different.

This has direct consequences for the development of schizophrenia (see also the Protective Mechanism "Neurofunctioning" in **Table 1**). If brain structures are not utilized according to their functional and structural specificity, this could result in psychosis or at least in subclinical symptoms. For example, as mentioned above, depriving sighted individuals of their vision for 5 days can lead to neurofunctional and cognitive changes (Kauffman et al., 2002). This means that not only eyerelated dysfunction but also functional changes in the primary and secondary occipital cortex can lead to hallucinations (Kazui et al., 2009; Schadlu et al., 2009). Further, individuals suffering from the so-called Charles Bonnet Syndrom (CBS) have reported visual hallucinations as a consequence of visual deterioration (ffytche and Howard, 1999; ffytche, 2009). Hallucinations due to the CBS can vary widely, comprising abstract geometric patterns, mosaic vision (tessellopsia), increased color vision (hyperchomatopsia), and miniturizations and magnifications of objects (micropsia and macropsia), and can occur for several minutes (70%), seconds

(18%), or hours (12%) (Hughes, 2013). There are two main theories about how CBS develops. The release theory claims that a mixture of impaired and unimpaired neuronal signals from the visual cortex lead to hallucinatory interpretations in the higher-order association cortices. This is similar to our argument that schizophrenia patients can profit from giving *more* weight to visual information in multisensory integration tasks. Hence vision training may be supplementarily useful to cognitive remediation programs. On the other hand, the deprivation theory of CBS argues that reduced sensory input may result in spontaneous image production in the visual association cortex, thus leading to visual hallucinations. This implies, as indicated by the PaSZ model, that perceptual deprivation *per se* is not sufficient for psychosis protection to occur. Instead, functional and possibly structural cortical reorganization need to be taken into consideration to avoid psychosis. In fact, there is strong evidence that brain regions are not adequately utilized in schizophrenia because patients use more sequential information processing strategies (Fatemi and Folsom, 2009; Landgraf et al., 2011a,c) possibly due to impaired higher-order cognitive deficits, i.e., in single- or multisensory integration (Park and Holzman, 1992; Park et al., 2002; Tek et al., 2002; Landgraf et al., 2008; Fuller et al., 2009).

#### **MULTISENSORY INTEGRATION**

The integration of information from multiple modalities has been found to improve performance compared to using one single perceptual channel alone (Calvert et al., 2003). In fact, multisensory integration allows the brain to generate a coherent amodal view of the self and the world (Smith and Gasser, 2005). Supramodal information, such as spatial and temporal information, is coded by all human perceptual systems and allows the comparison of multisensory integration between, for example, blind and sighted individuals. For the scope of this review, we restrict our considerations of multisensory integration to temporal and spatial processes that involve haptic, auditory, and, for sighted individuals, visual capacities. For a review of olfactory and gustatory perception regarding visual and auditory multisensory integration, please refer to Walla (2008) or Zampini and Spence (2012).

Efficient multisensory integration remaps information into amodal representations. In the absence of visual input, this remapping may develop differently (see Protective Mechanism "Multisensory Integration" in **Table 1**) (Hotting et al., 2004; Roder et al., 2004; Wallace et al., 2004). For example, in congenitally blind individuals, judging the temporal order of tactile stimuli with the right or left hand is not affected by whether the hands are in normal or crossed-over positions. By contrast, sighted individuals show longer reaction times when their hands are crossed, implying interference between visual and tactile external frames of reference (Roder et al., 2004, 2007; Collignon et al., 2007). Hotting et al. (2004) reported that the influence of task-irrelevant auditory tones for tactile discrimination is stronger in sighted than in congenitally blind individuals. Interestingly, humans deprived of vision between 5 and 24 months after birth due to retinal cataracts showed less auditory-visual interference and integration later in life than normally developing sighted individuals (Putzar et al., 2007). In line with these observations, the brain regions responsible for visual imagery (e.g., the fusiform face area) may retain their functional specificity even decades after the onset of blindness in late blind individuals (Goyal et al., 2006). In fact, there may be a specific critical period during which neurofunctional plasticity to sensory loss is maximal (Sathian, 2005). Until now, the duration of that time window has been unclear, but it has been proposed to be between 10 and 14 years for visual information processing (Cohen et al., 1999; Ofan and Zohary, 2007), implying that this may be a critical period for cortical changes regarding the protection and development of schizophrenia as well. Although changes in cortical functioning might not occur until several years after sensory deprivation (Cattaneo et al., 2008), the overall evidence implies that ontogenetic development must occur to establish the multisensory integration interferences observed in sighted individuals. Therefore, it can be hypothesized that this development may be impaired in patients with schizophrenia, thus leading to impaired or unusual performances in multisensory integration tasks.

In sighted individuals, auditory information processing is heavily influenced by other modalities and prior knowledge. Alain and Arnott (2000) distinguished between auditory attention (allocation of attentional resources to perceptual objects), auditory object discrimination (perception of sound attributes across a certain time period), and auditory event perception (perception of sound at a particular time, in a certain place, and having specific characteristics). The authors argue that auditory information quality impacts cognitive performance in all of these stages. For example, degradation of auditory information influences comprehension and memory differently in younger and older listeners (Pichora-Fuller and Singh, 2006). Furthermore, the characteristics of speakers are processed in parallel with semantic information. Whereas female voices are associated with more extraversion and openness, male voices are associated with higher emotional stability and greater agreeableness (Imhof, 2010). This is in line with other studies that have shown that identical semantic information is interpreted differently depending on whether the person is perceived as male or female (Addington, 1968; Knapp and Hall, 2002). Hence, auditory processing is inextricably linked to non-auditory information, thus implying that the working memory demands of listeners include the stream of the sound, semantic cues, and perceptual voice cues. Patients with schizophrenia have been shown to struggle when asked to integrate these perceptual cues with other cues (Hardoy et al., 2004; Leitman et al., 2005; Butler et al., 2008a, 2009). Interestingly, integrating non-auditory and auditory information directly from perceptual cues can be tested in multisensory integration tasks. For example, clustering task-irrelevant stimuli should not affect the performance of blind individuals if multisensory integration occurs in a manner that is similar to what has been observed in healthy individuals (Alain and Arnott, 2000).

Another auditory phenomenon is the so-called "right ear advantage," which refers to the observation that auditory information is processed with greater ease (more quickly, less erroneously) by the right ear compared to the left. In fact, it has been observed that left frontal lobe lesions diminish the advantage of the right ear (Hugdahl et al., 2003), indicating a functional preference for auditory information processing in the left frontal lobe. Congenitally blind individuals have shown a less pronounced right ear advantage, thus outperforming sighted individuals in, for example, dichotic listening tasks (Hugdahl et al., 2004; Castronovo and Seron, 2007). It may be interesting to investigate whether patients with schizophrenia would perform like sighted or blind individuals in this paradigm. If schizophrenia patients' performance more closely resembles the performance of sighted individuals, this would be an indication of their local processing preference. If, however, the patients' performance was more like that of blind individuals, this may be attributable to a lower reliance on multisensory integration.

Congenitally blind individuals have been found to rely less on multisensory integration, thus leading to strategic compensatory behavior. On the one hand, they recognize haptic stimuli more quickly than sighted individuals, due to, for example, more efficient sensory information processing during the first 100 ms after stimulus onset (Roder et al., 1996). On the other hand, however, if blind individuals were to employ an adequate multisensory integration strategy, they might perform equally as well as sighted individuals. In a haptic mental imagery memory task, Cornoldi et al. (2009) showed that congenitally blind individuals performed worse than healthy controls only when they adopted a spatial strategy but not when they adopted a verbal strategy. Mental representations of size are more erroneously influenced by multisensory information in sighted than in blind individuals (Bartley et al., 1955; Bolles and Bailey, 1956). Smith et al. (2005) showed that size estimations of real objects are less error-prone in congenitally blind individuals compared to sighted controls. Instead, blind individuals rely more on a haptic memory strategy to perform these tasks. This implies that multisensory integration involving touch depends on task characteristics, and a lack of visual input can be compensated for by adequate strategies (Postma et al., 2007). In fact, perceptual and long-term memory may be used to encounter multisensory integration deficits in blind individuals. Possibly patients with schizophrenia may profit from this strategy as well.

It is noteworthy that multimodal integration has been found to change with age. Warren and Pick (1970) found that auditory localization is strongly influenced by visual perceptual input in sighted adults but not in sighted children. Furthermore, auditoryproprioceptive integrations may be influenced by the increasing importance of vision with increasing age in sighted individuals. Pitch direction changes are better discriminated by blind individuals than by sighted controls even when the stimuli are presented 10 times more quickly but only when individuals became blind at an early age (Gougoux et al., 2004). Age at the onset of blindness has been shown to regulate the use of visual reference frames in haptic and auditory perception (Roder et al., 2004, 2007). However, these effects may also be attributable to the fact that blind individuals show better skills in processing unimodal perceptual inputs (Hotting and Roder, 2004; Hotting et al., 2004). More importantly, for congenitally blind individuals, if no change was observed across time in these functions, the weighing of intermodal information could be assumed to follow different developmental trajectories in blind and sighted individuals. Regarding schizophrenia, patients may start out with already impaired preconditions in their multisensory integration abilities. Hence, visual information in patients with schizophrenia may not be given enough multisensory integration weight. Instead, in their attempt to compensate perceptually, schizophrenia patients would integrate more perceptual information from other modalities than sighted controls, possibly leading to their preference for local information processing (Landgraf et al., 2011b).

# **THE VISION PERSPECTIVE ON SCHIZOPHRENIA THE VISION-SPACE-BODY-AND-COGNITIVE-IDENTITY MODEL OF SCHIZOPHRENIA**

Thus far, we have argued that the inability to perceive and process visual information prevents congenitally blind individuals from becoming psychotic. Specifically, we have shown that PaSZ in congenitally blind individuals may be associated with the acquisition of information and reorganizations of processing at cognitive, neurofunctional, and multisensory levels of integration. It appears that more than "zero" vision AND impaired visual processing must occur simultaneously in order to render the human brain susceptible to psychosis. This determines the visual impairment part of the PaSZ model (left side of the curve in **Figure 1**).

In our laboratory, we have developed a model that shows the processing levels at which the deteriorations in visual capacity contribute to the development of psychosis. The ViSBI, short for **Vi**sion-**S**pace-**B**ody-and-cognitive-**I**dentity, model of schizophrenia (Landgraf et al., 2012) shows that deficits in the acquisition and processing of visuo-cognitive information increase along with the progression of the disorder and are, indeed, associated with higher-order cognitive dysfunctions and expression of symptoms. We now turn our attention to "Biomoehrchen schmecken gut" the corresponding visual deterioration part of the PaSZ model (right side of the curve in **Figure 1**).

# **VISUAL INFORMATION ACQUISITION**

Visual information acquisition accounts for more than 80% of the perceptual input in humans and is therefore considered to provide the foundation for amodal world representations (Lüer et al., 1988; Becker, 1991; Leigh and Zee, 1999). Deficits in visual information acquisition have been described extensively in patients with schizophrenia (see "Vision" complex in **Figure 2**). Deficits deteriorate along the continuum from the prodrome to the full-blown syndrome, i.e., with the progression of the disease. One of the longest and best-studied types of deteriorations is that of oculomotor deficits. They not only jeopardize the availability of visual information but also impair the spatial and temporal accuracy of individuals. Hence, they are crucial for the formation of concepts of the self and the surrounding world.

Oculomotor deficits are least pronounced in prodromal individuals. Only a few studies have demonstrated that prodromal individuals show an increased rate of errors in the antisaccade paradigm (Nieman et al., 2007). In first-episode schizophrenia patients, deficits have been described with regard to antisaccade errors, reaction time, and accuracy (Broerse et al., 2001; Ettinger et al., 2004; Hutton et al., 2004). Furthermore, deficits such as a higher error rate in memory-guided saccades and lower acuity in predictive saccades have been reported (Krebs et al., 2001; Hutton et al., 2004; Keedy et al., 2006). In chronic patients, all these deficits have been observed (e.g., Karoumi et al., 2001; Brownstein et al., 2003; Reuter et al., 2006; Radant et al., 2007; Amado et al., 2008; Landgraf et al., 2008). In addition, chronic patients have shown lower acuity in memory-guided saccades (Crawford et al., 1995a; Park et al., 1995; McDowell and Clementz, 1996; Radant et al., 1997; Karoumi et al., 1998; McDowell et al., 2001). Although there is some debate about the usefulness of oculomotor deficits in schizophrenia research (e.g., Calkins and Iacono, 2000; Brownstein et al., 2003; Calkins et al., 2004, 2008; Levy et al., 2004, 2008; Boudet et al., 2005; Heydebrand, 2006), deficits have been identified as schizophrenia spectrum markers (Amador et al., 1991; Sweeney et al., 1994; Faraone et al., 1995; McDowell and Clementz, 1997; Rosenberg et al., 1997; Avila et al., 2002; Kathmann et al., 2003), independent of clinical state (Calkins et al., 2003; Kallimani et al., 2009), or medication (Crawford et al., 1995b; Muller et al., 1999). Furthermore, oculomotor aberrations have been found to fulfill some endophenotypic characteristics (Gottesman and Gould, 2003; Calkins et al., 2008).

Another interesting phenomenon regarding the acquisition of visual information is eye-movement strategies. Eye-movement strategies allow for a meaningful and useful succession of visual information input with regard to, for example, goal attainment or planning behavior (Land and Furneaux, 1997). Deficits in eye-movement strategies may lead to inefficient visual input and may thus alter an individual's perception of the world. As seen in oculomotor deficits, the severity of eye-movement-strategy

**FIGURE 2 | The Progressive Vision-Space-Body-and-Cognitive-Identity (ViSBI) model of schizophrenia.** Source: Reprinted with kind permission from Bentham Science Publishers. Note. The progressive ViSBI model stresses vision-related deteriorations from prodrome to chronic syndrome that may lead to schizophrenia. The model comprises four complexes: "Vision," "Space," "Body," and "cognitive Identity. "The "Vision," "Space," and "Body" complexes are reviewed in the sections on "The Vision Perspective on Schizophrenia" in the current paper. The "Cognitive Identity" complex (shadowed in gray) is hypothesized as a part of the model (see Landgraf et al., 2012). Complexes are sub-divided into specific research areas (e.g., in the "Vision" complex: oculomotricity and scanning) and disease progression status (prodromal, first episode, and chronic schizophrenia patients). Tasks listed in each row indicate deficient performance of the corresponding patient group. Some areas have not been investigated in schizophrenia patients ("suggestions for further research").The "Visual Input" triangle in the middle indicates that (congenital) blindness may prevent these mechanisms from taking place because visual perceptual input is required for this pattern of aberrations to occur. Abbreviations: prodromal = individuals identified at ultra-high risk for developing schizophrenia; first episode = patients with schizophrenia that have had one (identified) psychotic episode; chronic = patients with schizophrenia that have had at least three (identified) psychotic episodes; SPEM = smooth pursuit eye movements; n-back = spatial n-back task.

deficits increases with the increasing progression and severity of schizophrenia. Prodromal individuals have rarely been studied in visual scanning paradigms. Koethe et al. (2006, 2009) showed that abnormal binocular depth inversion was specific to prodromal individuals. Furthermore, first-episode schizophrenia patients have shown an abnormal clustering of fixations and shorter visual scanpaths when scanning faces, landscapes, and abstract patterns (Benson et al., 2007). The largest deteriorations have been reported for stabilized chronic patients with schizophrenia, with fewer fixations, shorter scanpaths, narrower clustering of fixations, and avoidance of predefined features in different visual scanning tasks (e.g., Gaebel et al., 1987; Gordon et al., 1992;

Kurachi et al., 1994; Phillips and David, 1997; Williams et al., 1999; Loughland et al., 2002, 2004; Minassian et al., 2005; Koethe et al., 2006).

In our own laboratory, we found that, in contrast to healthy controls, chronic schizophrenia patients do not adapt their eyemovement strategies to task demands but instead use similar strategies regardless of task difficulty (Landgraf et al., 2011b). In other words, patients employ the same visual scanning strategy no matter whether the task is easy or difficult. In another study, we even found that chronic schizophrenia patients use a less efficient information acquisition strategy more often than healthy controls (Landgraf et al., 2011a). This implies that inefficient visual information acquisition impedes patients from obtaining (task-) relevant information. Patients have probably noticed that inflexible visual scanning suits most of the demands they encounter in their daily lives. One intriguing question concerning the discussion of blindness and schizophrenia is whether patients are unable to develop other strategies from birth on or whether patients develop the use of only one strategy despite being able to develop others.

These findings imply the following. First, patients with schizophrenia attribute less weight to visual information and are not able to integrate visual information as well as healthy sighted individuals. This means that patients' deficient performances on higher-order cognitive tasks, therefore, can be accounted for by lower-level visual aberrations. Second, patients' visual channels are still "open." Hence, they are unable to obtain cognitive and neurofunctional reorganization in the same way as blind individuals. This suggests that PaSZ may be available in two directions. On the one hand, if patients learn to put *more* weight on visual information in the same way that healthy sighted individuals do, they may lower their risk for a psychotic episode. On the other hand, if patients are taught to put *less* weight on visual information and reorganize their neurofunctional information processing, they might be able to stabilize their self and world perceptions in the same way that blind individuals do (See "Vision Training: Decline vs. Improvement" for further details on vision training).

#### **VISUAL INFORMATION PROCESSING**

Aberrant visual information processing (see "Space" complex in **Figure 2**) deteriorates thought processes, contributing to cognitive dysfunctions. It has been acknowledged that cognitive deficits may be one of the best predictors of the development of psychosis (Elvevag and Goldberg, 2000; Insel, 2010; Keefe and Harvey, 2012), and this link could provide the basis for quantitative stage markers of the disease. Most of the paradigms used in schizophrenia cognition research are visual tasks. Moreover, fundamental associations between visuo-spatial (VS) abilities and oculomotor capacities (Leigh and Zee, 1999; Hutton et al., 2004; Lawrence et al., 2004; Pierrot-Deseilligny et al., 2005; Milea et al., 2007) indicate that disturbances in the VS domain may stem from oculomotor dysfunctions in patients with schizophrenia. Some of the cognitive domains deficient in patients with schizophrenia include but are not limited to VS memory (Goldman-Rakic, 1994; Green et al., 2000; Piskulic et al., 2007; Landgraf et al., 2011c), VS attention (Posner et al., 1988; Danckert et al., 2004; Granholm and Verney, 2004; Gouzoulis-Mayfrank et al., 2007), and VS executive functions (Laws, 1999; Eisenberg and Berman, 2010; Landgraf et al., 2011a,b,c). An impressive number of studies have actually shown that the development of schizophrenia may be associated with these cognitive deficits and that these cognitive deficits could be stage-specific (Heaton et al., 1994; Niendam et al., 2006; Fusar-Poli et al., 2007; Lysaker et al., 2007; Langdon and Ward, 2008; Picard et al., 2009; Barlati et al., 2012).

Visuo-spatial cognitive deficits in prodromal individuals are controversial. Some studies have shown a greater number of errors and inferior performance in prodromal individuals compared to healthy controls on spatial delayed-response tasks (Wood et al., 2003; Bartok et al., 2005; Smith et al., 2006; Kimhy et al., 2007;

Nieman et al., 2007). Other investigations have not confirmed this observation (Brewer et al., 2005; Lencz et al., 2006; Niendam et al., 2006; Pukrop et al., 2007). The reasons for this controversy have not yet been resolved. Possibly, heterogeneous results are due to imprecise diagnoses of the schizophrenia prodrome (high rate of false negatives), demographic differences between participating groups, or task simplicity (Wood et al., 2003; Conklin et al., 2005; Longevialle-Henin et al., 2005; Fusar-Poli et al., 2007; Pukrop and Klosterkotter, 2010). Nevertheless, prodromal individuals have been consistently reported to display deficient performances on the Trail Making Tests (TMT) A and B and the Wechsler Memory Scale (WMS-R) visual reproduction task parts I and II (Hawkins et al., 2004; Brewer et al., 2005).

First-episode patients have shown deficient performances on a number of VS cognitive tasks. For example, they have shown aberrant memory on VS delayed-response tasks (Goldman-Rakic, 1994; Park et al., 1995; Rybakowski and Borkowska, 2002; Simon et al., 2007), Gestalt perception (Parnas et al., 2001), and the TMT (Rybakowski and Borkowska, 2002; Simon et al., 2007). This implies moderate deficits in these patients.

Chronic patients have shown the strongest aberrations on VS tasks. Their deficits include abnormal performance on delayedresponse tasks (Park and Holzman, 1992; Goldman-Rakic, 1994; Glahn et al., 2003; Park et al., 2003; Saperstein et al., 2006; Genderson et al., 2007), figure search (Longevialle-Henin et al., 2005), mental rotation (de Vignemont et al., 2006; Halari et al., 2006), Gestalt perception (O'Donnell et al., 1996; Parnas et al., 2001; Cavezian et al., 2007; Kimhy et al., 2007), spatial span (Cannon et al., 2000; Perry et al., 2001; Manoach et al., 2005; Genderson et al., 2007; Thoma et al., 2007), 3-D real-world navigation Daniel et al. (2007), and referencing (Landgraf et al., 2010a; Mazhari et al., 2010). According to a meta-analysis, there is an overall effect size of −1.00 regarding VS working memory deficits in chronic schizophrenia patients (Piskulic et al., 2007), and these deficits are independent of gender differences (Albus et al., 1997; Reichenberg et al., 2002; Voglmaier et al., 2005; Halari et al., 2006; Wolitzky et al., 2006). Hence, visual information processing appears to be deteriorated in schizophrenia, increases with the progression of the disease, and is related to basic visual acquisition. It may be interesting to compare patients' performance to the performance of the blind in non-visual versions of these paradigms. Even though they have never been able to integrate visual information, blind individuals should outperform patients with schizophrenia on spatial tasks. This would indicate that schizophrenia patients' severely altered visual information processing is strongly associated with their deficits in visual information acquisition. We hypothesize that patients with schizophrenia could especially profit from the neurofunctional and cognitive reorganization observed in blind individuals. The most important question, however, for the consideration of PaSZ is whether or not VS deficits are associated with higher-order cognitive dysfunctions and the symptoms of schizophrenia.

#### **FROM VISUAL DETERIORATION TO SYMPTOMS**

Basic visual acquisition and processing dysfunctions in patients with schizophrenia suggest a relation to phenotypic symptomatology. In other words, someone who is not able to obtain accurate and precise (visual) information cannot process this information adequately and thus may suffer from deficient amodal representations of the world. Scientifically, visual scanning deficits and oculomotor deteriorations are related to higher-order cognitive dysfunctions (e.g., theory of mind, perspective taking) and social cognition (Adolphs, 2003; Amodio and Frith, 2006; Kluwe-Schiavon et al., 2013), as well as to functional outcomes in schizophrenia (Green et al., 2000; Benson et al., 2007). In fact, the temporo-parietal junction (TPJ), involved in perspective taking and theory of mind, and the insular cortex, involved in body-related multisensory integrations (Arzy et al., 2006; Cavanna and Trimble, 2006; Danckert and Ferber, 2006; Schwabe and Blanke, 2007; Mitchell, 2008), have been associated with psychotic states (Penfield, 1955; Blanke et al., 2005; Vercammen et al., 2010), and impaired whole body and body-part processing in patients with schizophrenia (Tan et al., 2006; Butler et al., 2008b; Suchan, 2008). It has been hypothesized that the failure to predict the sensory consequences of motor commands (Frith et al., 1992, 2000a,b; Friston and Frith, 1995; Blakemore et al., 2002; Frith, 2005) and the improper planning of motor sequences (Delevoye-Turrell et al., 2003; Coello and Delevoye-Turrell, 2007; Voss et al., 2010; Waters and Badcock, 2010) are both essential to schizophrenia symptoms. Furthermore, motor-related deficits in chronic and first-episode patients with schizophrenia include an altered pattern of self-recognition, for example, in the rubber hand illusion (Franck et al., 2001; Versmissen et al., 2007), the inability to distinguish between the self and others (Schwabe and Blanke, 2007; Ebisch et al., 2012), and the identification of the source of self- or externally generated movements (de Vignemont et al., 2006). This implies, on the one hand, that symptom-related deficits in schizophrenia encompass a cognitive component (Bowins, 2011). On the other hand, these results point toward the critical role of multisensory integration deficits in patients (Friston and Frith, 1995; Fourneret et al., 2002) (see "Body" complex in **Figure 2**).

Because multisensoriality and amodality, as well as self and world representations have already been discussed for blind individuals, there are two things to note. First, the neural sites for multisensory integration may differ between patients with schizophrenia and blind individuals. This would mean that dissimilar neurofunctional processes are conducted. These processes and sites may represent good candidates for early detection markers and possible interventions for schizophrenia. Second, blind individuals and patients with schizophrenia have something in common regarding multisensory integration: they assign less weight to visual information. Blind individuals do so due to the absence of visual input; patients with schizophrenia may do so because they have deficits in visual information acquisition and processing. However, the underlying neurofunctional processes differ and may be an indication of the protective mechanisms of blindness. Interestingly, there is a strong association between oculomotor function and multisensory integration (for reviews, see, e.g., Previc, 1998; Milner and Goodale, 2006). Dysfunctions of visual information acquisition and processing have actually been found to be correlated with multisensory integration deficits (Park and Holzman, 1993; Ross et al., 1998; Jansen et al., 2002; Nieman et al., 2007; Picard et al., 2009) and symptomatology (Gaebel et al., 1987; Lencz et al.,

2003; Semerari et al., 2003; Varga et al., 2007) in schizophrenia patients. Moreover, whereas cross-modal influences are dominated by visual information in patients (de Gelder et al., 2005), there is strong evidence that multisensory integration is compromised in chronic patients (Vrtunski et al., 1993; Marvel et al., 2004; Picard et al., 2008; Van den Stock et al., 2011; Castagna et al., 2012). Multisensory facilitation that is established potentially to compensate for deficient unisensory processing in patients with schizophrenia (Javitt, 2009; Williams et al., 2010; Stone et al., 2011) may need to be altered with regard to its reliance on visual information processing (de Gelder et al., 2005). Subsequently, patients might benefit from similar cognitive and perceptual protection against the disease as observed in blind individuals. The degree to which the neglect of visual information integration overlaps between blind and schizophrenic patients in multisensory integration may be an indicator of the severity of schizophrenia. We hypothesize that a greater degree of overlap would indicate a less severe schizophrenia outcome.

# **THE THERAPEUTIC PERSPECTIVE ON SCHIZOPHRENIA CONTINUOUS DIAGNOSTIC CRITERIA**

In the previous section, we demonstrated that deteriorations in the acquisition and processing of visual information increase the risk for developing schizophrenia. We established a critical relation between lower- as well as higher-order visual deteriorations and symptom expression in schizophrenia. Whereas fundamental oculomotor and strategic eye-movement deficits may impact the *acquisition* of visual information, information *processing* deficits point toward VS cognitive aberrations. Both visual information acquisition and processing dysfunctions have been found to increase in severity with the progression of the disorder and are correlated with symptomatic expressions of the disease.

The PaSZ model postulates a continuous relation between visual capacity and the risk of developing schizophrenia. This means that looking at disturbances in vision from a disease progression perspective obviates the need for symptom-based prodromal criteria. Instead, because cognitive deficits can be depicted on much more fine-grained continua than psychotic symptoms (Saperstein et al., 2006; Uhlhaas and Mishara, 2007), we argue that a graded stage model of schizophrenia based on visual functioning will contribute to more reliable diagnostic and especially prodromal criteria for schizophrenia. More importantly, however, are the therapeutic implications of the model, which we will discuss in this final section of the review.

# **VISION TRAINING: DECLINE VS. IMPROVEMENT**

So far, we have presented evidence for why the relative risk of developing schizophrenia is allegedly zero for congenitally blind individuals and for individuals with supernormal visual capacities. According to the PaSZ model, the risk of developing schizophrenia increases from both ends of the visual capacity continuum (congenital blindness and "supernormal" vision) toward a "peak risk" (**Figure 1**). Thus, depending on the person's initial visual capacity, a decrease in visual impairment or an increase in visual deterioration may similarly *elevate* the risk of developing schizophrenia. Consequently, therapeutic efforts may be differentially effective: a *decline* in a person's visual capacity (increase of visual impairment) may be more beneficial for an individual with visual impairment who never had "normal" visual skills in the first place. By contrast, an *improvement* in a person's visual capacity (decrease of visual deterioration) may be more beneficial for an individual with visual deterioration, that is, who had at some point developed "normal" vision. However, the difference between those two therapeutic approaches may not be clear cut as there is, until now, no clear agreement about (1) how much aberrations of visual functioning corresponds to the highest risk for schizophrenia and (2) whether or not an increase in visual impairment (e.g., via sensory substitution) actually corresponds to a decline in visual functioning.

Nevertheless, vision improvement training is indispensable for PaSZ when the affected individual suffers from visual deterioration. Regarding information *acquisition* (see "Acquisition of visual information" in **Table 2**), patients should be visually trained to obtain the necessary task-relevant visual information and should be given the tools needed to interpret this information. This means that it may be beneficial to train individuals to utilize taskspecific and successful eye-movement strategies. Patients should be taught to direct their attention (eye movements) and cognitive resources (pupil dilation, neurofunctionality) to task-relevant visual information: to look where the information is, to avoid information overload, and to access the "big picture" instead of focusing on attention-captivating details (Johnson et al., 2005; Longevialle-Henin et al., 2005; Cavezian et al., 2007; Coleman et al., 2009; Landgraf et al., 2011b). In addition, training should establish a link between how much weight is given to visual and non-visual information in multisensory integration tasks. Patients should learn when it is advantageous to rely on visual information (fast, parallel processing) and when it is advantageous to rely on non-visual information (slow, sequential processing). This may help them to build more reliable amodal representations of the world, to orient themselves better, and to avoid confusion.

Interestingly, visual deterioration has been associated with the risk for criminal behavior (Bachara and Zaba, 1978; Zinkus and Gottlieb, 1978; Lane, 1980; McKay and Brumback, 1980; Broder et al., 1981; Clack, 1990), one of the strongest indicators of a severe course (Steinert, 1998; Nedopil, 2007; Hutton et al., 2012), outcome (Leygraf, 1988; Haller et al., 2001; Soyka and Morhart-Klute, 2002; Soyka et al., 2002; Fazel et al., 2009; Nitschke et al., 2011; Kooyman et al., 2012), and relapse (Soyka et al., 2004; Witt et al., 2013) in schizophrenia. Specifically, it was argued that in very young children and juveniles, perceptual deficits lead to a higher rate of learning disabilities, specifically reading problems. These problems, in turn, may exclude children from further participating in social interactions and, subsequently, could lead to frustration and feelings of exclusion. It was concluded that visual deterioration in children facilitates the development of delinquent and criminal behavior (Slaton and Jorgensen, 1958; Dzik, 1966, 1975). Vision training was established to counteract this problem and, in fact, it has been found to be effective for reducing criminal behavior and recidivism (Berman, 1989). Because (i) schizophrenia is strongly associated with criminal behavior AND with visual deterioration and (ii) vision training may reduce criminal behavior, the implementation of vision training may be effective for reducing outcome severity in schizophrenia.

Vision training regarding information processing (see "Processing of visual information" in **Table 2**) may be achieved by functional reorganization. Patients can be taught to utilize different neurofunctional pathways via neurofeedback. Neurofeedback from functional real-time MRI can be used to regulate one's own brain activity, as has been shown predominantly for affective disorders (Linden et al., 2012; Micoulaud-Franchi et al., 2012) but also for the dopamine system (Sulzer et al., 2013) and in Tourette's syndrome (Messerotti Benvenuti et al., 2011). Patients with schizophrenia may learn to avoid certain brain structures associated with hallucinative experiences, such as the Insula and the TPJ. Instead, activation patterns from blind individuals could be mimicked, that is, patients could be taught to activate the occipital lobe for non-visual information processing. Moreover, lower- *and* also higher-order visual information processing may be a target for neurofeedback training in patients with schizophrenia.

Producing declines in a person's visual capacity may be a more radical but also a more effective intervention method especially when the affected individual is suffering from visual impairment. An interesting approach here entails sensory substitution devices for patients with schizophrenia. Intriguingly,


**Table 2 | "Protection-against-schizophrenia" (PaSZ) – contributions from visual information acquisition and processing.**

vision-deprived individuals can learn to behave in a manner similar to sighted individuals, that is, they may perceive depth, localization, and distance information in real-time from non-visual cues. In a series of inspiring reports, Amedi and colleagues as well as other research groups have shown that visual information can be transmitted via non-visual cues, thus allowing blind individuals to perceive and construct a 3-D image of their environment. The vision-deprived individual obtains VS information auditorily, that is, s/he *hears* visual information. Studies have shown that this type of sensory substitution is easily learned by blindfolded and blind individuals (Amedi et al., 2007; Collignon et al., 2007; Reich et al., 2012). With regard to the etiological considerations of schizophrenia, one interesting research question would be whether or not teaching blind individuals "to see" with sensory substitution devices could circumvent the protective mechanisms of congenital blindness against schizophrenia. This means that if individuals who have acquired the ability to decode VS information from non-visual (e.g., auditory) cues did not develop schizophrenia, cortical reorganization would contribute significantly to blind individuals' immunity against psychosis. However, if this was not the case, modality-unspecific world representations would be more crucial. Sensory substitution devices could then be used for patients with schizophrenia to induce neurofunctional compensation and reorganization for impaired (and possibly deteriorated) visual capacities, accordingly.

The fact that functional reorganization due to sensory deprivation is age-specific is important for all vision trainings (Cohen et al., 1999; Sathian, 2005; Ofan and Zohary, 2007; Cattaneo et al., 2008). For example, independent of sensory input or visual experience, certain brain areas maintain stimulus selectivity (Striem-Amit et al., 2012b), indicating that the human brain has intrinsic constraints with regard to functional plasticity (Striem-Amit et al., 2012d). These constraints must be taken into account when implementing vision training programs.

Finally, visual capacity and schizophrenia show gender effects. Interestingly, disease-related cognitive deficits are not different for men and women, implying that VS capacities do not affect patients' cognitive deficits (Albus et al., 1997; Halari et al., 2006; Wolitzky et al., 2006; Landgraf et al., 2010a). Nevertheless, there are some distinctions regarding the clinical trajectories of men and women who suffer from schizophrenia. Women are diagnosed at an average age of 26.5 years, whereas men are diagnosed earlier at around 21.4 years (Markowitch, 1997). Furthermore, comorbid substance abuse is more frequent in men than in women (Ochoa et al., 2012). Only 7% of female patients with schizophrenia compared to 22% of male patients are convicted of a crime after being discharged from the forensic facility (Soyka et al., 2004). Demographically, 55% of all female forensic patients are single, 15% are married, and 25% are divorced (Melzer, 2001). By contrast, the vast majority of male forensic patients are single (85%), and only 15% are married or divorced (Leygraf, 1988; Nowara, 1993). Suicide rates differ between male and female patients, and men more often successfully commit suicide than women (Markowitch, 1997). When researchers have examined the continuity perspective of schizophrenia, they have rarely taken these gender differences into consideration.

# **CONCLUSION**

The ability to see appears necessary but not sufficient for schizophrenia symptoms to develop. The PaSZ model provides a *continuous* measure for assessing the risk of schizophrenia. It suggests that, first, "absent" and "perfect" vision are associated with a lower risk of developing the disorder. Second, there is a peak in schizophrenia risk where visual capacity disturbances are "ideal" for the development of psychosis. Third, both declines AND improvements in visual functioning may improve PaSZ depending on the visual capacities (impaired, deteriorated) of the affected individuals. We argue that the understanding of the causes of schizophrenia and its development can be derived from a continuous vision-based model. And in this review, we presented evidence for this point of view from the "blindness" and the "vision" perspective, ultimately deriving interventional recommendations.

In the "blindness" part, we provided clues about what protects visually impaired, that is, congenitally blind individuals from psychosis. While blind individuals have cognitive experiences that are similar to sighted individuals, they show alterations in cognition (attentional capacities, inhibition of task-irrelevant stimuli, serial processing, strategic adaptation), neurofunctioning (amodal representations, information processing reorganization), and multisensory integration (interference, lateralization, temporal integration, imagery). In fact, these considerations raise the question of how much visual information processing is actually necessary for a person to become vulnerable to psychosis. Or, in other words, how little visual impairment is still protective against manifesting psychotic episodes. Future studies should investigate this and other questions regarding candidates for schizophrenia-specific developmental trajectories. In line with these assumptions, we already hypothesized in a former work that tasks that tap into multisensory integration and full-body motor and navigational control may improve discrimination rates between different disease stages (Landgraf et al., 2012). Nevertheless, individual differences may also contribute to qualitative and quantitative alterations in cognitive functioning regarding blind, vision impaired, and sighted schizophrenic patients (Heller and Kennedy, 1990; Cornoldi et al., 1991; Andreasen and Black, 2006; Andreasen, 2007; Cattaneo et al., 2008).

In the "vision" part, we demonstrated that deteriorated vision, in the form of disturbed visual information acquisition and processing, can lead to the development of schizophrenia symptoms. Specifically, we showed that basic functions, such as oculomotor control and strategic eye movements, are disturbed. Moreover, visuo-cognitive aberrations appear to be based on these deficits, thus resulting in a pattern that leads to motor and self-perception disturbances. These considerations allowed us to pinpoint the etiological underpinnings of schizophrenia because they show that deteriorated visual information acquisition and processing may contribute to the establishment of higher-order cognitive dysfunctions and subsequent symptoms. From a developmental point of view it could be argued that visual capacities may never develop normally in individuals with schizophrenia. If this was the case, the PaSZ model predicts that the majority (if not all) of patients with schizophrenia would be found on the "visual impairment" rather than the "visual deterioration" side. In addition, future research should identify vision- and cognition-related disease stages and determine the time point at which (or the time period in which) interventions may be most important and most effective.

In the "therapeutic" part, we proposed interventional strategies that resulted from the "blindness" and the "vision" perspectives. Decisively, declines in visual capacity (impairment) *and* improvements in visual capacity (deterioration) may increase PaSZ. We formulated treatment options, including vision training, with regard to visual information acquisition (scanning for task-relevant information, weighing information) and processing (neurofunctional sites, sensory substitution). In fact, cortical functional reorganization appears to be most crucial for successful interventions and may be induced by neurofeedback or sensory substitution. Hence, whereas patients would undoubtedly profit from being trained to improve and put *more* weight on visual functioning, we argue that learning how to decline and put *less* weight on visual functioning – the way (congenitally) blind individuals do – may lead to an even stronger protection against psychosis.

In this context, it has to be kept in mind that not all patients suffering from schizophrenia show visual dysfunctions and not all individuals with visual dysfunctions develop schizophrenia. On the one hand, this could be due to the fact that the PaSZ model may only apply to a specific subgroup of patients/individuals. For example, the influence of monocular vision, neurofunctional compensatory mechanisms, and genetic effects need to be investigated in the future. Further, it would be interesting to characterize specific subgroups along etiological dimensions. On the other hand, however, this may also indicate that the current diagnostic (and prodromal) criteria for schizophrenia are too coarse in order to dissociate between different etiological factors. Hence the

#### **REFERENCES**


O. (2008). Predictive saccades are impaired in biological nonpsychotic siblings of schizophrenia patients. *J. Psychiatry Neurosci.* 33, 17–22.


PaSZ model may provide a fine-grained tool that assesses psychotic symptomatology on the basis of developmental trajectories in much greater detail than current diagnostic procedures. This would result in greater diagnostic precision and, in turn, better therapeutic assignments for affected individuals.

Overall, this review stresses that although the ability to see makes us human (Hegel, 1807; Darwin, 1859; Plato, 380 BC; Wittgenstein, 1921), it also precludes our "PaSZ." In his "Metaphor of the Sun," the great Greek philosopher Plato lets his teacher, Socrates, argue that the most important of all senses, *vision,* determines our ways of thinking and how we experience the world (Halfwassen, 2006). Hence, our task of understanding the contribution of the visual system to psychosis will not only eventually lead us to be better able to predict, diagnose, and heal one of the most devastating mental disorders. It will also increase our understanding of how visual functioning influences our ways of thinking and, therefore, our mere existence as human beings.

#### **ACKNOWLEDGMENTS**

Christine M. Urbanski inspires this work. The paper profited from conversations with Elke van der Meer and the SCHAM group. The authors are deeply indebted to the members of the Department of Forensic Psychiatry and Psychotherapy in Regensburg, as well as to the three reviewers for their constructive comments on earlier drafts of this manuscript. Inka Bauer, Jonathan Moreno, and Jane Zagorski helped with language editing. Finally, the authors would like to openly thank the journal's guest editor, Steven Silverstein, for providing us with this fruitful opportunity of inviting and accepting our paper.

*Nat. Rev. Neurosci.* 7, 268–277. doi:10.1038/nrn1884


functions in prepsychotic patients. *Prog. Neuropsychopharmacol. Biol. Psychiatry* 29, 621–625. doi:10.1016/j.pnpbp.2005.01.008


cognition in schizophrenia using saccadic eye movements: a selective overview. *Neuropsychologia* 39, 742–756. doi:10.1016/ S0028-3932(00)00155-X


*Gen. Psychiatry* 62, 495–504. doi:10.1001/archpsyc.62.5.495


Cattaneo, Z., and Vecchi, T. (2011). *Blind Vision.* London: The MIT Press.


doi:10.1016/j.neuropsychologia. 2004.09.013


*Favoured Races in the Struggle for Life.* London: John Murray.


and antisaccade eye movements in siblings discordant for schizophrenia. *J. Psychiatr. Res.* 38, 177–184. doi:10.1016/S0022-3956 (03)00105-5


the symptoms of schizophrenia: abnormalities in the awareness of action. *Brain Res. Brain Res. Rev.* 31, 357–363. doi:10.1016/S0165- 0173(99)00052-1


in adolescents with schizophrenia: a study with event-related potentials and functional magnetic resonance imaging. *Arch.* G*en. Psychiatry* 64, 1229–1240. doi:10.1001/archpsyc.64.11.1229


pictures, and the blind. *Percept. Psychophys.* 48, 459–466. doi:10.3758/BF03211590


Bonnet Syndrome. *Psychogeriatrics* 9, 77–84. doi:10.1111/j.1479- 8301.2009.00288.x


schizophrenia: a critical systematic review. *J. Psychiatr. Res.* 47, 91–104. doi:10.1016/j.jpsychires.2012.10.001


doi:10.1016/j.neuropsychologia. 2010.05.034


performance on the standard antisaccade task meet the co-familiality criterion for an endophenotype? *Brain Cogn.* 68, 462–475. doi:10.1016/j.bandc.2008.08.027


*Ment. Dis.* 195, 618–621. doi: 10.1097/NMD.0b013e318093f437


doi:10.1093/oxfordjournals.schbul. a007046


*Schizophr. Bull.* 33, 772–781. doi:10.1093/schbul/sbm020


doi:10.1001/archpsyc.1995.0395022 0031007


676–681. doi:10.1016/j.pnpbp. 2009.03.012


*Psychol. Bull.* 124, 123–164. doi:10.1037/0033-2909.124.2.123


*Opin. Neurol.* 25, 86–95. doi:10. 1097/WCO.0b013e32834ed723


Neville, H. J. (1999). Improved auditory spatial tuning in blind humans. *Nature* 400, 162–166. doi:10.1038/22106


in the schizophrenia prodrome. *Schizophr. Bull.* 33, 761–771. doi:10.1093/schbul/sbm018


a high-density electrical mapping study. *Neuropsychologia* 49, 3178–3187. doi:10.1016/ j.neuropsychologia.2011.07.017


and working memory impairments in schizophrenia. *Arch. Gen. Psychiatry* 59, 146–153. doi:10.1001/archpsyc.59.2.146


*Learn Mem. Cogn.* 9, 269–282. doi:10.1037/0278-7393.9.2.269


construed as a potential conflict of interest.

*Received: 31 January 2013; accepted: 30 May 2013; published online: 01 July 2013.*

*Citation: Landgraf S and Osterheider M (2013) "To see or not to see: that is the question." The "Protection-Against-Schizophrenia" (PaSZ) model: evidence from congenital blindness and visuocognitive aberrations. Front. Psychol. 4:352. doi: 10.3389/fpsyg.2013.00352*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Landgraf and Osterheider. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*

# Is vision in schizophrenia characterized by a generalized reduction?

#### *Bernt C. Skottun1 \* and John R. Skoyles <sup>2</sup>*

#### *<sup>1</sup> Independent Scholar, Oslo, Norway*

*<sup>2</sup> Centre for Mathematics and Physics in the Life Sciences and Experimental Biology, University College London, London, UK*

*\*Correspondence: berntchrskottun@gmail.com*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Michael Herzog, École Polytechnique Fédérale de Lausanne, Switzerland Michael-Paul Schallmo, University of Minnesota, USA*

**Keywords: schizophrenia, vision, sensitivity, VEPs, magnocellular**

#### **A commentary on**

#### **Sensory contributions to impaired emotion processing in schizophrenia**

*by Butler, P. D., Abeles, I. Y., Weiskopf, N. G., Tambini, A., Jalbrzikowski, M., Legatt, M. E., et al. (2009). Schizophr. Bull. 35, 1095–1107. doi: 10.1093/schbul/sbp109*

How the visual capabilities of those with schizophrenia differ from those of individuals without schizophrenia is a topic of active research. Of special interest is the question of whether or not they might have a magnocellular deficiency. It has been concluded that contrast sensitivity in schizophrenic subjects is characterized by a general reduction in sensitivity, and so does not indicate a magnocellular deficiency (Skottun and Skoyles, 2007). Likewise, many of the reported cases of abnormal visual masking linked to schizophrenia can be described by a general reduction (see, e.g., Rassovsky et al., 2004). Also this is hard to reconcile with a magnocellular deficit since, according to the theory, such a deficit would have been expected to cause a reduction that was related specifically to the U-shaped Type-B masking function (Skottun and Skoyles, 2009). These observations prompt the question of whether or not other differences between schizophrenic subjects and controls that have been attributed to magnocellular deficiencies can also be accounted by a general reduction in sensitivity or response.

Differences between schizophrenic subjects and nonschizophrenic controls attributed to magnocellular deficiencies have been found using visually evoked potentials (VEP). Butler et al. (2009) obtained VEP data for schizophrenic subjects and controls under two conditions. One condition aimed at predominantly stimulating the magnocellular system, the other the parvocellular system. Butler et al. (2009) found statistically significant reduction in the responses from the schizophrenic subjects, relative to those of controls, under the condition favoring the magnocellular system but not under the parvocellular condition. The authors interpreted this as evidence for a magnocellular deficiency linked to schizophrenia. The present report examines the possibility that these results could reflect instead a general response reduction.

The data of Butler et al. (2009) have been re-plotted in **Figure 1**. The open and filled symbols give the results for the schizophrenic subjects and controls, respectively, while panels A and B give the data obtained under the magnoand parvocellular conditions. A general response reduction was modeled by a simple linear scaling of the response from the control group. The scaling factor was determined by the best fit to the data of the schizophrenic subjects by calculating the smallest sum of squared deviations. Analyses of both magno- and parvocellular data conditions were included, giving a single scaling value of 0.756. (The scaling factors for the magno- and parvocellular data sets computed separately were 0.737 and 0.797, respectively). The scaled data are indicated by the dashed lines in **Figure 1**. Even though the data for the

schizophrenic subjects in the magnocellular condition in some instances are slightly below the dashed line and the data in the parvocellular condition in some cases are slightly above the dashed line, the overall finding is that the scaled data give close fits to the data for the schizophrenic subjects for both conditions. This suggests that Butler et al.'s (2009) data for schizophrenic subjects in both the magnocellular and the parvocellular conditions are consistent with scaling by the same factor. This is consistent with a general reduction in the response rather than a difference (as suggested by Butler et al., 2009) between the magnocellular and parvocellular systems. No reason therefore exists to account for these data with a deficiency linked specifically to the magnocellular system.

One explanation for the differences between controls and schizophrenic subjects being significant in the magnocellular condition but not in the parvocellular one could be that the response differences were larger in the magnocellular condition. The fact that the responses were larger in this condition would have made the difference between the groups (as a result of scaling by a constant) larger under this condition, and so more likely to result in a statistically significant difference.

Further, it is an invalid inference to conclude that a magnocellular deficiency exists solely based on the finding that a magnocellular condition gives statistically significant differences whereas a parvocellular condition does not. To do so would require making an interpretation of the non-significant data—but a statistically non-significant result does not allow any conclusions to be inferred (Gill, 1999). The finding that the difference is statistically significant under the magnocellular condition and not under the parvocellular condition, moreover, is not itself strictly pertinent. What might have been somewhat more relevant would have been if the difference between the data obtained under the two conditions were statistically significant. (It should in this connection be kept in mind that the difference between a statistically significant result and a nonsignificant result need not itself be statistically significant. Gelman and Stern, 2006). However, it is not clear that even this would have solved the problem since it may be possible for a single factor to have effects that are different under different conditions (as shown in **Figure 1**). The difference between these effects may, or may not, be statistically significant.

#### **SUMMARY**

We have demonstrated that the VEP data of Butler et al. (2009) are very well described by a general response reduction. This argues that abnormal visual performance in schizophrenia reflects a general reduction in sensitivity or response amplitude rather than a magnocellular deficiency.

#### **REFERENCES**

Butler, P. D., Abeles, I. Y., Weiskopf, N. G., Tambini, A., Jalbrzikowski, M., Legatt, M. E., et al. (2009). Sensory contributions to impaired emotion processing in schizophrenia. *Schizophr. Bull*. 35, 1095–1107. doi: 10.1093/schbul/sbp109


*Received: 27 November 2013; accepted: 15 December 2013; published online: 27 December 2013.*

*Citation: Skottun BC and Skoyles JR (2013) Is vision in schizophrenia characterized by a generalized reduction? Front. Psychol. 4:999. doi: 10.3389/fpsyg.2013.00999 This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Skottun and Skoyles. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Windows to the soul: vision science as a tool for studying biological mechanisms of information processing deficits in schizophrenia

#### *Jong H. Yoon1 \*, Summer L. Sheremata2, Ariel Rokem3 and Michael A. Silver <sup>2</sup>*

*<sup>1</sup> Department of Psychiatry and Behavioral Sciences, Stanford University and Veterans Affairs Palo Alto Healthcare System, Palo Alto, CA, USA <sup>2</sup> School of Optometry, Helen Wills Neuroscience Institute, and Vision Science Graduate Group, University of California, Berkeley, Berkeley, CA, USA*

*<sup>3</sup> Department of Psychology, Stanford University, Stanford, CA, USA*

#### *Edited by:*

*Steven Silverstein, University of Medicine and Dentistry of New Jersey, USA*

#### *Reviewed by:*

*Michael Herzog, Ecole Polytechnique Fédérale de Lausanne, Switzerland Jonathan K. Wynn, University of California, Los Angeles, USA Carol Jahshan, VA Greater Los Angeles Healthcare System, USA*

#### *\*Correspondence:*

*Jong H. Yoon, Department of Psychiatry and Behavioral Sciences, Stanford University and Veterans Affairs Palo Alto Healthcare System, 3801 Miranda Avenue, Palo Alto, CA 94304, USA e-mail: jhyoon1@stanford.edu*

Cognitive and information processing deficits are core features and important sources of disability in schizophrenia. Our understanding of the neural substrates of these deficits remains incomplete, in large part because the complexity of impairments in schizophrenia makes the identification of specific deficits very challenging. Vision science presents unique opportunities in this regard: many years of basic research have led to detailed characterization of relationships between structure and function in the early visual system and have produced sophisticated methods to quantify visual perception and characterize its neural substrates. We present a selective review of research that illustrates the opportunities for discovery provided by visual studies in schizophrenia. We highlight work that has been particularly effective in applying vision science methods to identify specific neural abnormalities underlying information processing deficits in schizophrenia. In addition, we describe studies that have utilized psychophysical experimental designs that mitigate generalized deficit confounds, thereby revealing specific visual impairments in schizophrenia. These studies contribute to accumulating evidence that early visual cortex is a useful experimental system for the study of local cortical circuit abnormalities in schizophrenia. The high degree of similarity across neocortical areas of neuronal subtypes and their patterns of connectivity suggests that insights obtained from the study of early visual cortex may be applicable to other brain regions. We conclude with a discussion of future studies that combine vision science and neuroimaging methods. These studies have the potential to address pressing questions in schizophrenia, including the dissociation of local circuit deficits vs. impairments in feedback modulation by cognitive processes such as spatial attention and working memory, and the relative contributions of glutamatergic and GABAergic deficits.

**Keywords: schizophrenia, visual system, fMRI, psychophysics, magnetic resonance spectroscopy**

# **INTRODUCTION**

Schizophrenia is one of the most perplexing and important mysteries in modern medicine. This condition is associated with significant impairments across diverse functional domains, usually conferring to the affected individual a lifetime of disability and the need for long-term treatment. The prevalence of schizophrenia is nearly 1% of the general population, and it constitutes one of the largest public health burdens of any condition (Knapp et al., 2004). Among the diverse symptoms of schizophrenia, cognitive and information processing deficits are now widely recognized to be one of the most important causes of chronic functional impairments (Green et al., 2000). Unlike psychotic symptoms, deficits in cognition and information processing are largely refractory to currently available treatments (Goldberg et al., 2007). While intense research efforts have yielded tantalizing clues to the neural substrates of these deficits, we still possess limited understanding of the associated neural mechanisms. Further elucidation of these mechanisms will directly inform development of novel treatments. Thus, new approaches are needed to help spur advances in the discovery of neural mechanisms of cognitive and information processing deficits in schizophrenia.

The visual system has increasingly been recognized as having substantial advantages for conducting research on neural mechanisms of schizophrenia (Butler et al., 2008a; Green et al., 2009a; Silverstein and Keane, 2011a). Although visual disturbances are not generally considered to be among the most prominent clinical symptoms of schizophrenia, the visual system is nonetheless an important site of pathology and dysfunction in this disease. The high prevalence of visual perceptual abnormalities, including those occurring early in the course of illness, has been documented for many years by phenomenological studies (Chapman, 1966; Bracha et al., 1989). Moreover, a growing number of postmortem studies suggest that these visual perceptual symptoms could be the result of neuropathologic abnormalities within the visual cortex (Selemon et al., 1995; Glantz and Lewis, 2000; Dorph-Petersen et al., 2007; Hashimoto et al., 2008). As others have previously noted (Silverstein and Keane, 2011b), the combination of the relatively advanced state of knowledge of visual physiology and functional anatomy compared to other brain systems and the availability of sophisticated investigative tools make vision an excellent system to investigate links among neural circuitry, information processing, and behavior. Thus, the study of visual processing abnormalities in schizophrenia presents rich opportunities to translate the tools, knowledge, and concepts from basic vision science to elucidate the mechanisms of impaired information processing in this condition.

These attributes of the visual system have resulted in an increasing number of investigations that have employed visual perceptual and neuroscience techniques to study schizophrenia. In the first part of this paper, we will review this emerging literature and highlight the confluence of factors that have made the visual system a useful model for the study of information processing deficits in schizophrenia. We have selected for review some of the most productive lines of research, including studies from our group, illustrating the diversity of visual processes that have been investigated and the range of paradigms that are available to study structure-function relationships and their perturbations in schizophrenia.

In the second part of this review, we highlight the many advantages of the visual system for translational research and describe some of the methods that allow inferences at a level of specificity that are typically not possible in studies of other brain systems. At the heart of these methods is psychophysics, the science of quantitatively measuring and modeling the relationships between physical properties of visual stimuli and their perceptual consequences. Psychophysical methods allow investigators to quantify and control for the generalized deficit confound in task performance in schizophrenia (Knight and Silverstein, 2001; Silverstein, 2008). The psychophysical study of perceptual phenomena can also lead to a deeper understanding of the computational constraints on the system, which in turn can illuminate the associated physiological mechanisms. Given these unique advantages of psychophysical approaches, we will review how they have been used to infer the physiological and computational mechanisms that underlie visual information processing.

Perceptual tasks can easily be paired with non-invasive functional neuroimaging, allowing testing and confirmation of many such inferences *in vivo*. After describing evidence of visual dysfunction in several domains of visual perception in schizophrenia, we will explain how psychophysical analysis and design elements of psychophysical tasks can be translated to the study of specific information processing deficits in schizophrenia.

We also discuss concepts that inform our thinking in considering the visual system as an *in vivo* model system for understanding information processing deficits in schizophrenia. These include the canonical cortical circuit: the conservation of the morphological and neurotransmitter subtypes of neurons and patterns of local connectivity across all of the neocortex, which suggests that findings from the visual system may generalize to other neocortical systems (Douglas and Martin, 2004). In addition, the conservation of specific aspects of the functional architecture of the visual system across species offers unique opportunities for translational neuroscience studies. Although schizophrenia affects a number of different neurophysiological and neurochemical systems, in this review we highlight the magnocellular pathway and the neurotransmitter GABA in visual processing. The visual studies that have been conducted in these systems illustrate the refined neurobiological inferences that may be drawn from schizophrenia studies of well-characterized systems in the brain.

Finally, in the last part of this review, we present a discussion of the ways in which visual studies may inform some of the most important unresolved questions relating to the pathophysiology of schizophrenia (glutamate and GABA abnormalities and local circuit vs. feedback modulatory deficits).

# **STUDIES OF VISUAL PERCEPTION IN SCHIZOPHRENIA**

Research in schizophrenia has examined a wide variety of processes that have implicated abnormalities at various levels of the visual system, from the retina (Balogh et al., 2008) to higher-order extrastriate visual cortex (Butler et al., 2008b). An exhaustive survey of visual studies in schizophrenia is beyond the scope of this paper. Instead, we review a select sample of studies, including ones conducted by our group, that illustrates the translational potential of the visual system in elucidating the underlying mechanisms of information processing deficits in schizophrenia. We emphasize two recurring themes shared by the visual studies reviewed here: the well-studied functional neuroanatomy of the visual system and the highly developed investigational tools that are available in the vision sciences. The first feature is important because it provides a relatively extensive knowledge base, compared to other neural systems, with which findings in schizophrenia may be contrasted to help identify abnormalities in this condition. Both features are critical in allowing investigators to test hypotheses and draw inferences at a level of specificity and neurobiological detail that is often not possible in other systems in the brain.

# **MOTION PROCESSING**

Motion processing is among the most frequently studied visual processes in schizophrenia. Befitting such an ecologically critical function for primates, the primate brain has evolved a set of specific mechanisms for the processing of motion. Decades of basic research have identified a number of these mechanisms (Burr and Thompson, 2011), including the magnocellular visual processing stream that is specialized for detecting transient and dynamic changes in the visual environment (Livingstone and Hubel, 1988). Other brain regions involved in motion processing include visual cortical areas such as MT (medial temporal) and MST (medial superior temporal) that respond to moving stimuli and contain neurons with selectivity for motion features (Huk et al., 2002; Amano et al., 2009; Kolster et al., 2010). Note that the names of these areas refer to anatomical locations within the macaque brain, where they were initially discovered, and that homologous areas in the human brain are often referred to by the same names, even though these areas are not located in human medial temporal cortex. An extensive set of well-validated psychophysical methods for measuring components of motion processing has been established (Burr and Thompson, 2011), providing a foundation for the translation of motion processing paradigms into clinical studies.

Studying motion processing in patients with schizophrenia has proven important for several reasons. First, these studies show that visual paradigms can be used to reveal both impaired and preserved aspects of motion perception in schizophrenia, thereby providing a means of addressing the generalized deficit confound (Silverstein, 2008), a significant impediment to progress in schizophrenia research (see Overcoming the Generalized Deficit Confound section below). This confound reflects the fact that schizophrenia is characterized by deficits in almost any measured behavior (Knight and Silverstein, 2001). Thus, in the domain of motion perception, velocity discrimination (Chen, 2011) and spatial integration (Green et al., 2009a), but not direction discrimination (Chen et al., 2003), exhibit selective visual dysfunction in schizophrenia.

Patients with schizophrenia and their first-degree relatives have long been known to have impairments in the ability to track moving objects with their eyes. Although these tracking eye movements, known as smooth pursuit eye movements, represent a relatively simple and common behavior, they nonetheless require highly precise coordination of visual and motor systems (Lisberger et al., 1987). Specifically, the successful tracking of an object first requires the ability to determine the velocity and direction in which the object is moving. While velocity discrimination (**Figure 1A**) is generally impaired in schizophrenia patients and correlates with eye tracking dysfunction (Chen et al., 1999), motion direction discrimination itself is not generally impaired and only demonstrates deficits when direction signals must be integrated over spatial locations (Chen et al., 2003). Motion processing deficits have also been found for higher-order motion in schizophrenia, even in the absence of lower-order motion processing deficits (Chen et al., 2004; Kandil et al., 2013).

Motion perception deficits have been reported not only in patients but also in their unaffected first-degree relatives (Chen et al., 2005). This suggests that motion perception deficits may have a genetic component and do not result from generalized deficits or medication effects. In addition, the fact that patients with schizophrenia have increased activity during motion processing in areas involved in cognitive control, such as the prefrontal cortex (Chen et al., 2008a), indicates that perceptual impairments may increase demand on cognitive control systems.

# **MASKING**

Visual masking has been another rich area of research in schizophrenia. Masking refers to impaired perception of a behaviorally relevant stimulus (target) due to spatial and/or temporal proximity of a behaviorally irrelevant stimulus (mask) (**Figure 1B**). As extensively reviewed elsewhere (Green et al., 2011a), masking paradigms offer a number of advantages in elucidating neural mechanisms of information processing deficits in schizophrenia, including the ability to precisely control stimulus parameters, the presence of well established neurobiological models of masking, and the correlation of visual masking impairments with core symptoms in schizophrenia (Green and Walker, 1986; Braff, 1989).

While abnormalities in a variety of types of masking have been documented in schizophrenia, the best studied is backward masking. In backward masking, a mask is presented a fraction of a second after target onset. It is thought that two processes, interruption and integration, can contribute to visual masking and that these processes involve distinct neural mechanisms. Masking by interruption occurs after the target representation has already been formed and is based on interference with higherlevel, feedback processes that underlie conscious perception of the target (Enns and Di Lollo, 2000). Masking by integration has been suggested to occur as a result of formation of a single perceptual representation of both the target and the mask (Turvey, 1973).

A recent study has employed the 4-dot masking paradigm, thought to selectively involve interference by interruption, to show that patients with schizophrenia show larger and more prolonged masking effects (Green et al., 2011b). In addition, unaffected first-degree relatives exhibit impaired performance on a backward masking task (Green et al., 1997). Patients with schizophrenia show abnormalities even when the target and mask do not overlap and masking by integration cannot occur (Green et al., 2011b), suggesting that abnormal masking is due to impairments in feedback processing. Electroencephalographic (EEG) studies have implicated mechanisms involving gammaband synchrony in visual masking abnormalities in schizophrenia. In particular, patients with schizophrenia exhibited decreased evoked gamma-band activity during backward masking compared to controls but had intact gamma-band activity evoked by unmasked visual stimuli (Green et al., 2003; Wynn et al., 2005).

#### **PERCEPTUAL ORGANIZATION**

Perceptual organization refers to the integration of visual elements across spatial locations to form a representation of a coherent object (Palmer, 1999) (**Figure 1C**). The idea that impairments in this integration may reflect underlying neural disorganization in schizophrenia has been the motivation for a number of studies (Phillips and Silverstein, 2003; Uhlhaas and Silverstein, 2005; Silverstein and Keane, 2011a). A variety of perceptual tasks have been used in these investigations, including viewing of luminance-binarized, or Mooney, faces (Uhlhaas et al., 2006a), identification of objects from fragmented images (Doniger et al., 2002; Sehatpour et al., 2010), perception of illusory contours (Spencer et al., 2003), and integration of interrupted contours (Silverstein et al., 2000; Uhlhaas et al., 2005, 2006b). There are strong associations between impairments in perceptual organization in schizophrenia and symptoms, including behavioral/cognitive disorganization (Silverstein et al., 2000; Phillips and Silverstein, 2003; Uhlhaas et al., 2005).

Object perception may require feature-binding processes that rely on neural synchrony (Engel et al., 2001). Patients with schizophrenia exhibited reduced phase synchrony in the beta (Uhlhaas et al., 2006a) and gamma (Spencer et al., 2003) bands during Gestalt perception. Moreover, patients with schizophrenia showed decreased BOLD responses to objects defined by contour integration in higher-order visual cortical areas (V2–V4) and in cortical attention networks but not in primary visual cortex (Silverstein et al., 2009). Similarly, when identifying an object from fragmented images, patients with schizophrenia exhibited decreased BOLD signal in visual and attention cortical networks, as well as the hippocampus, but not in primary visual cortex (Sehatpour et al., 2010). Because perceptual organization relies on visual processing areas beyond primary visual cortex (Fang et al., 2008), these findings suggest that perceptual organization deficits

**FIGURE 1 | Examples of paradigms and stimuli used to study specific visual deficits in schizophrenia. (A)** Velocity perception. In velocity discrimination tasks, subjects compare the velocity (represented here by the size of the arrows) of two sequentially presented stimuli (in this case, gratings). **(B)** Masking. In paracontrast masking, the mask interrupts formation of the stimulus percept. Because the mask does not spatially overlap the target contours, paracontrast masking cannot occur through integration of the target and mask. In four-dot masking, the mask is thought to disrupt reentrant processing but not formation of object representations. **(C)** Perceptual grouping. A set of elements in the display is perceptually grouped via contour integration to form a coherent percept (in this case, an ellipse). As the amount of orientation jitter between the different elements

in schizophrenia do not result from impaired processing of individual visual elements but rather from abnormalities in specific early visual cortical circuits as well as in higher-order networks that modulate these circuits.

#### **CONTEXTUAL MODULATION OF VISUAL PROCESSING**

Contextual modulation in vision refers to the process by which perception or neural representation of a stimulus is influenced by the spatial and/or temporal context in which it is displayed (Albright and Stoner, 2002). Altered contextual modulation has been abundantly documented in schizophrenia using a wide variety of paradigms (Silverstein et al., 1996, 2013; Must et al., 2004; Dakin et al., 2005; Kéri et al., 2005; Tadin et al., 2006; Uhlhaas et al., 2006b; Chen et al., 2008b; Yoon et al., 2009; Schallmo et al., 2013; Tibber et al., 2013; Yang et al., 2013). The study of contextual modulation holds great promise for investigating the neural bases of impaired information processing in schizophrenia, as it arises from well characterized neural interactions that can be precisely manipulated by appropriate choices of stimuli and task parameters.

forming this shape is increased, the strength of the coherent percept decreases. **(D)** Orientation tuning. Adaptation to a grating with a specific orientation reduces the sensitivity for detection of a post-adaptation target grating with a similar orientation. The degree to which adaptation effects diminish with increasing adapter/target orientation differences provides an estimate of the width of orientation tuning in the visual system. **(E)** Orientation-specific surround suppression. The presence of a high-contrast surround decreases the perceived contrast of the center, and suppression is strongest when the orientation of the center and surround are parallel. **(F)** Magnocellular vs. parvocellular stimuli. Stimuli with low spatial frequencies preferentially engage the magnocellular pathway, while stimuli with higher spatial frequencies preferentially engage the parvocellular pathway.

One of the best studied forms of contextual modulation in schizophrenia is surround suppression of contrast. This process refers to the reduction in perceived contrast of a stimulus when it is surrounded by a high-contrast stimulus (Chubb et al., 1989). Surround suppression is thought to play an important functional role in scene segmentation (Walker et al., 1999), enhancing salience processing (Petrov and McKee, 2006) and the generation of perceptual constancies (MacEvoy and Paradiso, 2001). Abnormal surround suppression has been reliably demonstrated in patients with schizophrenia across multiple tasks and stimuli (Dakin et al., 2005; Tadin et al., 2006; Chen et al., 2008b; Yoon et al., 2009; Tibber et al., 2013; Yang et al., 2013).

A specific form of surround suppression, referred to as orientation-specific surround suppression (OSSS) (**Figure 1E**), represents a particularly appealing paradigm for investigating schizophrenia. This appeal is due to the detailed and specific neurobiological inferences that can be made relating to this phenomenon. For oriented grating stimuli, surround suppression is greater when the target stimulus and the suppressive surround share the same orientation (parallel condition) (Cannon and Fullenkamp, 1991; Solomon et al., 1993; Xing and Heeger, 2000, 2001; Kosovicheva et al., 2012). This orientation specificity of surround suppression suggests a locus in early visual cortex, where neurons exhibit orientation-selective responses. Indeed, OSSS is also evident in the responses of single neurons in primary visual cortex (Blakemore and Tobin, 1972; Cavanaugh et al., 2002). Moreover, fMRI (Zenger-Landolt and Heeger, 2003) and magnetoencephalographic (MEG) and EEG (Haynes et al., 2003) correlates of surround suppression have been reported in human primary visual cortex. Furthermore, modeling of psychophysical and fMRI surround suppression data indicates that activity in visual cortical area V1 provides a better quantitative account of the behavioral results than higher-order areas (Zenger-Landolt and Heeger, 2003). This study is an example of the power of combining psychophysical measurements, functional imaging, and quantitative modeling to elucidate the neural mechanisms of a perceptual phenomenon.

OSSS has been measured perceptually with contrast discrimination thresholds for center stimuli in different surround orientations. Relative to a no-surround condition, healthy control subjects showed larger increases in contrast discrimination thresholds (i.e., more surround suppression) than patients with schizophrenia, and this group difference was selective for the parallel orientation condition (Yoon et al., 2009). Importantly, the finding of intact or even better performance in patients, compared to control subjects, mitigates the generalized deficit confound (Knight and Silverstein, 2001; Silverstein, 2008).

Two recent studies (Yang et al., 2013; Tibber et al., 2013) reported weakened surround suppression for contrast but not other features in patients with schizophrenia. The specificity of abnormalities in schizophrenia for contextual processing of contrast may explain why some studies have not found deficits in contextual modulation for other visual features in schizophrenia (Schütze et al., 2007; Roinishvili et al., 2008).

# **ADVANTAGES OF THE VISUAL SYSTEM FOR TRANSLATIONAL RESEARCH**

#### **CONSERVATION ACROSS SPECIES**

The evolutionary preservation of functional characteristics of the visual system across mammals, such as orientation bandwidth of tuning functions in V1 and V2 (van den Bergh et al., 2010), provides motivation for studies of vision in translational research. Conservation of visual cortical organization across species allows translational conclusions to be made regarding neural circuits and molecular mechanisms from experiments in other mammalian species that cannot be conducted in humans. Molecular mechanisms of visual information processing can be studied using anatomical techniques, such as staining and visualization of particular molecular components in visual cortical circuits (Disney et al., 2007) and with pharmacological techniques that can be coupled with electrophysiological recordings of visual responses (Katzner et al., 2011).

Optogenetic studies allow selective stimulation of particular molecular components of visual circuits in a relatively temporally and spatially focal manner, thereby revealing links between molecular components and neural selectivity in the visual system. For example, GABAergic interneurons in the visual cortex are broadly divided into two different classes based on the relative expression of somatostatin and parvalbumin mRNA. In a rodent model, these two classes of cells have been associated with separate types of selectivity of responses to visual stimuli (Ma et al., 2010), and a recent study using optogenetic techniques (Lee et al., 2012) has provided causal evidence for this cell type-specific stimulus selectivity. Importantly, post-mortem studies have shown reduced expression of both somatostatin and parvalbumin mRNA transcripts in many cortical regions in patients with schizophrenia (Hashimoto et al., 2008).

In many cases, conservation of visual cortical and perceptual organization allows investigators to utilize similar or even identical brain measures and perceptual paradigms across species. Thus, the results of animal studies can directly inform human findings and strengthen neurobiological inferences. For example, OSSS has been studied in cats (Blakemore and Tobin, 1972) and non-human primates (Cavanaugh et al., 2002) to demonstrate surround suppression of responses of primary visual cortical neurons. Furthermore, pharmacological blockade of GABAergic receptors reduces surround suppression in primary visual cortex (Ozeki et al., 2004), while non-invasive measures of visual cortical GABA with magnetic resonance spectroscopy (MRS) is highly predictive of behavioral OSSS in humans (Yoon et al., 2010). Moreover, patients with schizophrenia exhibited lower behavioral OSSS as well as decreased visual cortical GABA levels (Yoon et al., 2010). Thus, the link between GABA and reduced OSSS in schizophrenia is significantly strengthened by the ability to investigate the same phenomenon (OSSS) using different experimental tools in a variety of species.

Synchronized oscillatory neural activity is another brain process with cross-species translational potential. In particular, oscillations in the gamma frequency range have been of particular interest, as they are considered to be a fundamental building block of neural computation subserving diverse functions [see Fries (2009) for review]. While a major focus of gamma-band synchrony research in schizophrenia has been higher-order cognition (Cho et al., 2006; Minzenberg et al., 2010), deficits in gammaband synchrony have also been demonstrated for visual processes in schizophrenia (Spencer et al., 2003). The fact that the visual system has been one of the most active areas of gamma synchrony research in animal models (Singer and Gray, 1995; Womelsdorf et al., 2006; Bosman et al., 2012) provides many opportunities for translational research in this area.

#### **THE CANONICAL CORTICAL CIRCUIT**

Studying the visual system to gain fundamental insights about basic information processing deficits in schizophrenia is motivated by the fact that the neocortex has a high degree of conservation of morphological and neurotransmitter subtypes of neurons as well as patterns of laminar and tangential connectivity (Douglas and Martin, 2004). This cortical "canonical circuit" has been extensively characterized (Douglas et al., 1989; Markov et al., 2011) and includes laminar-specific reciprocal local connections between excitatory glutamatergic neurons and inhibitory GABAergic neurons, local excitatory/excitatory connections, and long-range excitatory inputs from, and outputs to, other cortical and subcortical regions.

Further evidence for the relevance of conservation of cortical circuits for information processing comes from modeling of responses in the frontal eye fields (FEF), a cortical area with neurons containing oculomotor signals. A model derived from detailed anatomical and physiological studies of early visual cortical area V1 was used to generate neuronal and connectivity parameters that were then applied to FEF, quantitatively accounting for responses of FEF neurons and FEF-dependent oculomotor behaviors in macaque (Heinzle et al., 2007) and patterns of eye movements of humans during reading (Heinzle et al., 2010).

In schizophrenia, there is significant and growing evidence that abnormalities in visual processing are correlated with core symptoms and functional impairments, including disorganization (Silverstein et al., 2000; Phillips and Silverstein, 2003; Uhlhaas et al., 2005), community functioning (Butler et al., 2005), and working memory (Haenschel et al., 2007). One interpretation of these findings, considered in the context of the canonical neocortical circuit, is that the visual cortical circuit abnormalities that presumably underlie visual processing deficits in schizophrenia reflect more general circuit abnormalities that are also present in higher-order neocortical areas, where they more directly affect disorganization and real-life functioning.

#### **OVERCOMING THE GENERALIZED DEFICIT CONFOUND**

One of the main methodological concerns in the study of behavioral differences between patients with schizophrenia and healthy controls is the generalized deficit in performance that is found in practically all tasks. This deficit is attributable to impairments in a number of cognitive and/or affective processes (Heinrichs and Zakzanis, 1998). In studies aiming to focus on specific cognitive or sensory mechanisms, this generalized deficit is a particularly important but difficult confound that needs to be controlled (Knight and Silverstein, 2001).

Fortunately, vision research provides several methods for addressing the generalized deficit confound, including several strategies that have been previously described (Silverstein, 2008). First, information available from psychometric functions can help isolate specific deficits in schizophrenia. The psychometric function is a model that relates behavioral performance on a task to a selected property of the physical stimulus (**Figure 2**). The slope of the psychometric curve indexes the reliability of the psychophysical threshold measurement (Macmillan and Creelman, 2004), and this fact has been used as a diagnostic tool for investigating visual disturbances in various patient groups (Patterson et al., 1980; Chauhan et al., 1993). More specifically, a shallow slope of the psychometric curve indicates an unreliable estimate of threshold due to inconsistent task performance. If group differences are found in this parameter, this may indicate a generalized deficit confound. On the other hand, the use of an upper asymptote parameter to fit models of psychometric data allows estimation of the frequency of lapses in attention that are not related to task difficulty (as determined by a physical stimulus feature). This is because the upper asymptote parameter of the curve quantifies subject performance when the stimulus is relatively easy to discriminate (Barch et al., 2012).

Quantitative modeling of the psychometric function can also generate confidence intervals for the values of the model

**FIGURE 2 | Psychometric curves contain multiple sources of information about behavioral performance.** Behavioral performance, measured here as percent correct trials, is plotted as a function of a stimulus parameter (blue and green data points). In this example, the independent variable is the amount of decrement in contrast required to detect the presence of a target (see text and Yoon et al., 2009, 2010). Smaller contrast decrements are more difficult to discriminate, so the psychometric curve is monotonically increasing. In this example, cumulative Weibull functions are fit to the data (smooth green and blue lines). For each function, threshold is defined as the amount of contrast decrement that yields 82% correct trials. The percent correct corresponding to chance performance determines the lower asymptote of the curve (50% in this example). The blue curve has a lower threshold than the green curve, consistent with a higher percentage of correct trials at an equivalent contrast decrement compared to the green curve. Different slopes in the two curves indicate different levels of reliability across experimental conditions or groups of subjects. The upper asymptote is an index of attentional lapses, as the stimuli for these trials are well above perceptual threshold, and performance less than 100% indicates that attention was not properly engaged on some trials. The difference between the upper asymptote of the curve and 100% correct performance is a measure of the rate at which subjects are responding incorrectly on trials that should be easy to perform at 100% accuracy based on the physical level of the stimulus (e.g., large values of the contrast decrement in this task). This difference is comparable to estimates of the lapse rate in catch trials (e.g., Barch et al., 2012).

parameters. One potential manifestation of a generalized deficit is that patients with schizophrenia might fatigue more easily over the course of an experimental session, resulting in deteriorating task performance. This would not only result in elevated thresholds in patients but also more variability in the threshold estimate, as the threshold would be changing during the session. This variability can be quantified with bootstrapping analysis (Efron and Tibshirani, 1993). We used psychometric functions to assess the possibility that generalized deficits affect the stability of OSSS performance in patients with schizophrenia (Yoon et al., 2009). We found no difference between patients and controls, suggesting that although generalized deficits may affect overall task performance (thresholds were generally higher in patients than in controls), they did not manifest as increased variability in performance throughout the experimental session. A similar analysis was also used in Dakin et al. (2005) and Tibber et al. (2013).

Importantly, many visual paradigms allow systematic manipulation of stimulus features in a way that does not affect the cognitive demands of the task. In the case of surround suppression, the generalized deficit confound can be addressed by varying the relative orientation of the surround and the center stimuli (**Figure 1E**). The task structure and demands do not differ between the parallel and orthogonal conditions, yet psychophysical thresholds are substantially different. Hence, an orientation-selective component of surround suppression can be isolated that is relatively independent of the cognitive demands of the task. This approach has revealed that although patients with schizophrenia have higher contrast discrimination thresholds, they show less OSSS (Yoon et al., 2009; see also Dakin et al., 2005; Tibber et al., 2013; Yang et al., 2013). In addition, patients' perception is more veridical regarding the actual contrast of the stimulus, relative to the biased (suppressed) perception in healthy controls (Dakin et al., 2005). It is hard to conceive how a generalized performance deficit would result in more accurate perception. These findings add to numerous studies that have utilized visual paradigms in which an abnormality in schizophrenia results in superior performance, compared to healthy subjects (Place and Gilmore, 1980; reviewed in Uhlhaas and Silverstein, 2005). We also note that the finding of reduced OSSS in schizophrenia (Yoon et al., 2009, 2010) presumably does not result from group differences in attentional lapses (Barch et al., 2012), as the frequency of these lapses would need to depend on the relative orientation of the center and surround stimuli to account for the observed differences between patients and control subjects.

Another example of the utility of visual psychophysics to address the generalized deficit confound in schizophrenia is that of Rokem et al. (2011). In this study, we employed an adaptation paradigm to measure selectivity for stimulus orientation in patients with schizophrenia and matched healthy controls (see **Figure 1D**). Participants viewed a high-contrast oriented adapting stimulus, followed by a post-adaptation probe in one of several different orientations relative to the adapter. Replicating previous results using this paradigm (Fang et al., 2005), we found that, for both patients and control subjects, adaptation elevated detection thresholds for probes having the same orientation as the adapter (0◦ offset). In addition, this elevation decreased systematically with increasing adapter-to-probe orientation offset, allowing the construction of a psychophysical tuning curve for orientation for each participant. To control for group differences in detection threshold that could have reflected generalized deficits, we fit a tuning curve model for each subject, normalized to that subject's threshold elevation at 0◦ offset. This procedure provides a measure of orientation tuning that does not depend on other aspects of performance such as cognitive control, attention, memory, etc. With this procedure, we found that subjects with schizophrenia have broader (less selective) orientation tuning. A similar approach could be adopted to explore other domains of selectivity for more complex stimuli (Schwarzlose et al., 2008).

#### **MAGNOCELLULAR AND PARVOCELLULAR SYSTEMS**

The magnocellular processing stream is one of the most extensively studied visual pathways in schizophrenia and one that illustrates the refined neurobiological inferences that may be gained from study of the visual system. In the early stages of visual processing, it is thought that there are at least two distinct streams: magnocellular (M) and parvocellular (P) (Livingstone and Hubel, 1988; Felleman and Van Essen, 1991; Sincich and Horton, 2005). M neurons have relatively large receptive fields (i.e., they respond to a wide range of visual field locations) and therefore have low spatial resolution. These neurons respond best to coarse visual features (low spatial frequencies; **Figure 1F**), and are highly responsive even when stimuli have low luminance contrast. Furthermore, M neurons show strong responses to stimulus onsets and offsets and other visual transients. In contrast, P neurons have relatively small receptive fields and respond best to fine spatial details (high spatial frequencies; **Figure 1F**). P neuronal responses also require relatively high luminance contrast and are relatively insensitive to transient onsets of stimuli. M cells are thought to preferentially provide input to cortical areas that process motion (Livingstone and Hubel, 1988; but see Sincich and Horton, 2005).

The existence of these complementary processing streams allows investigators to create stimuli and conditions that are preferentially processed by either the M or P pathway. Using this approach, a series of studies has demonstrated selective processing deficits in the M pathway in schizophrenia. For example, Butler et al. (2001) found that patients with schizophrenia have decreased responses evoked by M-type stimuli, as measured with EEG. Furthermore, abnormal processing of M-type stimuli has been shown to contribute to motion processing deficits in schizophrenia (Kim et al., 2006). A recent fMRI study demonstrated a selective reduction in visual responses in the occipital cortex of patients with schizophrenia to contrast-reversing, low spatial frequency, sinusoidal gratings, especially at low contrast, consistent with a selective deficit in the M pathway (Martínez et al., 2008).

These findings suggest that further study of the M pathway could provide valuable information about the molecular and cellular bases of information processing deficits in schizophrenia. For example, some authors have interpreted the similarity between suppressed visual responses in subjects with schizophrenia and those of animals treated with NMDA receptor antagonists (Kwon et al., 1992) as indicating that NMDA receptor dysfunction may underlie impairments in feedforward signal amplification in schizophrenia (Butler et al., 2005). However, it should be noted that the M pathway deficit hypothesis is not universally accepted (Slaghuis, 1998; Kéri et al., 2002; Skottun and Skoyles, 2007). For example, masking deficits are associated with functional abnormalities within areas involved in object processing that are thought to receive preferential input from the P pathway (Green et al., 2009b). Specifically, in a backward masking paradigm, patients with schizophrenia exhibited decreased fMRI BOLD responses in lateral occipital cortex, an area involved in object identification, but normal responses in early visual cortical areas. This finding is consistent with other studies that have found visual masking deficits in schizophrenia that cannot be readily accounted for by M pathway deficits (Herzog et al., 2004; Chkonia et al., 2010).

#### **FUTURE DIRECTIONS: GABA AND GLUTAMATE DEFICITS IN SCHIZOPHRENIA**

Some of the advantages of the visual system for schizophrenia research are illustrated by its potential utility in elucidating the relative contributions of inhibitory GABAergic and excitatory glutamatergic deficits in this condition. While multiple neurotransmitter systems are likely involved in schizophrenia, including dopamine (Davis et al., 1991), serotonin (Igbal and van Praag, 1995) and acetylcholine (Adams and Stevens, 2007), the GABA and glutamate systems are of particular interest because the interactions and balance of activity between these systems are fundamental phenomena underlying cortical information processing. Perturbation of this balance in schizophrenia may be a key pathophysiologic feature underlying cognitive and information processing deficits.

Post-mortem studies have consistently shown reductions in the concentration of the major synthetic enzyme for GABA within specific cortical interneuron subtypes, implying a deficit in GABAergic neurotransmission (Akbarian et al., 1995; Guidotti et al., 2000; Volk et al., 2000; Hashimoto et al., 2008). GABA release from these interneurons plays a critical role in controlling the activity of excitatory pyramidal neurons and in the generation of gamma-band synchrony (Whittington and Traub, 2003). As noted above, these coordinated oscillations are thought to be critical in supporting diverse processes, including higherorder cognition (Tallon-Baudry et al., 1998; Howard et al., 2003). However, the association between gamma-band synchrony and higher-order cognition can be confounded by gamma activity that is induced by microsaccades (Yuval-Greenberg et al., 2008). In schizophrenia, GABA deficits are thought to result in impairments in gamma-band synchrony, which in turn may be responsible for cognitive and information processing deficits (Lewis et al., 2005).

An alternative theory, the NMDA receptor hypofunction hypothesis, proposes that impairments in NMDA-type glutamate receptors are primary causal factors in schizophrenia. Specifically, it postulates that dysfunction of NMDA receptors on GABAergic interneurons leads to diminished GABAergic neurotransmission and, consequently, disinhibition and dysregulation of neural activity (Olney and Farber, 1995; Lewis and Moghaddam, 2006; Moghaddam and Javitt, 2012). Indirect supportive evidence for this hypothesis comes from animal studies in which induction of NMDA receptor hypofunction on GABAergic interneurons, either through pharmacologic (Homayoun and Moghaddam, 2007) or genetic (Belforte et al., 2010) manipulations in rodents, decreased excitatory drive onto GABAergic interneurons, and the resulting disinhibition caused an increase in spontaneous activity in excitatory pyramidal neurons.

In healthy humans, NMDA receptor antagonists such as PCP and ketamine can result in a behavioral syndrome that closely mimics the most salient clinical features of schizophrenia (Javitt and Zukin, 1991; Krystal et al., 1994). The disinhibition of excitatory activity that is hypothesized to occur in the initial stage of schizophrenia is thought to eventually lead to excitotoxic damage to cortical neurons and diminished excitatory drive (Olney and Farber, 1995). This may lead to compensatory homeostatic reductions in GABAergic tone in order to maintain the appropriate excitatory-inhibitory balance in cortical networks (Lewis et al., 2011). A more complete understanding of the relative contributions of glutamatergic and GABAergic deficits in schizophrenia would have obvious and important implications for the associated neural mechanisms of schizophrenia and for developing new interventions. However, as indicated above, there is extensive recurrent connectivity between excitatory and inhibitory neurons in cerebral cortex (for example, Ozeki et al., 2009), making it difficult if not impossible to manipulate activity in one system without eliciting secondary responses in the other.

# **DIRECT MRS MEASUREMENTS OF GABA AND GLUTAMATE CONCENTRATIONS**

Measurement of glutamate and GABA levels with MRS promises to be a powerful tool for dissociating the unique contribution of each neurochemical system in schizophrenia. This method allows non-invasive quantification of neurotransmitter levels in the brains of patients with schizophrenia, and unlike post-mortem studies, provides the opportunity to correlate measurements in the living brain with behavior in the same group of participants. Furthermore, MRS provides information about concentrations of neurotransmitters and other molecules in the native state, unaltered by pharmacological manipulations.

For the glutamate system, a large animal literature has documented activity-dependent changes in glutamate levels with invasive methods (Carder and Hendry, 1994; Qu et al., 2003), and MRS studies in humans have confirmed that dynamic increases in glutamate/glutamine (Glx) concentration occur in response to interventions designed to increase excitatory transmission (Mullins et al., 2005; Gussew et al., 2010; Maddock et al., 2011). Of direct relevance to visual system studies in schizophrenia is the recent demonstration at ultra high magnetic field that viewing of a contrast-reversing checkerboard increased visual cortical Glx concentration (Mangia et al., 2007).

Other studies have used MRS to document changes in GABA concentrations in response to cognitive, pharmacological, or other manipulations targeting the GABA system. For example, administration of vigabatrin, an inhibitor of GABA metabolism that increases GABA levels and is used in the treatment of epilepsy, caused dose-dependent changes in GABA levels (Verhoeff et al., 1999; Weber et al., 1999). In addition, motor cortical GABA MRS levels were reduced following a 30-min motor sequence learning paradigm thought to reduce tonic neural inhibition by decreasing GABA synthesis (Floyer-Lea et al., 2006). Finally, a number of studies have demonstrated strong correlations between GABA concentration and cognitive or perceptual functions (Edden et al., 2009; Sumner et al., 2010; Yoon et al., 2010; van Loon et al., 2013).

Taken together, these studies suggest that *in vivo* spectroscopic measurements of glutamate and GABA may index the underlying functional status of excitatory and inhibitory neural systems, respectively. Furthermore, the use of these non-invasive neuroimaging methods in conjunction with carefully selected visual paradigms and/or glutamatergic and GABAergic pharmacological agents can help dissociate excitatory and inhibitory deficits in schizophrenia. We employed this approach to document reduced visual cortical GABA levels as well as decreased behavioral OSSS in schizophrenia (Yoon et al., 2010). Across a sample of patients and control subjects, the amount of visual cortical GABA was highly predictive of OSSS, but no such correlation was observed for Glx and behavior (Yoon et al., 2010). Another application of GABA MRS comes from the study of receptor tyrosine kinase ErbB4, which is only expressed in GABAergic interneurons. A genetic variant of ErbB4 that is a risk factor for schizophrenia was shown to predict cortical GABA but not Glx concentrations (Marenco et al., 2011).

Despite the many advantages of MRS, the method is still being actively developed, and many issues remain unresolved. These include contamination of the measurement by other molecules and difficulties in obtaining reliable quantitative measurements of GABA concentrations, due to variability in GABA signal isolation as well as variability between individuals and between brain locations in the molecules used as baseline for comparison (see Mullins et al., 2013, for a recent review of these issues).

# **FUTURE DIRECTIONS: LOCAL CIRCUIT AND FEEDBACK MODULATION DEFICITS IN SCHIZOPHRENIA**

The visual system affords distinct advantages for experimental dissociation of local cortical circuit processing and feedback modulatory inputs from higher-order regions. In particular, manipulations of visual spatial attention allow comparison of neural responses to the identical visual stimulus when it is attended and when attention is directed elsewhere. Modulation of visual responses by spatial attention in healthy individuals has been shown in early visual cortex with fMRI (Gandhi et al., 1999; Somers et al., 1999), EEG (Martínez et al., 1999; Kelly et al., 2008), and MEG (Poghosyan and Ioannides, 2008). Moreover, sustained allocation of visual spatial attention to specific visual field locations increases fMRI signals in early visual cortical regions representing the attended locations, even in the absence of a visual stimulus (Kastner et al., 1999; Silver et al., 2007), thereby allowing a pure measure of top-down signals without any contribution from responses evoked by sensory stimulation.

Early visual cortical areas contain a continuous topographic map of visual field locations on the cortical surface. That is, a visual stimulus presented at a particular visual field location activates a corresponding location in each of these cortical areas, and the layout of visual field locations reflects the two-dimensional representation of visual field locations in the retina (retinotopic maps; **Figure 3A**).

**FIGURE 3 | An example of application of fMRI and a spatial attention paradigm to study impaired neural mechanisms of surround**

**suppression in schizophrenia. (A)** Functional MRI data displayed on a flattened occipital cortical surface from an example subject. The circular grating stimulus is divided into two regions: the annulus and the "surround" (in this example, the "surround" includes visual field locations both inside and outside the annulus). An annulus was used as the "center" stimulus instead of a central circle or patch because it is difficult to definitively map retinotopic early visual cortical areas with fMRI for locations close to the fovea (central vision; where the gaze is directed). In these localizer data, subjects viewed the annulus stimulus in block alternation with the surround stimulus. The fact that early visual cortical areas (in this example, V1, V2, and V3) contain retinotopic maps of the visual field means that the responses to the annulus and surround portions of the stimulus are spatially separated on the cortical surface. Red indicates regions that exhibited a stronger response to the annulus than to the surround, and blue indicates regions that responded more to the surround than to the annulus. The foveal confluence

is where the visual field maps in areas V1, V2, and V3 converge. Isoeccentricity lines form concentric arcs around the foveal confluence. This type of localizer experiment can be used to objectively define regions of interest in each visual cortical area in each subject that represent either the annulus or surround. **(B)** Neural circuit model of surround suppression of responses to grating stimuli in cortical area V1. Neurons representing the annulus or surround are largely spatially segregated within the V1 retinotopic map but project to common target neurons in higher-order visual cortical areas. These areas send feedback projections back to V1 that activate local GABAergic inhibitory neurons. The resulting inhibition suppresses responses in V1 neurons that represent annulus locations. **(C)** Possible circuit-level abnormalities contributing to reduced surround suppression in schizophrenia. Left, reduction in the number and/or efficacy of GABAergic neurons within V1 could decrease inhibition of the pyramidal neurons. Right, impaired feedback modulation to V1 could result in reduced excitation of GABAergic interneurons. Experimental manipulations of spatial attention can help distinguish between these possible mechanisms.

More than 20 distinct cortical areas have now been identified in the human brain based on their topographically-organized visual field maps (Silver and Kastner, 2009; Wandell and Winawer, 2011).

The existence of topographically-organized visual field maps in the visual system, combined with appropriate manipulations of spatial attention, allows fMRI responses to attended and unattended visual field locations to be simultaneously measured in independent sets of voxels within a given cortical area. Feedback modulation of fMRI responses in early visual cortex by spatial attention is spatially-specific: it is restricted to those portions of the topographic visual field maps that represent the attended portions of the visual field (Tootell et al., 1998; Silver et al., 2007). One potential source of these top-down spatial attention signals is intraparietal sulcus areas such as IPS1 and IPS2 (Silver et al., 2005). These areas contain topographic maps of the locus of spatial attention and transmit spatially-selective top-down attention signals to early visual cortex (Lauritzen et al., 2009; Greenberg et al., 2012).

The utility of using spatial attention manipulations to separately and simultaneously measure responses to attended and unattended stimuli can be extended to other paradigms in which stimuli are presented in different visual field locations. For example, attention manipulations can be included in fMRI studies of surround suppression that retinotopically identify portions of visual cortical areas that represent visual field locations corresponding to either the center or surround (**Figure 3A**).

The combination of OSSS and spatial attention holds great promise in characterizing local circuit vs. feedback modulation abnormalities in schizophrenia. The fact that OSSS is selective for stimulus orientation suggests an early visual cortical locus, and finding a neural correlate of reduced OSSS in early visual cortex in schizophrenia would establish an excellent experimental model in a portion of the brain that, relative to other brain regions, is well understood.

In addition, comparison of responses to the suppressed center stimulus when it is attended vs. when attention is drawn elsewhere (for example, maintaining attention at central fixation by requiring subjects to perform a difficult task there) provides a highly controlled measure of feedback modulation of visual cortical activity by spatial attention in schizophrenia. The use of fMRI or other neurophysiological techniques uniquely allow measurement of responses to a stimulus when it is being ignored. Analogous measures of the impact of unattended stimuli are very difficult to obtain with behavioral methods, because requiring a subject to make a behavioral response to a stimulus necessarily requires the allocation of some attentional resources (typically an unknown amount) to that stimulus.

Finally, combining spatial attention manipulations with OSSS is likely to provide insights regarding the causes of reduced OSSS in schizophrenia at the local cortical circuit level. For grating stimuli, the neural substrates of surround suppression are thought to include feedback projections from higher-order visual cortical areas to area V1 (Angelucci and Bressloff, 2006; Nassi et al., 2013), with these feedback connections suppressing V1 pyramidal cell responses through local GABAergic interneurons (Ozeki et al., 2004, 2009; Schwabe et al., 2006) (**Figure 3B**). In this framework, reduced OSSS in schizophrenia could be due to impaired feedback modulation of V1 and/or diminished GABAergic transmission (Yoon et al., 2010) (**Figure 3C**). Spatial attention is a wellcontrolled and thoroughly studied means of manipulating topdown inputs, and single-unit studies indicate that directing spatial attention to the center vs. surround modulates surround suppression of neural responses in cortical area V4 (Sundberg et al., 2009). In fMRI studies of schizophrenia, finding reduced OSSS that is independent of whether the center stimulus is attended or not would be evidence of a local cortical circuit mechanism. On the other hand, if reduced OSSS in schizophrenia is only evident when the center stimulus is attended, this would implicate feedback modulation by spatial attention.

# **CONCLUSION**

While contributions from multiple disciplines and experimental approaches will likely be required to overcome the formidable challenges in elucidating the neural mechanisms of cognitive and information processing deficits in schizophrenia, the study of the visual system has a number of distinct advantages in this area. This review has highlighted some of the most productive lines of research within the rapidly growing body of literature on visual processing in schizophrenia, illustrating the diversity of visual processes that have been studied as well as the sophisticated methods available in the vision sciences. We also discussed several key factors that make the visual system such an appealing model system for the discovery of neural mechanisms. The convergence of the well-developed body of knowledge in structurefunction relationships in the visual system, the conservation of the functional architecture across species, the preservation of basic local circuit architecture across neocortical regions, and the availability of quantitative methods to control for generalized deficits allow for inferences at a level of detail and specificity that is usually impossible in other neural systems. In the near future, the combination of vision science paradigms with modern neuroimaging methods may allow us to test some of the most compelling hypotheses on the neural origins of cognitive and information processing deficits in schizophrenia.

# **ACKNOWLEDGMENTS**

Jong H. Yoon was supported by grants from the National Institutes of Health (K08 MH076174) and NARSAD (Young Investigator Award). Ariel Rokem was supported by a postdoctoral NRSA award from the National Institutes of Health (F32 EY022294). Summer L. Sheremata was supported by a training grant from the National Institutes of Health (T32 EY007043).

# **REFERENCES**


synchronization between monkey visual areas. *Neuron* 75, 875–888. doi: 10.1016/j.neuron.2012.06.037


top-down processing. *Nat. Rev. Neurosci.* 2, 704–716. doi: 10.1038/35094565


paradigms from CNTRICS. *Schizophr. Bull.* 35, 163–181. doi: 10.1093/schbul/sbn156


Time-resolved functional 1H MR spectroscopic detection of glutamate concentration changes in the brain during acute heat pain stimulation. *Neuroimage* 49, 1895–1902. doi: 10.1016/j.neuroimage.2009.09.007


memory load in humans. *Cereb. Cortex* 13, 1369–1374. doi: 10.1093/cercor/bhg084


resulting from general performance deficiencies in schizophrenia. *J. Abnorm. Psychol.* 110, 15–30. doi: 10.1037/0021-843X.110.1.15


processing and sensory-motor integration for smooth pursuit eye movements. *Annu. Rev. Neurosci.* 10, 97–129. doi: 10.1146/annurev. ne.10.030187.000525


P. D., Guilfoyle, D. N., et al. (2008). Magnocellular pathway impairment in schizophrenia: evidence from functional magnetic resonance imaging. *J. Neurosci.* 28, 7492–7500. doi: 10.1523/JNEUROSCI.1852-08.2008


between excitation and inhibition underlying size tuning and contextual response modulation in the cat primary visual cortex. *J. Neurosci.* 24, 1428–1438. doi: 10.1523/JNEUROSCI.3852-03.2004


8:e68090. doi: 10.1371/journal. pone.0068090


*Abnorm. Psychol.* 107, 49–62. doi: 10.1037/0021-843X.107.1.49


*Rev.* 80, 1–52. doi: 10.1037/ h0033872


maps in the human brain. *Vision Res.* 51, 718–737. doi: 10.1016/j.visres.2010.08.004


Green, M. F. (2005). Event-related gamma activity in schizophrenia patients during a visual backward-masking task. *Am. J. Psychiatry* 162, 2330–2336. doi: 10.1176/appi.ajp.162.12.2330


specific surround suppression. *J. Neurosci.* 30, 3777–3781. doi: 10.1523/JNEUROSCI.6158-09.2010


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 15 May 2013; accepted: 09 September 2013; published online: 31 October 2013.*

*Citation: Yoon JH, Sheremata SL, Rokem A and Silver MA (2013) Windows to the soul: vision science as a tool for studying biological mechanisms of information processing deficits in schizophrenia. Front. Psychol. 4:681. doi: 10.3389/fpsyg. 2013.00681*

*This article was submitted to Psychopathology, a section of the journal Frontiers in Psychology.*

*Copyright © 2013 Yoon, Sheremata, Rokem and Silver. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The coherent organization of mental life depends on mechanisms for context-sensitive gain-control that are impaired in schizophrenia

# *William A. Phillips 1,2\* and Steven M. Silverstein3*

*<sup>1</sup> Psychology, School of Natural Sciences, University of Stirling, Stirling, UK*

*<sup>2</sup> Theoretical Neuroscience, Frankfurt Institute of Advanced Studies, Frankfurt, Germany*

*<sup>3</sup> Rutgers University School of Biomedical and Health Sciences, University of Medicine and Dentistry of New Jersey, Piscataway, NJ, USA*

#### *Edited by:*

*Michael Green, University of California, Los Angeles, USA*

#### *Reviewed by:*

*Yue Chen, Mclean Hospital, USA Jonathan K. Wynn, University of California, Los Angeles, USA Joshua Kantrowitz, Columbia University, USA*

#### *\*Correspondence:*

*William A. Phillips, Psychology, School of Natural Sciences, University of Stirling, Cottrell Building, Stirling FK9 4LA, UK e-mail: wap1@stir.ac.uk*

There is rapidly growing evidence that schizophrenia involves changes in context-sensitive gain-control and probabilistic inference. In addition to the well-known cognitive disorganization to which these changes lead, basic aspects of vision are also impaired, as discussed by other papers on this Frontiers Research Topic. The aim of this paper is to contribute to our understanding of such findings by examining five central hypotheses. First, context-sensitive gain-control is fundamental to brain function and mental life. Second, it occurs in many different regions of the cerebral cortex of many different mammalian species. Third, it has several computational functions, each with wide generality. Fourth, it is implemented by several neural mechanisms at cellular and circuit levels. Fifth, impairments of context-sensitive gain-control produce many of the well-known symptoms of schizophrenia and change basic processes of visual perception. These hypotheses suggest why disorders of vision in schizophrenia may provide insights into the nature and mechanisms of impaired reality testing and thought disorder in psychosis. They may also cast light on normal mental function and its neural bases. Limitations of these hypotheses, and ways in which they need further testing and development, are outlined.

**Keywords: cognitive coordination, context-sensitivity, gain-control, perceptual grouping, coherence, vision, schizophrenia, cortical computation**

# **INTRODUCTION**

Schizophrenia is well-known to be associated with disorganized and incoherent patterns of thought and behavior, but it is also associated with perceptual impairments that are less well-known because they are less obvious to casual observation. Substantial perceptual impairments have been rigorously demonstrated using many different experimental paradigms, however, and may constitute a core component of the illness. Furthermore, as so much is known at both psychological and neurobiological levels about the perceptual processes involved it may be easier to gain a deep understanding of those impairments than it is of the more obvious higher-level symptoms. If, as we argue here, perceptual and higher cognitive impairments arise from the same underlying pathophysiology, then this will show how perceptual impairments can provide a window on the disorder in general. The pathophysiology that we hypothesize to underlie many aspects of schizophrenia is centered on context-sensitive gain-control. Therefore, before reviewing evidence implicating it in schizophrenia, we provide an in-depth review of the arguments and evidence for its central role in normal mental life. Furthermore, we claim that schizophrenia research casts new light on contextsensitive gain-control, and assessment of that claim requires adequate knowledge of current views concerning its functions and mechanisms.

Put most simply, the perspective that we have been developing since the mid 1990s proposes that the coherent organization of mental life depends upon coordinating neural interactions of context-sensitive gain-control that amplify relevant signals and suppress irrelevant signals. Thus, disorganization and incoherence occur when those coordinating interactions are impaired, as in schizophrenia. Several other researchers in basic neuroscience have also argued that gain modulation, or gain-control, is a major principle underlying brain function (e.g., Salinas and Sejnowski, 2001; Chance et al., 2002). Gain-control changes the rate at which a neuron's output increases with the strength of the driving inputs to which it is selectively tuned. This suggests that neurons have two classes of input: one specifying selectivity and the other controlling gain. There is evidence that neuronal selectivity is specified by synaptic inputs that are few but strong whereas gain-control depends on many inputs that are individually weak (Lee and Sherman, 2010; Phillips et al., 2010). Gain-controlling inputs include those from the classical neuromodulators that have long been associated with the psychoses. Our theory also emphasizes gain-controlling interactions within and between glutamatergic and GABAergic neurons, however, because it is only they that convey the detailed cognitive content whose coherence is compromised in schizophrenia. It is not the inputs from sub-cortical neuromodulatory systems that convey that content, even though they do have modulatory effects on the intra-cortical interactions that do. Furthermore, there is ample evidence that impairments of the glutamatergic and GABAergic systems are central to schizophrenia.

Previous papers have distinguished gain-control from dynamic Gestalt grouping or "integration" (e.g., Butler et al., 2008), but here we use the phrase context-sensitive gain-control to cover both, because grouping can be seen as a form of contextsensitive gain-control on a fast time-scale that may use essentially the same mechanisms as other coordinating interactions (Phillips et al., 2010). The close relations between grouping and other forms of gain-control are clear in the similar dependence of contour integration and flanker facilitation on collinearity. Furthermore, there is clear evidence for deficits in both contour integration (Silverstein et al., 2009) and flanker facilitation (Must et al., 2004) in schizophrenia.

As context-sensitivity is central to our hypotheses we must also make clear what we mean by "context." Some influential researchers identify "context" with information about time and place, or with information in working memory (WM) (e.g., Cohen and Servan-Schreiber, 1992; Gonzalez-Burgos and Lewis, 2008; Lewis, 2012). From our perspective those conceptions are far too narrow. Instead, we define context as any information that is used to modulate the strength with which a pyramidal cell, or local cortical circuit, transmits the information to which it is selectively sensitive. Thus, inputs arising from concurrent sensory input (e.g., Schwartz et al., 2007), as well as those arising from many other ongoing activities, can also control gain.

We do not provide a comprehensive review of the vast amount of empirical and theoretical research relevant to all the issues discussed in this paper. Our goal is to cite representative examples of the relevant findings. The perspective taken here overlaps in various ways and to various extents with several other theories. We do not claim priority for any of its essential components. One previous theory that we should relate it to, however, is our own. The central hypotheses examined are similar to those that we proposed ten years ago (Phillips and Silverstein, 2003), except that we now emphasize a greater variety of possible functions and mechanisms. Then we emphasized contextual disambiguation, now we emphasize various other possible functions. Then we emphasized NMDA receptors (NMDARs) for glutamate (the main excitatory neurotransmitter in cortex) as the main mechanism for gain-control. Now we emphasize several other possibilities also, including various classes of inhibitory interneuron and various intracellular mechanisms.

Thus, motivated by the above considerations, we examine five central hypotheses in this paper. First, context-sensitive gain-control is fundamental to brain function and mental life. Second, it occurs in many different regions of the cerebral cortex of many different mammalian species. Third, it has several computational functions, each with wide generality. Fourth, it is implemented by various mechanisms at cellular and circuit levels. Fifth, impairments of context-sensitive gain-control produce many of the well-known symptoms of schizophrenia and change basic processes of visual perception. To the extent that schizophrenia arises from widespread impairments of contextsensitive gain-control, insights gained from studying it can also inform our basic understanding of brain function and mental life in general.

These hypotheses are guided by various theoretical attitudes and working assumptions. First, our perspective is resolutely multi-disciplinary. Second, we assume that there are two-way causal interactions between macroscopic events at a psychological level and microscopic events at a cellular level. Third, we assume that impairments at synaptic and local circuit levels may be subtle, such as those related to differences between subtypes of cortical neuron or synaptic receptor. For example, there are various subtypes of NMDAR with the 2A and 2B subtypes being the most common in cerebral cortex. Small parametric differences in their biophysical properties suggest that the 2A subtype is better suited to operate on signals with high temporal precision. In addition, there is evidence that adolescence is associated with dramatic changes in NMDAR distribution (Wang and Gao, 2009), including a switch from the 2B to the 2A subtype (Liu et al., 2004). This suggests that disorders with an adolescent onset, as is the case with psychosis in schizophrenia, might be related to such subtle differences.

The remainder of this paper is organized into three sections. Section Context-Sensitive Gain-Control Plays a Central Role in Brain Function and Mental Life outlines our theory of contextsensitive gain-control, and reviews evidence concerning its role in normal cognition and brain function. Central aims of this section are to clarify our understanding of both its functions and its mechanisms. Section The Functions and Mechanisms of Context-Sensitive Gain-Control are Impaired in Schizophrenia reviews evidence that both the functions and the mechanisms are impaired in schizophrenia. Section Difficulties for the Hypotheses Proposed and Major Aspects to be Further Developed outlines some difficulties faced by the theory, and suggests ways in which it can be further tested and developed.

## **CONTEXT-SENSITIVE GAIN-CONTROL PLAYS A CENTRAL ROLE IN BRAIN FUNCTION AND MENTAL LIFE**

Contextual inputs to a local neural processor that arise from other ongoing activity in the system are in effect implicit predictions about the current state of the activity of that local processor. Mental life as a whole would be fully coherent if all local activities were as predicted by other ongoing activity. Though that state is an unrealistic ideal, we do depend on our ability to make sufficiently accurate inferences about distal things from proximal signals, about our own mental activities and intentions, and about the likely consequences of possible actions.

The centrality of unconscious inference to perception was emphasized by Helmholtz more than a 100 years ago. This perspective has now been extended into several prominent theories of cognition and brain function that are often referred to collectively as the "Bayesian brain" (e.g., Feldman, 2001; Purves et al., 2001; Körding and Wolpert, 2004; Yuille and Kersten, 2006; Friston and Stephan, 2007; Friston, 2010; Brown and Friston, 2012). The central tenet of these theories is that interpretation of sensory input depends upon the beliefs about the world that have been acquired from prior experience. Many psychophysical, neurobiological, and computational studies support this perspective (Clark, 2013). These theories usually imply a central role for context-sensitive gain-control, and they provide the basis for several recent and influential theories of schizophrenia (e.g., Friston, 2005; Corlett et al., 2007, 2010; Fletcher and Frith, 2009; Synofzik et al., 2010; Seth et al., 2011).

Our theory agrees with these main-stream theories in emphasizing probabilistic inference, but it differs in important ways. First, instead of relying primarily on "Bayes theorem," we build on the foundations laid by the American statistical physicist Jaynes (2003). The many riches in Jaynes's analysis have been ignored by most neuroscientists and psychologists, but one notable exception is the neurophysiologist Fiorillo (2012) who sees the implications of Jaynes's logic as requiring radical changes in the current consensus concerning the "Bayesian brain." We agree. The relevance of Jaynes's work to our perspective has therefore been examined in depth elsewhere (Phillips, 2012). Second, we assume that the requirements for optimal inference can rarely be fully met. For example, between what options must higher cortical regions choose when perceiving things? What prior data are relevant? On what should likelihoods be conditioned? These are not questions to which we expect optimal answers, even though, given answers to them, the principles of Jaynes's logic specify the optimal way to draw inferences from them. Third, context operates via the likelihoods, not via the prior probability, as shown formally by Kay and Phillips (2010). Fourth, we are not committed to the view that prediction error is the common currency of feed-forward signals between cortical regions as proposed by Rao and Ballard (1999). Instead, we assume that what is fed-forward from any level to the next is information about the current interpretations or predictions as inferred at that level. Fifth, inferences depend on contextual predictions, and hierarchical Bayesian theories emphasize feed-back signals from higher regions as the source of the predictions. We also emphasize predictions from lateral contextual interactions both within and between cortical regions, however, and this applies to all levels of the system as clearly demonstrated by the ubiquitous distribution of the mechanisms by which contextual interactions are implemented. All this is fully compatible with amplification of attended signals (Spratling, 2008; Spratling et al., 2009), and also with amplification of feed-forward signals that contradict strong predictions.

Our previous claims concerning the role of context-sensitive gain-control in cognitive coordination (Phillips and Silverstein, 2003) have been rigorously formalized using information theoretic concepts (e.g., Kay et al., 1998; Kay and Phillips, 2010), and are founded on cognitive, neurobiological, and clinical evidence, including that from visual psychophysics. The ubiquity of local ambiguity in visual perception and its resolution by context is well-established by many reviews of neurobiological and psychophysical studies (e.g., Phillips and Singer, 1997; Phillips and Silverstein, 2003; Butler et al., 2008; von der Malsburg et al., 2010). This applies to the early stages of visual processing, and also to the higher levels of perceptual interpretation, such as in the dependence of object recognition on scene context (Bar, 2004). Taking these well-established findings as a given, the perspective outlined here formalizes conceptions of context-sensitive gain-control rigorously in terms of computations that neural systems can perform, and relates them to detailed neurobiological mechanisms at both intracellular and local-circuit levels. These

computations are described in formal terms to show that they can in principle deliver the capabilities claimed for them, and that they have sufficient generality to underlie many different domains of cognition.

Several prominent theorists have previously argued that gain modulation is a major computational principle underlying brain function (e.g., Salinas and Their, 2000; Salinas and Sejnowski, 2001; Chance et al., 2002), and some of its many computational functions will be listed in sub-section Context-Sensitive Gain-Control has Several Computational Functions. Within computational neuroscience gain modulation, or gain-control, is usually defined as a non-linear change in the response amplitude of a neuron that does not change its receptive field (RF) selectivity, i.e., its tuning function. Mathematically this has general utility because a population of responses to any RF variable, *x*, modulated by any context, *y* provides a basis set from which any function of *x* and *y* can be computed, and in many relevant cases it can be computed simply as a linear weighted sum (Salinas and Their, 2000; Salinas and Sejnowski, 2001). Salinas and Their (2000) note that to some researchers it can seem difficult to draw the line between selectivity and modulation, however, and that would greatly weaken any theory using it as a fundamental distinction. Fortunately, this crucial distinction is clear to others, such as Lamme (2004) whose extensive electrophysiological findings on contextual modulation led him to the conclusion that it bears no relation to the neuron's RF properties and is mediated by mechanisms far removed from those that shape and tune the local RF. Furthermore, the distinction can be rigorously formulated using information theoretic concepts (Smyth et al., 1996). Primary driving RF input is that determining the variables and values to which the neuron is selectively tuned, and about which it thus transmits information. Gain-control changes the rate at which the neuron's output increases with the strength of the driving inputs to which it is selectively tuned, but without fundamentally changing that selectivity. In short, selective driving inputs are both necessary and sufficient to produce an output signal; contextual inputs are neither necessary nor sufficient.

#### **NEUROBIOLOGICAL EVIDENCE THAT CONTEXT-SENSITIVE GAIN-CONTROL OCCURS IN MANY DIFFERENT CORTICAL REGIONS**

There is ample evidence that context-sensitive gain-control occurs within the mammalian brain. Its wide distribution throughout the cortex is shown by evidence from single units, multiple units, local-field potentials, intra-cortical potentials, and macroscopic neuroimaging [see reviews by Phillips and Singer (1997); Phillips and Silverstein (2003); Lamme (2004); Schwartz et al. (2007); Salinas (2009); Lee and Sherman (2010); Feldman and Friston (2010); von der Malsburg et al. (2010)]. Studies using recent and revolutionary optogenetic techniques to control the activity of cortical cells also show that context-sensitive gaincontrol occurs in various cortical regions. They also provide clear evidence on the mechanisms by which gain-control is achieved, and so are reviewed in sub-section There are Various Local-Circuit and Cellular Mechanisms for Context-Sensitive Gain-Control where possible mechanisms are discussed. In addition to all the physiological evidence, common anatomical features of the canonical cortical circuit also suggest that gain-control is a general principle of cortical computation (Douglas and Martin, 2007, 2008).

Interpretation of these electrophysiological and anatomical findings has been strengthened by many computational studies of the role of context-sensitivity and gain-control in perceptual and higher cognitive functions. Examples include studies by Huang and Grossberg (2010) in learning and visual search, and many others reviewed by Schwartz et al. (2007) in relation to the perception of orientation. Context-sensitive gain-control is central to the computational model by which Schwartz et al. (2009) account for dynamic Gestalt grouping, the effects of context, and their dependence on natural scene statistics, all of which have been observed in visual cortex. Their model uses normalization in the form of divisive gaincontrol, and they argue that it is relevant to various levels of the visual system. Furthermore, contextual disambiguation and the dynamic grouping of coherently related elements may be of even greater importance to higher cognitive functions, such as language. Our working assumption is therefore that, as argued by Phillips and Singer (1997), context-sensitive gaincontrol provides a common foundation for cortical computation in general.

Evidence for a form of gain modulation that combines retinal and gaze signals multiplicatively was first observed in singleunit recordings in neurons of the parietal cortex of the macaque monkey, and computational studies showed that it could provide a basis for converting the position of stimuli relative to the retina into position relative to the head (Andersen et al., 1985). Since then several other coordinate transformations that could also be based on multiplicative gain modulation have been seen in other cortical areas (e.g., Galletti and Battaglini, 1989; Salinas and Sejnowski, 2001). This is widely assumed to be yet another form of context-sensitive gain-control, and that assumption will be reconsidered after our review of schizophrenia-related impairments because they cast new light on it.

#### **CONTEXT-SENSITIVE GAIN-CONTROL HAS SEVERAL COMPUTATIONAL FUNCTIONS**

Context-sensitive gain-control has several computational functions; each with wide generality. Here we simply list some of the most well-known, making no attempt to review the substantial body of research available on each.

First, contextual disambiguation is one of its main functions. This could be achieved by multiplicatively increasing the gain on interpretations that are coherently related to the context and reducing the gain on those that are not. Examples of this include the enhancement of low-contrast edge detection by collinear flankers (Polat and Sagi, 1993), sensitivity of object recognition to scene context (Bar, 2004), word-sense disambiguation, and many other examples. Our broad conception of contextual disambiguation includes coordination of multiple distinct probabilistic decisions so that they form a coherent whole. An example of this at the level of object perception is the interpretation of ambiguous figures, such as the duck-rabbit figure or Necker cube. When perception switches between alternative interpretations it does so as a whole, implying that all the distinct decisions that this involves are coordinated by some form of context-sensitive gain-control that operates so as to maximize coherence over the whole figure that is being interpreted (Klemm et al., 2000).

Second, divisive normalization is another function of gaincontrol that has been described as a canonical computation. This has various uses from low levels of sensory processing to high levels of cognition such as value encoding (Carandini and Heeger, 2012). It includes surround suppression (Heeger, 1992; Simoncelli and Schwartz, 1999), invariant object recognition (Kouh and Poggio, 2008), the reduction of redundancy (Schwartz and Simoncelli, 2001), and various other ways of producing efficient codes. Recent neurophysiological findings show that input normalization by feedforward inhibition can expand the dynamic range of cortical activities by enabling populations of pyramidal cells to remain sensitive to weak inputs without saturating in response to stronger inputs (Pouille et al., 2009). Within the theory of Coherent Infomax, which provides the foundations on which our current hypotheses are built, the driving RF is equivalent to the driving summation field in normalization theory. The suppressive field in normalization theory is included within the contextual field (CF) that is an essential component of the Coherent Infomax theory. For an outline and peer-evaluation of that theory see Phillips and Singer (1997); for an outline and evaluation of its relevance to schizophrenia see Phillips and Silverstein (2003).

Third, context-sensitive gain-control can also play a role in dynamic Gestalt grouping. Grouping, sometimes referred to as "integration," can be treated as a separate class of functions quite distinct from gain-control (e.g., Butler et al., 2008), but here we include it within a broad conception of gain-control for the reasons noted in section Introduction. Lamme (2004), who has studied contextual modulation extensively, argues that perceptual grouping is one of its main functions. Furthermore, there is much evidence that gain-control on a fast time-scale so as to synchronize coherent subsets could provide the basis for many cognitive functions including Gestalt figural organization (von der Malsburg et al., 2010). Finally, as we will show below, dynamic Gestalt grouping depends upon some of the same mechanisms as other forms of context-sensitive gain-control, such as fast-spiking inhibitory interneurons.

Fourth, context-sensitivity contributes to object and face recognition because the probability of seeing any given object or face depends so strongly upon the context (Bar, 2004). It may also contribute to the invariance of object recognition because normalization can be used to compute outputs that are insensitive to irrelevant stimulus dimensions (Salinas, 2009).

Fifth, selective attention also requires gain-control because it enhances the selected signals and suppresses the irrelevant signals. We therefore assume that the context that controls gain includes attention, which is in harmony with both the biasedcompetition theory (Desimone and Duncan, 1995), and its recent re-interpretation as a form of divisive normalization (Reynolds and Heeger, 2009).

Sixth, context-sensitive gain-control can produce efficient codes by using predictions to suppress the feed-forward transmission of signals that are highly probable, and thus not informative. Predictions are often assumed to be computed using hierarchical Bayesian inference (e.g., Lee and Mumford, 2003), and that possibility has now been developed into several highly influential theories. It may seem that such predictive coding theories are in conflict with the biased-competition theory of selective attention because they imply the suppression of predicted data, rather than its enhancement. It has been shown computationally that predictive coding and biased-competition are in principle compatible, however. This was done by combining them in a single model in which prediction-error processing occurs within, rather than between, cortical regions. Selective attention can then modulate those signals, so as to enhance, rather than suppress, the selected interpretations (Spratling, 2008; Spratling et al., 2009). Recent developments of that model show that by combining both driving and modulating inputs to the local processors it can account in detail for many well-established neurophysiological and psychophysical phenomena, including surround suppression, contour integration, predictive coding, and selective attention (Spratling, submitted).

Seventh, a neural network model has shown that contextual modulation can be used to select one of a number of possible arbitrary mappings from sensory stimuli to motor actions by controlling gain, thus helping to explain how higher organisms can rapidly and flexibly adapt their actions to current conditions (Salinas, 2004). Though that model is concerned with the selection of motor commands, the same computations could apply equally well to the selection of inner percepts and thoughts as assumed by the closely related theory of Coherent Infomax. Though these two theories were developed independently, they use essentially the same mathematical function to specify how the gain of the response to driving inputs is modulated by the context. Both theories are strengthened by this convergence because each provides further grounds for supporting the other.

Finally, another possible function is coordinate transformation, which was one of the first uses of gain modulation for which there was both empirical and theoretical evidence. This has been widely assumed to be a paradigmatic example of gain-control in general (e.g., Salinas, 2009), but the validity of that assumption remains to be determined.

It may also be possible to relate these basic computational functions of context-sensitive gain-control to more subjective aspects of human conscious experience. One recent development suggesting how that might be done is a theory of interoceptive inference which offers a unified account of emotion, the sense of presence, and the sense of agency (Seth et al., 2011), all of which are impaired in schizophrenia (e.g., Hauser et al., 2011) and can be impaired by drugs that block (modulatory) activity at NMDARs (e.g., Moore et al., 2013). By analogy with predictive coding theories of visual perception, interoceptive inference is hypothesized to involve a hierarchy of top-down predictions that guide the interpretation of bottom-up interoceptive signals. The subjective sense of the reality of the self and of the external world, referred to as conscious "presence," is hypothesized to depend on the successful suppression of interoceptive signals by precise top-down predictions (Seth et al., 2011). Similarly, the subjective sense of agency is hypothesized to arise from precise predictions of the sensory consequences of actions, as proposed by Fletcher and Frith (2009). The theory of (Seth et al., 2011) synthesizes much of the relevant phenomenology, neurobiology, and psychopathology, and the precision of prediction error signals plays a key role in their theory. This is optimized by using context to control the gain of prediction error units. Seth et al. emphasize the role of the classic neuromodulators in doing this, and dopamine in particular, but more locally-specific coordinating interactions must also play a role. Their theory is particularly relevant here because it depends on the modulation of precision by gain-control, and explicitly shows how impaired gain-control could produce positive symptoms of psychosis. It is important to note that although Seth et al. emphasize feed-forward transmission of prediction-errors, rather than of the inferences used to make the predictions, that is not essential to predictive inference as explained in our above discussion of predictive coding.

## **THERE ARE VARIOUS LOCAL-CIRCUIT AND CELLULAR MECHANISMS FOR CONTEXT-SENSITIVE GAIN-CONTROL**

There are various mechanisms for controlling gain within the cortex (e.g., Salinas, 2009; Silver, 2010). There is no simple one-to-one mapping between these mechanisms and the various functions of gain-control because one mechanism may contribute to more than one function and one function may be performed by more than one mechanism. Different mechanisms are suited to different roles, however, because they collect contextual information from very different sources, operate on very different time-scales, and vary greatly in the distribution of their effects, with some exerting gain-control that is highly local while others have widely distributed effects.

The simplest way in which pyramidal cells could increase the gain of other pyramidal cells so as to amplify coherent activities is via direct connections between them. It is likely that such a mechanism is used because it is the fastest and most energy efficient. In addition, it requires the transmission of a great deal of information and about 75% of all cortical connections are between pyramidal cells (Braitenberg and Schuz, 1991). Furthermore, NMDARs, which provide the means by which such connections can control gain (Phillips and Silverstein, 2003), are highly expressed on pyramidal cells. Recurrent excitation mediated by NMDARs also contributes to sustained neuronal firing, which is a potential substrate for WM (Gonzalez-Burgos and Lewis, 2008). Finally, a crucial role for direct gain-controlling interactions between pyramidal cells is shown by Self et al. (2012) using the well-established phenomenon of figure-ground modulation. This was almost abolished when NMDARs were blocked in V1, whereas the purely feedforward component of pyramidal cell response was largely unaffected. Conversely, blocking AMPA receptors did not affect figure-ground modulation, but did greatly reduce the feedforward component. Direct NMDAR-mediated interactions between pyramidal cells are therefore likely to be a widely used mechanism for controlling gain so as to amplify coherently related activities.

Recurrent connection between pyramidal cells requires tight inhibitory control to prevent runaway excitation, however, and this is provided by inhibitory interneurons. Much is now known about their role in shaping cortical activity (Isaacson and Scanziani, 2010), and they play a central role in various gaincontrol mechanisms. This is easy to understand intuitively: as inhibitory mechanisms for suppressing activity must be present it is likely that evolution has used them to control gain. It has recently been shown that this is so by combining advanced transgenic and optogenetic techniques with classical recording methods. Using optogenetic techniques experimenters can control the activity of genetically specified subtypes of cortical neuron in specified cortical layers of awake behaving animals, and they can do so with millimeter and millisecond precision. Studies using these techniques show that different classes of inhibitory interneuron provide mechanisms for different forms of context-sensitive gain-control. One major class, referred to as PV interneurons, are those expressing parvalbumin (PV), a low-weight protein involved in various physiological processes, including neuronal signaling. They have been identified with chandelier and basket cells, which are fast-spiking local-circuit inhibitory interneurons with synapses on perisomatic parts of pyramidal cells. Optogenetic studies show that under natural conditions PV interneurons can either amplify or suppress the gain of pyramidal cell activity (Atallah et al., 2012). They show that PV interneurons control the gain of the response of layer 2/3 pyramidal cells in an essentially simple way. Optogenetically suppressing PV interneuron activity increased layer 2/3 pyramidal cell activity multiplicatively by a factor of 1.2 and added a constant amount. Optogenetically activating PV interneurons decreased pyramidal cell activity divisively by a factor of 1.4 and subtracted a constant amount (Atallah et al., 2012). Furthermore, small changes in PV interneuron-mediated inhibition can lead to robust changes in the gain of pyramidal cell response without having any major impact on the selectivity of their tuning (Atallah et al., 2012). This provides direct and independent support for theories proposing that cortical computation is founded on the ability to control gain without fundamentally changing selectivity (Phillips and Singer, 1997; Kay et al., 1998; Phillips and Silverstein, 2003; Kay and Phillips, 2010).

Given that PV interneurons control gain, we need to consider the source of their inputs. One likely possibility is that this includes input from layer six cells in the same cortical column. Other optogenetic studies have revealed that excitatory cells in layer six of visual cortex control the gain of visually evoked activity in pyramidal neurons in the higher layers of the same cortical column (Olsen et al., 2012). This establishes pyramidal cells in layer six as a major mediator of cortical gain-control, so a major task for the future is to discover more about their inputs.

In addition to the PV-expressing class of inhibitory interneuron there is another large class, including Martinotti cells, which express the neuropeptide somatostatin (SOM interneurons). They are not fast-spiking and have axonal arbors on the distal dendrites of pyramidal cells. They are widely distributed across mammalian cortex, including that of humans, and are involved in the regulation of various processes. Of particular relevance here is their role in visual gain-control. By selectively reducing SOM interneuron activity using optogenetic techniques, it has been shown that they contribute to surround suppression (Adesnik et al., 2012). Reducing their activity significantly reduced surround suppression of layer 2/3 neurons by between 10 and 30%. SOM interneurons are but one of several mechanisms for surround suppression, however, as it is also in part inherited from earlier stages of visual processing, and is also in part due to other types of inhibitory interneuron and circuit mechanisms (Adesnik et al., 2012). A plausible default assumption is that visual surround-suppression by divisive normalization is but one example of a general computational strategy for suppressing highly probable signals, thus making less probable signals more salient (Seriès et al., 2003; Carandini and Heeger, 2012).

At the intracellular level there are several mechanisms by which gain-control can be implemented. These include shunting inhibition, background noise induced by balanced excitatory and inhibitory input, nonlinear dendritic integration such as dendritically localized NMDAR-mediated spikes, and short-term depression (STD) which can provide a mechanism for multiplicative gain-control if the contextual inputs are received on synapses distant from the cell body (Silver, 2010). Two mechanisms may be of particular relevance to the general computational issues considered here, as well as to schizophrenia (see below). One involves PV interneurons because they have the ability to amplify or suppress pyramidal cell activity. They also modulate temporal precision and generate synchronized gamma rhythms. They do this by controlling the "window-of-opportunity" within which pyramidal cells can generate spikes given their driving inputs (Gonzalez-Burgos and Lewis, 2008; Phillips et al., 2010). Computational modeling shows that by synchronizing the local activity of PV interneurons to a greater or lesser degree this window can be opened more or less. This is because PV neurons exert a powerful veto on spiking, so synchronizing their bursts also synchronizes the periods between bursts. This synchronized disinhibition therefore provides a "window of opportunity" for spiking that could provide a means by which contextual inputs, such as those from selective attention, could control the gain of pyramidal cell responses to their driving inputs (Tiesinga et al., 2008). It is known that in rodent primary somatosensory cortex their excitatory inputs are on distal dendrites, and come from both thalamic and intracortical sources, whereas their inhibitory inputs are somatic and perisomatic (Kameda et al., 2012), but we now need to know far more about those sources.

Another mechanism that may be of particular relevance is modulation of proximally driven activity by distal nonlinear dendritic currents, because that can either increase or decrease response gain at the soma (Silver, 2010). The possibility that distal dendritic tuft inputs might modulate response gain to inputs at the soma and basal dendrites was explored computationally by Körding and König (2000). They showed that this enables the learning and processing of information that is relevant to the context. Lee and Sherman (2010) distinguished two classes of glutamatergic pathways in the auditory cortex, termed "drivers" and "modulators." Driving inputs are the informationbearing pathways, while modulators regulate transmission of the driving information. Driving inputs are received by proximal dendrites, whereas modulatory inputs are received by distal dendrites. Lee and Sherman (2010) also note that these two glutamatergic pathways are fundamentally different in other ways. Driving inputs are received from thick axons at ionotropic synapses, and produce large EPSPs via depressing synapses and dense synaptic arbors. Modulatory inputs are received from thin axons at ionotropic and metabotropic synapses, and they produce small EPSPs via facilitating synapses and sparse synaptic arbors. All these differences are in agreement with the distinction between driving inputs and context-sensitive gain-control on which our hypotheses here are based. Lee and Sherman (2010) argue that their distinction between drivers and modulators clarifies the function of the many parallel and descending pathways in the auditory and other sensory pathways. We agree, and argue for the potential relevance of such a distinction to cortical processing in general. Further support for the view that some contextual influences operate via thin distal dendrites is that the cortico-cortical projections that are likely to convey them terminate preferentially in superficial cortical layers and on the distal segments of apical dendrites of pyramidal cells, which are especially rich in NMDARs (Monaghan and Cotman, 1985; Rosier et al., 1993). We do not suggest that all contextual influences operate via distal dendrites, however. Inhibitory modulatory influences from PV cells are received on or proximal to the soma, so they do not operate via distal synapses. Furthermore, other mechanisms that are both modulatory and proximal may yet be discovered. A simple summary of the current evidence is that direct modulatory interactions between pyramidal cells seem to be predominantly distal, as does modulation by inhibitory SOM interneurons, whereas modulation by inhibitory PV interneurons is proximal to or on the soma.

Though much remains to be learned about the functions and mechanisms of context-sensitive gain-control, one important conclusion is already clear. It is not a single function with a single mechanism. It is a family of regulatory functions served by a variety of mechanisms, and with complex interactions between them; there is no need for evolution to produce only mechanisms that are easy for us to understand. The inhibitory interneuron activity that produces changes in pyramidal cell gain can itself be modulated by NMDAR-mediated input to the inhibitory interneurons. Thus pyramidal cells modulate each others' activities directly via NMDAR-mediated connections, and indirectly via their effects on the modulation produced by inhibitory interneuron activity. The different mechanisms nevertheless make different contributions as they are suited to different functions. It is going to be difficult to find out exactly which mechanisms do what because their capabilities depend on so many things (Silver, 2010). These include: (1) whether it is input or output gain that is modulated; (2) the morphological complexity of the cell whose activity is modulated; (3) whether the modulatory inputs are proximal to the soma or on distal apical dendrites; (4) whether the modulatory synapses are clustered or widely distributed; (5) whether the gain is to be increased multiplicatively or decreased divisively; (6) the time-scale over which gain is modulated; and (7) whether it operates on sustained highfrequency rate signals or on sparse and brief but temporally correlated population signals. Therefore, a major task for the cognitive neuroscience of the future is find out which of the various mechanisms for context-sensitive gain-control contribute to each of its various uses. This is a difficult task, but it may be greatly facilitated by studying disorders, such as schizophrenia, in which functions and mechanisms of context-sensitive gain-control are both impaired.

## **THE FUNCTIONS AND MECHANISMS OF CONTEXT-SENSITIVE GAIN-CONTROL ARE IMPAIRED IN SCHIZOPHRENIA**

Here we discuss visual and other impairments in schizophrenia in the light of the functions and mechanisms of context-sensitive gain-control reviewed above. As our focus is on impairments of basic capabilities common to many different cognitive domains and cortical regions, we are not constrained to consider only impairments that are specific to perception. We do need to ask whether or not they are specific to context-sensitive gaincontrol, however. The evidence suggests that schizophreniarelated impairments are rarely all-or-none, so our default working assumption is that the capabilities impaired are still operating to some extent, though less effectively.

#### **IMPAIRMENTS OF VISUAL PERCEPTION IN SCHIZOPHRENIA INVOLVE CONTEXT-SENSITIVE GAIN-CONTROL**

Many studies show that impairments of visual perception in schizophrenia involve reduced context-sensitivity and gaincontrol. There is no need for a comprehensive review of all these impairments here because they will be the focus of other papers within this *Frontiers* Research Topic. Here it is sufficient to comment on recent assessments of this issue (Butler et al., 2008, 2012; Green et al., 2009), and to outline a few further findings.

Butler et al. (2008) divide the visual functions that are impaired in schizophrenic disorders into two groups, "gain control" and "integration." They define gain control as processes optimizing response to stimuli within a particular surrounding context. One form of this is that in which the neurons' dynamic range is modulated so as to increase responses to differences between adjacent and successive stimuli, as seen, for example, in "pop-out" and "surround suppression" paradigms. Divisive gain normalization is the appropriate form of gain-control in that case, and center-surround suppression has been shown to be reduced in schizophrenia (Dakin et al., 2005; Yoon et al., 2009). Another form of gain control (Butler et al., 2008) is the amplification of driving inputs that are present but weak such as those produced by near-threshold stimuli, as shown, for example, by facilitation of the detection of a low-contrast edge by collinear flankers. This form of gain-control can be studied in various psychophysical and electrophysiological paradigms that measure contrast-sensitivity under conditions designed to reveal the operation of either the magnocellular or parvocellular visual pathway, and with either transient, moving, or steady-state stimulation. In general these paradigms include any in which the preceding, concurrent, or following context amplifies signals coherently related to that context. Multiplicative gain amplification is appropriate for this form of gain-control. It is clearly impaired in schizophrenia but not in other forms of serious mental illness (Butler et al., 2005; Kéri et al., 2005a,b, 2009). Impairment may be greater in magnocellular than in parvocellular pathways (Butler et al., 2005, 2008).

Butler et al. (2008) define "integration" as the process linking the output of neurons into globally coherent subsets, where their individual activities are assumed to code for local attributes. This is therefore equivalent to what is here and elsewhere referred to as dynamic Gestalt grouping. There are many paradigms for studying such grouping, with contour integration being an example that is often used because it can be rigorously controlled. Since 1961, many of these paradigms have been used to study visual grouping in schizophrenia (Snyder, 1961; Snyder et al., 1961), with the general conclusion being that it is impaired, as reviewed by Silverstein and Keane (2011). Impaired grouping in schizophrenia has been demonstrated in studies of perceptual organization of static forms, fragmented forms, completion of occluded objects, illusory correlations, and coherent motion. This evidence includes psychophysical, electrophysiological, and brain imaging data (e.g., Spencer et al., 2003; Silverstein et al., 2009; Sehatpour et al., 2010; Chen, 2011).

It is well-established that face processing is impaired in schizophrenia (e.g., Uhlhaas et al., 2006a; Turetsky et al., 2007; Silverstein et al., 2010; Soria Bauser et al., 2012), including changes in the perception of emotion (e.g., McBain et al., 2010), which can contribute much to disordered interpersonal interactions. As perceptual deficits are not confined to higher levels of processing, deficits at lower levels may account for some of the face processing impairments (Turetsky et al., 2007; McBain et al., 2010; Silverstein et al., 2010). Schizophrenia patients need more visual information and use it atypically (Lee et al., 2011), with configural or holistic processing being particularly impaired (Shin et al., 2008; Joshua and Rossell, 2009). All these findings harmonize well with the hypotheses we propose. Impairments in face perception are also observed in body dysmorphic disorder, the only other psychiatric condition in which perceptual organization impairments have been observed (Feusner et al., 2007, 2010), and where half of the patient population also exhibits delusional psychotic symptoms (Phillips et al., 2006).

Schizophrenia-related deficits have been shown to be specific to context-sensitive gain-control in experiments that use conditions in which context is misleading. If performance deficits are specifically due to reduced effects of context then performance may be supra-normal when context is misleading. This was shown to be the case in a size perception task where surrounding figures provided a context that was helpful in some conditions and misleading in others (Silverstein et al., 1996; Uhlhaas et al., 2006b). Schizophrenia patients in those studies were neither helped by helpful context nor hindered by misleading context. Similar results were reported by Dakin et al. (2005) who found that schizophrenia patients had decreased center-surround antagonism in a contrast perception task. High-contrast surrounds reduced perceived contrast of the central target in control subjects but not for most of the patients, with the consequence that patient's judgments were then more veridical than normal. Finally, Tadin et al. (2006) found that schizophrenia patients had reduced surround suppression in a motion perception paradigm, including more veridical performance in conditions where context was misleading.

Figure-ground segregation using brief temporal cues is also severely impaired in many but not all schizophrenia patients (Hancock et al., 2008). This was demonstrated in a task based on figure-ground segregation by onset-asynchrony. Performance in this task is likely to be particularly sensitive to the function of magnocellular pathways because it is concerned with rapid attentional capture, at low spatial resolution, of overall stimulus organization. Most people can segregate figure from ground when the asynchrony of their onsets is about 24 ms, but 7 of 9 chronically disorganized schizophrenia patients required asynchronies of at least 50–100 ms. (Hancock et al., 2008). Furthermore, 7 of 63 undergraduate students also showed poor temporal resolution in this task, four of whom had schizotypy disorganization scores well into the clinical range, suggesting that this psychophysical paradigm may provide a useful endophentoype for the disorder.

Eight possible uses for context-sensitive gain-control were listed in sub-section Context-Sensitive Gain-Control has Several Computational Functions. So far we have cited evidence that four are impaired in schizophrenia. What of the other four, i.e., selective attention, modulation of precision in probabilistic inference, arbitrary input-output mappings, and coordinate transformation? All are relevant to vision, though none to vision alone. Selective attention is clearly one of the major impairments in schizophrenia, and is related to positive symptoms (e.g., Cornblatt et al., 1985). Imprecise signaling in probabilistic inference may also make a major contribution to positive symptoms (e.g., Fletcher and Frith, 2009; Seth et al., 2011), as discussed further below. The use of context to guide selection of one from a number of possible mappings is also likely to be impaired, though we know of no work explicitly relating that to the model of Salinas (2004).

Finally, although it is often emphasized as a paradigmatic example of gain modulation in general (Salinas and Their, 2000; Salinas, 2009), coordinate transformation seems the least likely to be impaired in schizophrenia. Schizophrenia patients show no signs of disordered gaze or reaching, or of inadequate coordinate transformation in any other domain. Schizophrenia patients do demonstrate heightened spatial frame illusions, and this may suggest abnormalities in visuo-motor functioning (Chen et al., 2011). Schizophrenia patients do not demonstrate the normal degree of attenuation of sensory feedback during self-initiated movements, and this has been proposed as a factor in the formation of delusions of control by external entities (Landgraf et al., 2012). Similarly, schizophrenia patients do show heightened susceptibility to the rubber hand illusion, suggesting a more dynamic and flexible representation of their body in space (Thakkar et al., 2011). None of these findings suggest primary impairments in coordinate transformation, however. Maybe there are none. The obvious prediction from our theory is that coordinate transformation will only be impaired to the extent that it depends upon the same neuronal mechanisms as the forms of context-sensitive gain-control that are impaired. As many forms of context-sensitive gain-control are impaired but coordinate transformation seems not to be, this raises the possibility that, instead of being a paradigmatic example of gain-control in general, it is somehow quite different. Further thought on this issue reveals that neither of the two classes of input on which coordinate transformation depends meet our long-used criteria for classifying an input as "context," because, according to our definitions, contextual inputs are neither necessary nor sufficient, whereas in coordinate transformation both classes of input that it combines are necessary. Knowledge of stimulus position relative to the retina and of eye-position relative to the head are both necessary to compute stimulus position relative to the head, for example. Thus, coordinate transformations depend on multiplicative interactions that are inherently symmetric as both terms are necessary. Driving and contextual interactions are inherently asymmetric, with context having a secondary, dependent, status (Phillips, 2012). Thus, this seems to provide a clear case where studies of visual impairments in schizophrenia contribute to our understanding of context-sensitive gain-control in general, because they indicate that, contrary to previous assumptions, it needs to be distinguished from coordinate transformation.

Overall, we can conclude that the context-sensitive perceptual operations of divisive gain suppression, multiplicative gain amplification, dynamic Gestalt grouping, and face and object perception are all impaired in schizophrenic disorders, though to different extents in different cases and conditions. This evidence shows that such impairments can occur at multiple levels of visual processing, and it suggests that they probably also occur in other modalities. These impairments are not constant over time. Some have been demonstrated to be state-sensitive in that they are more pronounced when patients are acutely psychotic compared to when their symptoms are in remission (Uhlhaas et al., 2005; Silverstein and Keane, 2009; Keane et al., in press; Silverstein et al., submitted, this research topic). Moreover, some of these state-sensitive impairments also occur in healthy volunteers administered ketamine, an NMDA antagonist (Uhlhaas et al., 2007; Morgan et al., 2011), as expected given the major contribution of NMDARs to gain-control according to our theory.

#### **NEURONAL MECHANISMS FOR CONTEXT-SENSITIVE GAIN-CONTROL ARE IMPAIRED IN SCHIZOPHRENIA**

The classical neuromodulators that have long been implicated in schizophrenia, such as dopamine and acetylcholine, provide an obvious form of gain-control. Their effects are slow and diffuse, however, whereas the cognitive interactions that are most obviously impaired in schizophrenia must have high temporo-spatial specificity because they convey detailed cognitive content. Gaincontrolling interactions within and between the glutamatergic and GABAergic systems that convey that content must therefore also be involved, so our focus here is on the evidence that they do indeed play a central role in the pathophysiology of schizophrenia.

Some of the interactions that control gain are produced via direct NMDAR-mediated interactions between pyramidal cells, as reviewed in sub-section There are Various Local-Circuit and Cellular Mechanisms for Context-Sensitive Gain-Control. There is ample evidence that NMDAR-mediated signaling is impaired in schizophrenia as reviewed many times elsewhere (e.g., Phillips and Silverstein, 2003; Loh et al., 2007; Corlett et al., 2010; Kantrowitz and Javitt, 2010; Moghaddam and Javitt, 2012). Furthermore, a review of the evidence on genetic susceptibility and gene expression concluded that, although there are probably direct and indirect links to both dopaminergic and GABAergic signaling, glutamate transmission via NMDARs is especially implicated (Harrison and Weinberger, 2005). Conclusive evidence that NMDAR hypofunction can produce many schizophrenic symptoms comes from an autoimmune disease first reported in 2007. This is an anti-NMDAR encephalitis that progressively reduces the activity of NMDARs by capping and internalizing them (Hughes et al., 2010). Patients present with acute schizophrenia-like symptoms, including paranoia. They are often admitted to psychiatric institutions and later develop severe catatonia, catalepsy, and stereotyped movement disorders. This provides unequivocal evidence that NMDARhypofunction can produce symptoms of schizophrenia, as we have long supposed.

Though less dense than on pyramidal cells, there are also NMDARs on the inhibitory interneurons on which several of the other mechanisms for context-sensitive gain-control discussed above depend. There is plenty of evidence that interneuron activity is also impaired in schizophrenia. Multiple studies have reported alterations in markers of inhibitory GABAergic neuronal activity (e.g., Lewis et al., 2005; Gonzalez-Burgos et al., 2010; Lewis, 2012), including their association with reduced centersurround suppression in visual cortex (Yoon et al., 2010). This deficit appears to be particularly pronounced in the subset of GABAergic neurons that express the calcium-binding protein PV (Hashimoto et al., 2003). Thus, this provides another route by which NMDAR-hypofunction could contribute to some of the deficits in schizophrenia. For example, when transgenic mice are generated in which NMDARs are selectively deleted from cortical and hippocampal GABAergic PV interneurons this produces selective molecular, physiological, and behavioral changes similar to some in schizophrenia (Nakazawa et al., 2012). Behrens and Sejnowski (2009) review evidence suggesting how dysregulation of PV interneurons in the developing cortex could explain the late onset of schizophrenic symptoms as well as the differences between the effects of brief and prolonged exposure to NMDA antagonists (Jentsch and Roth, 1999). The division of PV interneurons into two major classes is based on the principal target of their axon terminals. The axon terminals of the basket cell class target the cell body of pyramidal neurons and their proximal dendrites. The other major class, chandelier cells, gives rise to terminals that exclusively target the axon initial segments of pyramidal cells. There is evidence that both classes are impaired in a way that is specific to schizophrenia (Lewis et al., 2005; Lewis, 2012). PV interneurons also play a major role in setting the levels of temporal precision. This suggests that their impairments may play a major role in the reduced temporal precision of figure-ground segregation in schizophrenia reported by Hancock et al. (2008). Further evidence for the role of PV interneurons and synchronized rhythms in the development of schizophrenia is provided by Lee et al. (2012) who report that, in a neurodevelopmental rat model of schizophrenia, adolescent cognitive training changed PV-labeling in mature prefrontal interneurons, normalized the synchrony of neural oscillations between the left and right hippocampi, and prevented adult cognitive impairment.

Impairments of inhibitory interneuron activity could thus have several cognitive consequences. Many researchers, such as Lewis (2012), focus on consequences for WM and executive functions of the dorsolateral prefrontal cortex. We agree that dysfunctions of PV interneurons have consequences for WM and executive function, but from our perspective that provides far too narrow a focus as argued above. There is no good evidence linking the many selective impairments of perception reviewed here and elsewhere to WM or executive impairments. Effects of PV GABAergic impairment could also include many other cognitive functions as a consequence of their pivotal role in temporally precise activities including the generation and timing of rhythmic activity in the gamma frequency range (Cobb et al., 1995; Pouille and Scanziani, 2001). It is well-established that a wide range of cognitive deficits are associated with NMDARhypofunction and changed gamma-band activity in schizophrenia (Dzirasa et al., 2009; Uhlhaas and Singer, 2010). Uhlhaas and Singer (2012) review further evidence showing that synchronization of high-frequency rhythms is essential for the dynamic coordination of activities that are impaired in schizophrenia. They also summarize evidence suggesting that impaired longrange dynamic coordination of activity across brain-regions may be central to both of these disorders. The effects of impaired NMDAR-mediated neurotransmission on pyramidal cells and PV interneurons are particularly implicated. For example, the correlation between reduced GABAergic tone and reduced surround suppression in schizophrenia (Yoon et al., 2010) is probably mediated by gamma frequency oscillations, as recent research indicates a strong relationship between these three phenomena in healthy humans (Edden et al., 2009). Uhlhaas and Singer (2012) note many similarities between schizophrenia and autistic spectrum disorders (ASD) with respect to changes in rhythmic synchronization. Most of the paradigms used to study vision in schizophrenia have also been used extensively to study autistic perception, but very few firm conclusions can be drawn from all that evidence (Simmons et al., 2009). The possibility that a greater emphasis on context-sensitive gain-control as attempted here may reveal more order in all the evidence concerning both functions and mechanisms in ASD remains to be explored.

In addition to PV interneurons, other classes of inhibitory interneuron also contribute to context-sensitive gain-control. Evidence that somatostatin expressing interneurons (SOM interneurons), such as Martinotti cells, play a major role in surround suppression (Adesnik et al., 2012) was reviewed in subsection There are Various Local-Circuit and Cellular Mechanisms for Context-Sensitive Gain-Control. There is evidence for SOM interneuron impairment in schizophrenia (Morris et al., 2008), and surround suppression is one of the forms of context-sensitive gain-control shown to be impaired in schizophrenia. Therefore, that impairment may be due to impairments of SOM interneuron activity.

Overall, the neurobiological evidence suggests that schizophrenia involves impairments of: (1) NMDAR-mediated interactions between cortical pyramidal cells, (2) the activities of PV inhibitory interneurons, and (3) the activities of SOM inhibitory interneurons. The interneuron impairments could, at least in part, be due to reductions in their NMDA-mediated synaptic input. All three mechanisms play a major role in context-sensitive gain-control, as outlined in section Context-Sensitive Gain-Control Plays a Central Role in Brain Function and Mental Life. An important direction for future research is therefore to relate these physiological impairments to the various signs and symptoms of schizophrenia. As noted above, there is growing evidence for: (1) state-sensitivity of impairments in context-sensitive gaincontrol in schizophrenia (Silverstein et al., 1996; Uhlhaas et al., 2005; Silverstein and Keane, 2009; Keane et al., in press; Silverstein et al., submitted); (2) relationships between reduced contextual effects in perception and fragmentation in thinking (Uhlhaas et al., 2006b; Horton and Silverstein, 2011; Silverstein and Keane, 2011); and (3) relationships between abnormal GABAergic activity and context-sensitive gain-control in schizophrenia (Yoon et al., 2010). Symptoms are, by definition, state related, and many theories now relate positive and disorganized symptoms of psychosis to altered states of NMDARs and interneuron activity. However, the development of pharmacotherapy on the basis of these theories, though promising, has not yet clearly improved on clozapine, which has been available for 50 years (Barch, 2010; Moghaddam and Javitt, 2012). This may, in part, be due to the difficulty of specifying clinically optimal doses. It could also be related to the need to distinguish between subtypes of receptor and post-synaptic cell. For example, if impairments are due to reduced activity of only a particular NMDAR subtype on a particular class of post-synaptic cell, then that would not be overcome by a systemic enhancement of NMDAR activity in general. Therefore, we need a better understanding of the different functional roles and developmental trajectories of the different subtypes of NMDAR.

## **THE POSITIVE SYMPTOMS OF SCHIZOPHRENIA ARE RELATED TO CONTEXT-SENSITIVE PROBABILISTIC INFERENCE AND GAIN-CONTROL**

Psychiatrists have often concluded that contextual regulation of ongoing processing is particularly relevant to the induction of thought disorder (e.g., Barrera et al., 2005). Over the last few years this possibility has been developed into rigorously formulated theories that focus on the use of context to guide processing toward inferences that are both coherently related to each other and well adapted to the current circumstances. These theories often assume a form of hierarchical Bayesian inference that adapts and learns by reducing prediction error, where the predictions arise from higher levels of processing (e.g., Friston, 2010). In addition, we emphasize that predictive inputs are also provided by lateral interactions within levels. Such theories have been used to explain hallucinations (e.g., Friston, 2005) and various forms of delusion (e.g., Hemsley and Garety, 1986; Garety et al., 2001; Corlett et al., 2007; Fletcher and Frith, 2009; Clark, 2013). In essence, to the extent that perception is under-constrained by prior experience of statistical regularities in the world, misperceptions and false attributions of meanings can result. These can produce a sense that the world is changing, giving rise to delusional explanations for these subjective changes. Delusions of agency are well-explained by these models on the assumption that they arise from reduced precision in the predictions of selfinduced sensory signals (Fletcher and Frith, 2009; Stephan et al., 2009; Synofzik et al., 2010). In section Context-Sensitive Gain-Control Plays a Central Role in Brain Function and Mental Life we cited work showing how theories of this kind can explain the normal sense of conscious presence as arising from the correct prediction of interoceptive signals (Seth et al., 2011). That theory explains how disorders of both conscious presence and emotion could arise from reductions in the context-sensitivity and precision of probabilistic inference. Such theories can explain many of the psychotic symptoms that are seen in schizophrenia patients. Thus, they may provide important insights into the well-established symptoms of schizophrenia, and all depend upon context-sensitive gain-control. They imply a distinction between drivers and modulators because the predictions that are central to these accounts are thought to be modulatory and implemented by specialized synaptic interactions, such as those using NMDARs and inhibitory interneurons. Though we are not convinced by some aspects of the theories based on predictive coding (Phillips, 2013; Silverstein, in press), we agree with their emphasis upon the necessity of using probabilistic inference to interpret interoceptive inputs as well as those from the external world, and we emphasize the role of context-sensitive gain-control in doing that. Many of the positive symptoms of schizophrenia can thus be seen as arising from predictions that are pathologically imprecise because inadequate use is made of context to make them more precise. The use of contextual modulation can also enable the selection of perceptual interpretations or motor commands that have low probability overall, but high probability in special contexts. Thus, in addition to the symptoms noted above, weakened context-sensitivity could also lead to various other impairments of perception, thought, and action. Recent evidence in support of this is that reduced application of a convexity prior during perception of a hollow mask can lead to more veridical perception of such stimuli by schizophrenia patients. Furthermore, the extent of veridical perception by such patients was related to higher levels of hallucination and delusion, and to fewer days since last hospital discharge (Keane et al., in press). Moreover, this reduced sensitivity to the "hollow-mask illusion" has been shown, in dynamic causal modeling analyses of ERP and fMRI data, to be due to reduced top-down modulation of occipital lobe output in people with schizophrenia (Dima et al., 2009, 2010), as our theory predicts.

#### **DIFFICULTIES FOR THE HYPOTHESES PROPOSED AND MAJOR ASPECTS TO BE FURTHER DEVELOPED**

Hypotheses as general and abstract as ours cannot be confirmed or refuted by a single definitive experiment. Nevertheless, they can be strengthened or weakened by further evidence. For example, if further studies reveal many perceptual deficits in schizophrenia that are neither primary nor secondary consequences of impaired context-sensitive gain-control then our hypothesis concerning the functional impairments in schizophrenia would need to be amended. It will therefore be of great interest to see whether other papers published as part of this *Frontiers* Research Topic reveal such deficits. If schizophrenia were shown to be due to impairments of mechanisms unrelated to context-sensitive gain-control then our hypothesis concerning the neuronal bases of schizophrenia could be rejected. Our hypotheses carry many implications concerning mechanisms that can be tested and developed by further work. Indeed, differences between our emphases now and those in Phillips and Silverstein (2003) show this clearly. Then we placed great emphasis on the role of NMDAR-mediated interactions between pyramidal cells as the mechanism for context-sensitive gain-control. Now we also place great emphasis on the role of PV interneurons because recent findings, such as those using optogenetic techniques, demonstrate that they are well-suited to perform context-sensitive gain-control (Atallah et al., 2012).

Most fundamentally, our hypotheses depend upon the distinction between context-sensitive gain-control and the driving signals that convey content. If that distinction were shown to be misleading or of no use then our perspective could be justifiably ignored. Though many arguments and findings have been offered in favor of such a distinction by ourselves and others, some researchers remain unconvinced, so we acknowledge that this fundamental distinction remains open to question. Furthermore, we assume that presentation of the distinction as dichotomous is merely a heuristic simplification, but we have not given examples of cases that are intermediate between the two poles of the distinction.

Theories founded on the notion of optimal Bayesian inference have been challenged in various ways (e.g., Jones and Love, 2011). For example, Bowers and Davis (2012) argue that such theories are difficult to test because *post-hoc* assumptions about priors or likelihoods can be used to explain almost anything. They also argue that human inference is often not optimal, and that the neurobiological evidence for such theories is weak. Clark (2013) also notes that, being founded on the narrow goal of reducing prediction error, these Bayesian theories present a bleak desert-landscape view of mental life. Though most commentators on his *Behavioural and Brain Sciences* target article support his enthusiasm for predictive processing, several raise other difficulties that our perspective may help reduce. First, one difficulty often raised concerns optimality, but we do not assume optimality. On the contrary, we argue that the conditions for optimality at the systems-level can be met only in exceptionally simple cases (Phillips, 2012). Second, the neurobiological evidence for our hypotheses is strong and rapidly becoming stronger as it is supported by the optogenetic evidence that is now being used to explore the mechanisms of context-sensitive gain-control. Third, the theory of Coherent Infomax that underlies the hypotheses proposed in this paper avoids the desert-landscape criticism by emphasizing the objective of maximizing coherent inference rather than that of reducing prediction error (Phillips et al., 1995; Kay and Phillips, 1997, 2010; Phillips, 2013). Finally, another difficulty facing any simple unifying theory is the need to explain the endless diversity of cognitive capabilities. Our perspective has to some extent met this need by showing that context-sensitive gain-control in visual size-perception varies greatly across people of different ages (Doherty et al., 2009), sex (Phillips et al., 2004), and culture (Doherty et al., 2008), but those studies are merely the first few steps into a largely unexplored territory.

Plenty of other difficulties and undeveloped possibilities remain. We cannot yet claim that all of the symptoms associated with schizophrenia are due to impairments of context-sensitive gain-control or their secondary consequences. Nor do we yet have fully adequate answers to questions concerning relations between schizophrenia-related impairments and the coordinate transformations that some see as a foremost function of gain-control. Is coordinate transformation impaired in schizophrenia or not? If not, why not? Is it because the form of gain-control involved in coordinate transformation is not context-sensitive in the way that the others are? Relations between classical neuromodulation and the more locally specific gain-control that we have emphasized also need to be further clarified. We expect them to be complex, and to operate in both directions. There is also much that needs to be clarified concerning the full range of schizophrenia-related deficits in visual perception. For example, it is well-established that these include changes in visual masking (Green et al., 2011). Such deficits may be related to the reduced temporal precision shown by Hancock et al. (2008) and to the impairments of PV inhibitory interneurons emphasize above, but we have not yet examined that possibility adequately.

# **REFERENCES**


cornsweet effect. *Front. Psychol*. 3:43. doi: 10.3389/fpsyg.2012.00043 Butler, P. D., Chen, Y., Ford, J. M., Geyer, M. A., Silverstein, S. M., and Green, M. F. (2012). Perceptual measurement in schizophrenia: promising electrophysiology and neuroimaging paradigms from CNTRICS. *Schizophr. Bull*. 38, 81–91. doi: 10.1093/schbul/sbr106


Overall, our view of the difficulties and immaturities faced by our perspective is that they offer far more opportunity for healthy growth than they do for fatal decline. It will be of great interest to see whether developments over the coming years justify that optimism.

#### **ACKNOWLEDGMENTS**

We thank Bassam Atallah, Christopher Fiorillo, Lowana Phillips, Emilio Salinas, Anil Seth, and Mike Spratling for their expert and insightful comments on a draft of this paper. The editor, Michael Green, and three reviewers provided invaluable advice on presentation of the hypotheses proposed.


H., Emrich, H. M., et al. (2009). Understanding why patients with schizophrenia do not perceive the hollow-mask illusion using dynamic causal modelling. *Neuroimage* 46, 1180–1186. doi: 10.1016/j.neuroimage.2009.03.033


and free-energy. *Front. Hum. Neurosci.* 4:215. doi: 10.3389/fnhum.2010.00215


*Schizophr. Bull.* 35, 163–181. doi: 10.1093/schbul/sbn156


in schizophrenia. *Cogn. Neuropsychiatry* 14, 77–86. doi: 10.1080/13546800902757936


*Cereb. Cortex* 18, 1575–1587. doi: 10.1093/cercor/bhm186


eds C. von der Malsburg, W. A. Phillips, and W. Singer (Cambridge: MIT Press), 1–24.


T., Charbonnier, D., and Gur, R. C. (2007). Facial emotion recognition in schizophrenia: when and why does it go awry? *Schizophr. Res.* 94, 253–263. doi:


interneurons of rat prefrontal cortex. *Neuropsychopharmacology* 34, 2028–2040. doi: 10.1038/ npp.2009.20


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Received: 20 February 2013; accepted: 13 May 2013; published online: 29 May 2013.*

*Citation: Phillips WA and Silverstein SM (2013) The coherent organization of mental life depends on mechanisms for context-sensitive gain-control that are impaired in schizophrenia. Front. Psychol. 4:307. doi: 10.3389/fpsyg. 2013.00307*

*This article was submitted to Frontiers in Psychopathology, a specialty of Frontiers in Psychology.*

*Copyright © 2013 Phillips and Silverstein. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.*