Edited by: Sharon Gilaie-Dotan, University College London, UK
Reviewed by: Jody C. Culham, University of Western Ontario, Canada; Holly Bridge, University of Oxford, UK
*Correspondence: Alan J. Pegna
†These authors have contributed equally to this work.
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.
Fast and automatic behavioral responses are required to avoid collision with an approaching stimulus. Accordingly, looming stimuli have been found to be highly salient and efficient attractors of attention due to the implication of potential collision and potential threat. Here, we address the question of whether looming motion is processed in the absence of any functional primary visual cortex and consequently without awareness. For this, we investigated a patient (TN) suffering from complete, bilateral damage to his primary visual cortex. Using an fMRI paradigm, we measured TN's brain activation during the presentation of looming, receding, rotating, and static point lights, of which he was unaware. When contrasted with other conditions, looming was found to produce bilateral activation of the middle temporal areas, as well as the superior temporal sulcus and inferior parietal lobe (IPL). The latter are generally thought to be involved in multisensory processing of motion in extrapersonal space, as well as attentional capture and saliency. No activity was found close to the lesioned V1 area. This demonstrates that looming motion is processed in the absence of awareness through direct subcortical projections to areas involved in multisensory processing of motion and saliency that bypass V1.
Moving stimuli are thought to be highly salient and several studies have suggested that motion is a powerful attractor of attention. For example, in a visual search paradigm, Franconeri and Simons (
A likely explanation for the heightened attentional capture of looming stimuli may be provided by evolutionary psychology, as the early and rapid detection of looming stimuli appears to be an essential factor for the survival of the individual and, on a larger scale, of the species. Indeed, looming stimuli are highly relevant in that they indicate possible collision with the observer. Stereotypical avoidance behavior, such as head withdrawal and covering of the face with the arms have been observed in monkeys for approaching but not receding or randomly moving stimuli (King and Cowey,
From a general perspective, unconscious processing of visual stimuli has also been observed in patients suffering from cortical blindness. This has been termed “blindsight” and is “a condition in which the sufferer responds to visual stimuli without consciously perceiving them” (Weiskrantz et al.,
Less is known about the effects of direction of motion in blindsight, in particular looming motion. In monkeys, looming was investigated after unilateral ablation of the occipital cortex (King and Cowey,
To the best of our knowledge, no studies have addressed the effect of looming motion in cortical blindness in humans. We therefore examined the response to moving dot patterns in a patient who suffers from complete cortical blindness following bilateral strokes to the posterior cerebral arteries. This patient, TN, has been described elsewhere and was shown to present remarkable residual visual abilities including affective blindsight (Pegna et al.,
TN is a well-studied cortically blind patient that we have recently described in detail (Van den Stock et al.,
Clinically, TN continues to behave as a blind person, for example using a cane as a tool to guide himself while walking, locating food on his plate with his fingers, etc…When in presence of his wife or friends, he will rely on them for navigation, for example by placing his hand on their shoulder when walking. It should be pointed out that nothing in his behavior suggests the spontaneous use of vision to guide his actions.
Despite these observations, he sporadically claims to “feel” the occurrence of visual movement. Despite his alleged impressions of visual movement, a careful clinical assessment revealed numerous misses and false positives in his responses. For example, when asked to attend to the possible movement of the examiner's (AP) finger, he would frequently signal its occurrence before any motion had occurred, while on other occasions, he would fail to respond to entirely. Subsequently, in order to determine the extent of actual movement perception that was possible, TN was presented with a series of random dot stimuli consisting of 1000 ms video clips of static or random movement stimuli. Each stimulus consisted of 13 light points placed randomly on the screen that could move randomly at different speeds. TN was instructed to respond if he detected movement or not. He was informed of the presence of the stimulus on the screen by a beep indicating onset and offset. This was followed by a verbal prompt by the examiner.
TN's performance on this task revealed a hit rate of 63%. Out of 192 stimuli presented (96 moving and 96 static stimuli), TN responded correctly to 121 stimuli. With respect to a binomial distribution, this rate yields a Z score of 3.61 with a probability of chance occurrence at
He participated in the present experiment aged 62, i.e., 10 years after the loss of conscious visual perception. He gave his informed consent to participate in this study that was approved by the Ethics Committee of Geneva University Hospital.
Videos of dot-fields on a black background were generated to mimic various types of motion, namely: looming, receding, rotation (either clockwise or anticlockwise), and a static condition. Stimuli were back-projected onto a screen of size 40.5 × 24.2 cm with a resolution of 1680 × 1050 pixels, placed behind the participant's head. The participant viewed the screen through a mirror attached to the head coil, with a viewing distance of approximately 75 cm. Dots (diameter 0.18° of visual angle) were presented in a neutral gray (defined by RGB values set to 46.7%) on a black background. Stimuli were generated using Matlab 7.14 (The Mathworks, Natick, MA, USA). Video clips lasted 2 s at 30 frames per second.
Dots were distributed placed on the perimeter of three imaginary concentric squares. Eight points were placed in this manner, four on the apices and four on the midpoints of each side, such that all stimuli were composed of 24 dots (see Figure
Equiluminance of the stimuli was preserved by always having the same number of dots present on screen at any time, irrespective of condition (for the looming conditions, dots passing the edge of the screen wrapped around to the center of the screen, and for receding stimuli, dots disappearing to the center of the screen wrapped around to the edges).
A block design was employed in which videos of a given motion type were presented for 16 s (each speed of motion was presented twice in a random sequence, for rotation trials clockwise and anticlockwise stimuli were presented in equal numbers at each speed in a random sequence), followed by a 16 s period of no stimulation (a black screen). Block (i.e., motion condition) sequence was randomized. The fMRI session was divided into four runs, each of which contained two blocks of each motion condition, and the associated null events. Stimulus delivery was controlled using custom routines developed for PsychoPy (Peirce,
Anatomical and functional MR images were acquired in Maastricht University, with a 3T whole-body scanner (Magnetom Trio, Siemens, Erlangen, Germany) equipped with a 20-element head-neck coil.
T2*-weighted functional images were acquired using a gradient echo EPI sequence, with 2 × 2 × 2 mm3 resolution, covering the whole brain (75 slices without gaps,
In addition, a T1-weighted anatomical image (1 × 1 × 1 mm3 isotropic) was acquired with using an MPRAGE sequence (
Stimuli were back-projected onto a screen placed behind the participant's head. The participant viewed the screen through a mirror attached to the head coil, with a viewing distance of approximately 75 cm. The patient was asked to keep his eyes open and oriented straight ahead during the experimental presentation. His direction of gaze was monitored with an eye tracker camera, although it was not possible to obtain calibration values due to his inability to see the calibration points.
Preprocessing and analysis of the data was carried out in SPM8 (
Rigid realignment of each EPI volume to the first in the session, using a two-pass procedure in which the images are realigned to the mean image created after a first-pass realignment,
Coregistration of the MPRAGE image to the mean EPI image,
Spatial smoothing using a Gaussian kernel of 8 mm at full-width half-maximum height.
In order to be able to compare the patient's neuroanatomy with that of existing atlases, we normalized the T1 anatomical image to the standard MNI template included in SPM8, using the “unified segment” procedure (Ashburner and Friston,
The preprocessed EPI images were submitted to a GLM analysis, in which each scan was coded for a condition (looming, receding, rotating, static) and null events were left unmodeled (after Josephs and Henson,
A control group, composed of eight right-handed male participants (mean age: 58 years, range: 55–62) with normal or corrected-to-normal vision and no history of head injury was scanned using the same experimental paradigm. Control participants were scanned at Geneva University Hospitals, on a 3T Siemens Prisma, with a 20-channel head-and-neck coil. Screen-size and resolution for stimulus presentation were as for TN. Since, for operational reasons, these participants could not be scanned on the same scanner as TN, only qualitative comparisons of these data will be presented.
The control participants' data was analyzed at the first, fixed-effects level as TN's data. In order to obtain estimates of reliable activity over the group, contrast estimates from each of the participants were subsequently analyzed at the second, random-effects level. Second-level analysis comprised a repeated-measures ANOVA, in which the first-level condition vs. baseline images for each participant were entered into a design matrix to test for between-condition differences. In order to account for between-participant variability, “subject” was included as a factor in the analysis. Analysis was carried out in SPM8, following the methods described by Henson and Penny (
Since the primary focus of this paper is on the detection of looming stimuli by the brain's subcortical pathways, analyses will focus on the areas selectively activated by these stimuli, and particularly in contrast to those engaged by receding stimuli. The effect of speed of movement on BOLD response in each of the different movement conditions was tested and no significant effects were found. As a result, this condition was collapsed.
For completeness, results are presented at a relatively liberal threshold of uncorrected
The subtraction of looming and receding stimuli revealed a difference in BOLD response in a number of areas (see Figure
− |
− |
||||||
Posterior middle temporal gyrus [BA20] | R | 50 | −36 | −10 | 4.25 | 4.22 | |
Cerebellum crus I [BA37] | R | 52 | −50 | −30 | 4.02 | 3.99 | |
− |
|||||||
Angular gyrus [BA7] | R | 36 | −62 | 42 | 4.83 | 4.79 |
|
− |
− |
− |
|||||
Occipitotemporal area [BA37] | L | −42 | −60 | −10 | 4.76 | 4.72 |
|
Inferior occipital cortex [BA19] | L | −38 | −76 | −6 | 4.18 | 4.15 | |
Posterior superior temporal sulcus [BA21] | L | −44 | −46 | 10 | 4.15 | 4.12 | |
− |
|||||||
Middle cingulate gyrus [BA23] | R | 2 | −40 | 36 | 4.66 | 4.62 |
|
Precuneus | R | 16 | −44 | 40 | 4.2 | 4.17 | |
− |
− |
It can be seen from the plots of relative activation levels presented in Figure
Surprisingly and in sharp contradistinction with TN, control participants did not manifest any significant increase in BOLD response to looming over receding stimuli.
In order to determine whether any differential activation was reliably present for movement and therefore if a differential BOLD response could be observed in comparison with baseline, we examined the main-effect of motion in both TN and the control group. This result is described in the next section.
The results of this analysis is illustrated in Figure
− |
− |
||||||
Fusiform gyrus [BA18], V4 | R | 26 | −74 | −6 | 14.58 | 4.07 | |
Inferior occipital [BA37], area FG2 | R | 44 | −62 | −12 | 12.72 | 3.85 | |
− |
|||||||
Middle occipital gyrus [BA19] | R | 44 | −76 | 6 | 24.4 | 4.9 |
|
Posterior middle temporal [BA39], | R | 46 | −72 | 20 | 22.56 | 4.77 | |
Middle occipital [BA18] | R | 34 | −82 | 6 | 18.45 | 4.45 | |
Inferior temporal [BA19], V5/MT | R | 44 | −70 | −2 | 12.4 | 3.81 | |
Middle occipital [BA18] | R | 26 | −82 | 10 | 12.29 | 3.79 | |
− |
− |
− |
|||||
Middle occipital [BA37], V5/MT | L | −48 | −72 | 2 | 13.81 | 3.98 |
− |
− |
||||||
− |
|||||||
Angular gyrus [BA7] | R | 34 | −64 | 44 | 8.98 | 4.33 | |
− |
− |
||||||
Inferior occipital gyrus [BA37], area FG2 | L | −42 | −60 | −10 | 9.46 | 4.47 | |
Middle temporal gyrus [BA21] | L | −44 | −48 | 10 | 8.35 | 4.13 |
This study investigated the cerebral response to moving (looming) stimuli in a patient (TN) who became cortically blind after having suffered a bilateral loss of his primary visual cortex. In addition, an age- and gender-matched control group was scanned for comparison purposes. The results in TN showed that looming activated a network that included the posterior middle temporal lobes bilaterally, the left inferior parietal region, the right angular gyrus and the right precuneus. This occurred despite the fact that the patient could not detect the direction of motion of the stimuli or even their presence. Surprisingly, data obtained in the control group failed to produce any significant activation for looming, although motion
The importance of the motion selective region MT/V5 in humans was identified after a patient with bilateral damage to these areas was described. The patient was subsequently impaired in motion processing, suffering from what is known as akinetopsia (Zihl et al.,
By contrast, data from TN was slightly more surprising. Indeed, activation for motion in TN was not comparable to controls. The superimposition of the control group's regions of activation onto TN's brain revealed a certain amount of overlap of these clusters with the damaged areas. In TN, regions of significant activation for motion were in fact situated in the left middle temporal gyrus, the right inferior temporal gyrus and the angular gyrus. Furthermore, looming yielded several distinct areas of activation, including two clusters in the posterior middle temporal regions, one of them (on the left) coinciding with the area activated for movement
The activation of these middle temporal areas in TN were rather unexpected. Indeed, we had anticipated the activation of more posterior regions compatible with MT/V5. However, even allowing for the possibility of errors occurring in the normalization and warping procedures, the localization of the middle temporal clusters appears too posterior and therefore do not appear to be compatible with known hMT+ regions in the healthy brain (Watson et al.,
More recently, another hypothesis has suggested that blindsight may be linked to the transfer of visual information between healthy and lesioned hemispheres (Leh et al.,
Our current findings do not allow us to determine whether the geniculo-extrastriate or colliculo-pulvinar routes are involved in conveying information to the cortical regions activated in TN's brain. Nevertheless, activation was observed in the middle temporal regions (bilaterally for looming and on the left for motion) demonstrating the recruitment of typically non-visual cortex for movement processing and suggesting that cortical reorganization may have led more anterior regions to perform some of the functions usually performed by hMT+/V5, and more importantly, showing a heightened response to radially approaching movement.
The increased sensitivity of modality-specific cortical regions for looming has been observed on different occasions. For example, in a study addressing looming from a multimodal perspective, Tyll et al. (
In another brain imaging study with healthy controls, the existence of a collicular route for processing of looming was evidenced. Billington et al. (
The evidence for an enhanced activation of modality-specific cortical regions in response to looming is in line with TN's heightened cortical response. The preference of the sensory system for looming stimuli shows that an organism is biased toward stimuli reflecting possible collisions and thus potential threat. In point of fact, such enhanced processing for threatening stimuli has received much attention in the field of affective science, and findings have revealed that stimuli such as threatening facial expressions enhance stimulus-specific visual cortical areas and can in fact do so both when stimuli are consciously detected or not (Vuilleumier et al.,
Little is known about whether looming may be processed in the absence of awareness. One study addressed this using a binocular rivalry task. In binocular rivalry, the viewer can usually only report the stimulus in one of the two eyes, while the stimulus in the other is not consciously detected. Although the conscious percept shifts from one eye to the other in equal proportions, a more salient stimulus tends to be consciously perceived more often, catching the viewer's awareness. Following this line of reasoning, Parker and Alais (
Consistent with the idea of a saliency-driven response, our findings in TN additionally pointed to an involvement of the IPL, an area that also appears to be strongly linked with both saliency and multisensory processing. Data from the IPL is derived in large part from studies in the ventral inferior parietal area (VIP) of the monkey, which is located in the depth of the intraparietal sulcus (Bremmer et al.,
Motion selective cells recorded in area VIP appear to be sensitive to the direction of the light points moving across the visual field (expanding/approaching vs. contracting/receding or other movement), and of particular relevance in our study, to stimuli representing real or apparent self-motion (Colby et al.,
It has been proposed that a high level, saliency-based motion mechanism may be opposed to low level, luminance-based processes (Cavanagh,
Interestingly, a hemispheric specialization dissociation concerning saliency based attentional processes has been put forward (Mevorach et al.,
The presence of middle cingulate activity for looming motion in TN is notable. Several studies in healthy controls have found activity in the posterior cingulate for moving stimuli, in particular when coherent motion was represented (Sunaert et al.,
In conclusion, our findings revealed that, despite the absence of a functional primary visual cortex, looming motion activated a neural network that included bilateral clusters in the middle temporal lobes anterior to hMT+/V5 areas in the healthy brain that may have developed through cortical plasticity to respond to motion. In addition, cortical regions that generally participate in the processing of visual saliency (IPL) and visual guidance in space (IPL and cingulate) were found to be active. The fact that looming may continue to be processed without V1 may be due to its high level of salience and the importance of this direction of motion as a signal of imminent collision or threat.
The Review Editor Holly Bridge declares that, despite being affiliated to the same institution as author Marco Tamietto, the review process was handled objectively and no conflict of interest exists. 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.
This study was supported by a Swiss National Foundation for Scientific Research (grant 320030-144187) to AP. MT is supported by a Vidi grant from the Netherlands Organization for Scientific Research (NWO; grant 016.125.391) and by a FIRB—Futuro in Ricerca 2012—grant from the Italian Ministry of Education University and Research (MIUR; grant RBFR12F0BD). BG is partly supported by FES and FP7-FET-Open grants and by an Adv ERC grant.
The Supplementary Material for this article can be found online at: