Goal-Directed Movement Enhances Body Representation Updating

Wen, Wen; Muramatsu, Katsutoshi; Hamasaki, Shunsuke; An, Qi; Yamakawa, Hiroshi; Tamura, Yusuke; Yamashita, Atsushi; Asama, Hajime

doi:10.3389/fnhum.2016.00329

ORIGINAL RESEARCH article

Front. Hum. Neurosci., 28 June 2016
Sec. Sensory Neuroscience
Volume 10 - 2016 | https://doi.org/10.3389/fnhum.2016.00329

Goal-Directed Movement Enhances Body Representation Updating

Wen Wen^*†

Katsutoshi Muramatsu^†

Shunsuke Hamasaki

Qi An

Hiroshi Yamakawa

Yusuke Tamura

Atsushi Yamashita

Hajime Asama

Department of Precision Engineering, University of Tokyo, Tokyo, Japan

Body representation refers to perception, memory, and cognition related to the body and is updated continuously by sensory input. The present study examined the influence of goals on body representation updating with two experiments of the rubber hand paradigm. In the experiments, participants moved their hidden left hands forward and backward either in response to instruction to touch a virtual object or without any specific goal, while a virtual left hand was presented 250 mm above the real hand and moved in synchrony with the real hand. Participants then provided information concerning the perceived heights of their real left hands and rated their sense of agency and ownership of the virtual hand. Results of Experiment 1 showed that when participants moved their hands with the goal of touching a virtual object and received feedback indicating goal attainment, the perceived positions of their real hands shifted more toward that of the virtual hand relative to that in the condition without a goal, indicating that their body representations underwent greater modification. Furthermore, results of Experiment 2 showed that the effect of goal-directed movement occurred in the active condition, in which participants moved their own hands, but did not occur in the passive condition, in which participants’ hands were moved by the experimenter. Therefore, we concluded that the sense of agency probably contributed to the updating of body representation involving goal-directed movement.

Introduction

People’s perceptions of their own bodies constitute a fundamental aspect of self-consciousness and have been discussed by numerous researchers in multiple fields such as psychology, neuroscience, and engineering. Mental representation of the body involves both percepts and abstract knowledge, and beliefs and cognition about the body. Two concepts are usually used to describe body representation: body image and body schema (Gallagher, 1986; Paillard, 1999; de Vignemont, 2010). Usually, body image refers to perceptual, cognitive, or emotional awareness of the body, while body schema is more holistic and operates in a non-conscious manner, for example, containing certain habitual postures and movements (Gallagher, 1986). However, the definition and usage of these concepts is often unclear and confusing. Further, a prior study clarified body representation as belonging to two classes: perceptual representation (what the body is felt to be like), and cognitive representation (what the body is believed to be like; Longo et al., 2010). According to Longo et al. (2010), perceptual body representation involves online updating of the mental body, including superficial percepts and percepts of body size, shape, and posture, and influences our perception of external objects and behaviors. In contrast, cognitive body representation refers to the cognition of the body as a physical object in the external world (Longo et al., 2010), and is considered to be constructed and updated at the cognitive level. For example, in phantom limb pain, patients who have lost a limb by accident or amputation still have vivid percepts of the missing limb but know that it does not exist. In this case, the perceptual body representation in the brain has not been updated, but the cognitive representation has been updated (Longo et al., 2010). In the present study, we focused on the online updating of perceptual body representations in Longo et al.’s (2010) definition.

The underlying process of online body representation updating can be examined using the rubber hand illusion. In the original rubber hand illusion, after a fake rubber hand is brushed synchronously with the individual’s hidden real hand for a certain period, the individual feels the touch of the viewed brush, as if the rubber hand had sensed the touch (Botvinick and Cohen, 1998). Further, in the rubber hand illusion, individuals also experience the rubber hand as belonging to themselves (i.e., sense of ownership), and show a distorted sense of position of their real hands drifting to the rubber hand (Botvinick and Cohen, 1998). The rubber hand illusion reflects an interaction between vision, touch, and proprioception (Botvinick and Cohen, 1998). When the tactile input from the real hand matches visual information for the rubber hand, a sense of ownership of the real hand is extended to the rubber hand, and the incorrect position of the fake hand is integrated into the individual’s body representation and influences proprioception. Proprioceptive drift has been used widely as a behavioral indicator of body representation updating in the rubber hand illusion (Botvinick and Cohen, 1998; Tsakiris and Haggard, 2005a; Tsakiris et al., 2006; Costantini and Haggard, 2007; Kalckert and Ehrsson, 2012). In addition, the extension of body ownership in the rubber hand illusion has been examined not only via subjective rating of ownership (Shimada et al., 2009; Guterstam et al., 2011; Kalckert and Ehrsson, 2012, 2014) but also via objective measures such as skin conductance responses (Armel and Ramachandran, 2003; Ehrsson et al., 2008) and muscle activity (Perez-Marcos et al., 2009; Slater et al., 2009; Moretto et al., 2011). Although previous studies suggested common mechanisms for proprioceptive drift and the illusory sense of ownership (Botvinick and Cohen, 1998; Kalckert and Ehrsson, 2012, 2014), recent findings reported a distinction between them, suggesting that different multisensory mechanisms underlie the two phenomena (Kammers et al., 2009; Rohde et al., 2011; Blanke, 2012). Blanke (2012) suggested that body consciousness involves three aspects: self-identification (i.e., body ownership), self-location (where “I” am in space), and first-person perspective (the experience of the position from where “I” perceive the world), and that these aspects are probably based on distinct brain mechanisms.

Tsakiris and Haggard (2005b) examined the role of body schema on the rubber hand illusion, and found that proprioceptive drift occurred only when the rubber hand’s posture was similar in appearance to that of the real hand. The results indicated that the bottom-up integration of multiple sensory input is necessary, but not sufficient for body representation updating, and the illusion is modulated by top-down influences originating from one’s body representation (Tsakiris and Haggard, 2005b). Moreover, a recent study using immersive virtual reality demonstrated that the illusion of ownership could be generated from the realistic appearance of a fake (virtual) body and first-person perspective, even without any other congruent visuotactile or sensorimotor cues (Maselli and Slater, 2013), revealing the strong influence of body schema and first-person perspective in the illusion. Recent research using visuo-motor adaptation showed that synchronous movements can also produce the rubber hand illusion (Tsakiris et al., 2006; Kalckert and Ehrsson, 2012, 2014). In such paradigms, both the sense of ownership (i.e., the feeling that the fake hand belongs to oneself; Gallagher, 2000) and the sense of agency (i.e., the feeling that the movements of the fake hand are generated by oneself; Gallagher, 2000) are generated for the rubber hand. In Tsakiris et al.’s (2006) experiment, the participants watched a projection of their own hands on a screen and moved their index or little fingers. When the finger movements were generated by external forces, proprioceptive drift occurred only for the moving finger; in contrast, when participants moved their fingers themselves, proprioceptive drift also occurred for unmoving fingers, indicating that motor agency integrated distinct body parts into a coherent and unified whole (Tsakiris et al., 2006). Moreover, Tsakiris et al. (2007) pointed out that although afferent signals alone could generate sense of ownership and modify body representation, a more coherent experience of the body depends on the integration of both effect and afferent signals in an action context.

In the present study, we aimed to examine the influence of goal-related instruction and goal-attainment feedback on body representation updating, using the visuo-motor adapted version of the rubber hand paradigm. Goal-directed movements are very common in daily life. Prior studies have reported that goals and goal-attainment feedback greatly influenced sense of agency during action (Aarts et al., 2005; Metcalfe and Greene, 2007; Metcalfe et al., 2013; van der Weiden et al., 2013; Wen et al., 2015b,c). Sense of agency could play an important role in integrating body representation during body movement (Tsakiris et al., 2006, 2007). In the present study, we hypothesized that goal-directed movements may improve body representation updating, and sense of agency may play a role in this process. This issue is important in understanding the processes underlying body representation and helps to clarify the internal relationship between different factors underlying self-consciousness including sense of agency, ownership, and body representation. In the present study, we first clarified the effect of goal-directed movement on body representation updating in Experiment 1 by examining whether proprioceptive drift differed between conditions with and without a specific goal or goal-attainment feedback. Thereafter, we further examined the role of sense of agency in Experiment 2, in which we compared the effect of goals on proprioceptive drift between the active and passive conditions.

Experiment 1

Methods

The present study used the rubber hand paradigm with virtual reality stimuli. The participant moved his or her hidden left hand forward and backward, while a virtual left hand was presented 250 mm above the real hand and moved in synchrony with it. Participants were instructed to move their hands according to cued timing (without a specific goal) or an instruction to touch a moving object (with a specific goal). We examined the effects of having a goal and goal-attainment feedback on proprioceptive drift. The study was approved by the ethics committee of the School of Engineering at the University of Tokyo, and written informed consent was obtained from all participants prior to participation.

Participants

Twenty students with normal or corrected-to-normal visual acuity participated in the experiment (three women, mean age = 22.8 years, SD = 2.0, range 20–25 years). The sample size was chosen because it provided power of 0.95 in revealing the difference between the conditions with and without a goal, based on the results for the first five samples (a priori power was computed using G*Power 3, Faul et al., 2007).

Stimuli and Tasks

A virtual left hand was presented on the left side of a 597 × 336 mm (width × height) blue screen using OpenGL (an application programing interface for rendering graphics, Figures 1A,B). The size of the virtual hand was 105 × 145 mm (width × height), similar to the average size of a normal hand. The virtual hand moved between the bottom and center of the screen, in synchrony with the participant’s hand, and an arm was presented when the virtual hand moved to the center of the screen, giving the appearance of the virtual hand being stretched out from the bottom of the screen. Two horizontal lines were represented at the bottom and center of the screen, 149 mm apart. The color of the two lines alternated between red and white at random intervals of 4–6 s in the conditions without goals. Participants placed their hands on a sliding rail to restrict hand movements in a horizontal direction (Figure 1C). The distance between the sliding rail and the surface of the screen was 250 mm. A 3-D position sensor (PHANToM premium 1.5, Sensable Inc.) was attached to the participant’s index finger to measure hand position. The average delay in response in synchronous conditions was below 30 ms and was unperceivable for participants.

FIGURE 1

Figure 1. Example of the screen that presented a virtual hand and flying object in the conditions with the goal involving touching an object (A), example of the screen showing the virtual hand touching the flying object (B) and arrangement of the experimental devices (C).

In each trial, participants placed their left hand on the sliding rail and moved it forward and backward according to the stimuli on the screen. They were instructed to move the hand forward until the root of the index finger reached the upper line on the screen (e.g., Figure 1B) and backward until the root of the thumb reached the lower line on the screen (e.g., Figure 1A). Participants were told that they were not strictly required to move the distance indicated, as it was for use as a reference, and they should move the hand while maintaining a comfortable posture. Participants were also told to keep the hand at the initial position (Figure 1A) between movements.

There were five conditions between trials (Table 1). In the random condition, participants moved the left hand upon the same instruction in the no-goal and delay conditions; however, the virtual hand moved forward and backward at random times without responding to the position of the participant’s hand (the number of times the virtual hand moved was equal to the number of cues). In the random condition, the rubber hand illusion was not considered to have occurred and responses served as a baseline for other conditions. In the no-goal and delay conditions, participants were told to move the left hand forward and backward when the lines changed color (every 4–6 s). They were told to complete this movement within approximately 1.5 s. After the movement, they rested the left hand on the rail to wait for the next cue. In the no-goal condition, the movement of the virtual hand responded to the position of the real hand immediately. In the delay condition, the virtual hand responded to the position of the real hand after a 1000 ms delay. We used the delay condition to manipulate agency on an intermediate level (Farrer et al., 2013; Wen et al., 2015a) to examine the hypothesis that agency could contribute to body representation updating.

TABLE 1

Table 1. List of conditions in the experimental task.

In the goal-without-feedback and goal-with-feedback conditions, the two lines on the screen remained white in color, and a red block emerged from the left of the screen and along the upper line at a speed of 100 mm/s, every 4–6 s. The participant was instructed to move his or her hand forward to touch the flying object with the virtual hand. The task was easy to achieve, and all participants were able to touch the flying object once they had practiced. For participants in the goal-without-feedback condition, the object continued to fly, even after being touched, but disappeared when it passed the center of the screen. In the goal-with-feedback condition, the object stopped after being touched and disappeared 1 s later. The object disappeared after passing the center of the screen if participants did not touch it. The five conditions are listed in Table 1. The number of cues (color changing of the lines, flying object) in each trial of each condition was equal (36 in each trial); therefore, the number of hand movements in each trial should have been the same when participants followed the instruction.

In each trial, participants performed hand movements for 3 min. Previous studies observed proprioceptive drift after a visuo-motor adaption for 1.5–3 min (Tsakiris et al., 2006; Kalckert and Ehrsson, 2012); therefore, we believe 3 min should be enough to observe the rubber hand illusion. Thereafter, participants pointed to the perceived height of the left hand on a ruler, which was placed to their right, using the right index finger with their eyes closed. We did not measure pre- or post-drift for each trial. Instead, proprioceptive drift in the random condition served as a baseline, and responses in other conditions were compared with the baseline. After the 3 min hand movement, participants verbally rated two statements that referred to the feeling of ownership (“I felt as if the virtual hand was my hand”) and agency (“the virtual hand moved just like I wanted it to, as if it was obeying my will”) on a 7-point Likert scale (−3: totally disagree; 0: uncertain; +3: totally agree). The two statements were modified from those used in prior studies examining the rubber hand illusion (Botvinick and Cohen, 1998; Kalckert and Ehrsson, 2012).

Procedure

Participants were tested individually in a quiet room, seated in front of a 27″ LCD monitor. The height of the chair was adjusted to ensure that participants’ eyes were approximately 350 mm above the surface of the monitor. Participants’ arms and the bottom of the monitor were covered with a blanket, to ensure that participants could not see the movement of their own arms. Prior to each trial, the initial position of the participant’s left hand was adjusted to ensure that the virtual hand appeared above the real hand. Each condition comprised three blocked trials, resulting in a total of 15 trials. Prior to each condition, participants received an explanation regarding the task and practiced the movement for 1 min. After each trial, participants detached the 3-D position sensor from the finger, exposed the hand, and took a 3 min break. The order of conditions was randomized between participants. The experiment lasted for approximately 110 min on average.

Results

Proprioceptive Drift

Average proprioceptive drift (difference from actual height) and standard errors for each condition are shown in Figure 2. We conducted a repeated measures analyses of variance (ANOVA) to examine the influence of condition on proprioceptive drift. The main effect of condition was significant (F_(4,76) = 23.58, p < 0.001, $η_{p}^{2}$ = 0.55; random: M = 68.18 mm, SD = 43.46 mm; delay: M = 79.77 mm, SD = 50.82 mm; no-goal: M = 95.92 mm, SD = 46.80 mm; goal-without-feedback: M = 100.03 mm, SD = 45.82 mm; goal-with-feedback: M = 111.13 mm, SD = 50.51 mm). For post hoc comparisons between conditions, the significance level was set at 0.005 according to the Bonferroni correction. Proprioceptive drift in the no-goal, goal-without-feedback, and goal-with-feedback conditions differed significantly from those in the random and delay conditions (random and no-goal conditions, t₍₁₉₎ = −4.70, p < 0.001; random and goal-without-feedback conditions, t₍₁₉₎ = −5.78, p < 0.001; random and goal-with-feedback conditions, t₍₁₉₎ = −6.49, p < 0.001; delay and no-goal conditions, t₍₁₉₎ = −4.34, p < 0.001; delay and goal-without-feedback conditions, t₍₁₉₎ = −5.33, p < 0.001; delay and goal-with-feedback, t₍₁₉₎ = −6.82, p < 0.001). More importantly, the proprioceptive drift in the goal-with-feedback condition was significantly larger relative to that observed in the no-goal condition (t₍₁₉₎ = 3.91, p = 0.001) but did not differ from that observed in the goal-without-feedback condition (t₍₁₉₎ = 2.82, p = 0.011). The difference between the no-goal and goal-without-feedback conditions was nonsignificant (t₍₁₉₎ = −1.02, p = 0.320).

FIGURE 2

Figure 2. Mean proprioceptive drift in each condition in Experiment 1. Error bars represent standard errors. Proprioceptive drift in the goal-with-feedback condition differed significantly from that of the no-goal condition, but the difference between the goal-without-feedback and no-goal conditions was nonsignificant.

Ownership and Agency Ratings

The means and standard errors for ownership and agency ratings for the virtual hand are presented in Figure 3. With respect to the sense of ownership rating, the repeated measures ANOVA revealed a significant main effect of condition (F_(4,76) = 53.93, p < 0.001, $η_{p}^{2}$ = 0.74; random: M = −1.97, SD = 1.07; delay: M = −1.10, SD = 1.68; no-goal: M = 1.42, SD = 1.60; goal-without-feedback: M = 1.10, SD = 1.50; goal-with-feedback: M = 1.32, SD = 1.46). A significance level of 0.005 was used for post hoc multiple comparisons according to the Bonferroni correction. Ratings in the no-goal, goal-without-feedback, and goal-with-feedback conditions were significantly higher relative to those observed in the random and delayed conditions (random and no-goal conditions, t₍₁₉₎ = −9.77, p < 0.001; random and goal-without-feedback conditions, t₍₁₉₎ = −9.39, p < 0.001; random and goal-with-feedback conditions, t₍₁₉₎ = −10.14, p < 0.001; delay and no-goal conditions, t₍₁₉₎ = −7.52, p < 0.001; delay and goal-without-feedback conditions, t₍₁₉₎ = −6.26, p < 0.001; delay and goal-with-feedback, t₍₁₉₎ = −7.21, p < 0.001). The differences between the random and delay conditions, and the differences between no-goal, goal-without-feedback, and goal-with-feedback conditions were nonsignificant (random and delay: t₍₁₉₎ = −2.01, p = 0.059; no-goal and goal-without-feedback conditions: t₍₁₉₎ = 2.65, p = 0.016; no-goal and goal-with-feedback conditions: t₍₁₉₎ = 0.69, p = 0.500; and goal-without-feedback and goal-with-feedback conditions: t₍₁₉₎ = −1.66, p = 0.114).

FIGURE 3

Figure 3. Mean ratings for sense of ownership and agency in each condition in Experiment 1. Error bars represent standard errors. Sense of ownership ratings in the conditions with synchronous movement of the virtual hand were significantly higher relative to those observed in the conditions in which movement of the virtual hand was delayed or completely random. The sense of agency rating in the delayed condition was significantly higher relative to that observed in the random condition, and the ratings in the conditions with synchronous movement of the virtual hand were significantly higher relative to those observed in the delayed and random conditions.

With respect to sense of agency ratings, the main effect of condition was significant (F_(4,76) = 114.39, p < 0.001, $η_{p}^{2}$ = 0.86; random: M = −2.27, SD = 0.99; delay: M = −0.15, SD = 1.73; no-goal: M = 2.62, SD = 0.62; goal-without-feedback: M = 2.33, SD = 1.16; goal-with-feedback: M = 2.40, SD = 0.96). A significance level of 0.005 was used for post hoc multiple comparisons according to the Bonferroni correction. The rating in the random condition was significantly lower relative to those observed in the other conditions (random and delay conditions, t₍₁₉₎ = −4.47, p < 0.001; random and no-goal conditions, t₍₁₉₎ = −20.80, p < 0.001; random and goal-without-feedback, t₍₁₉₎ = −15.20, p < 0.001; random and goal-with-feedback, t₍₁₉₎ = −17.62, p < 0.001). The rating in the delay condition was significantly lower relative to those observed in the no-goal, goal-without-feedback, and goal-with-feedback conditions (delay and no-goal conditions, t₍₁₉₎ = −8.43, p < 0.001; delay and goal-without-feedback, t₍₁₉₎ = −7.94, p < 0.001; delay and goal-with-feedback conditions, t₍₁₉₎ = −8.78, p < 0.001). However, the differences between the no-goal, goal-without-feedback, and goal-with-feedback conditions were all nonsignificant (no-goal and goal-without-feedback conditions: t₍₁₉₎ = 1.69, p = 0.108; no-goal and goal-with-feedback conditions: t₍₁₉₎ = 1.49, p = 0.153; and goal-without-feedback and goal-with-feedback conditions: t₍₁₉₎ = −0.456, p = 0.654).

Discussion

The purpose of Experiment 1 was to examine the influence of goals and goal-attainment feedback on body representation updating, using the rubber hand paradigm with virtual reality stimuli. We found that when the participants moved their hands according to the instruction to touch a virtual object and received goal-attainment feedback, proprioceptive drift was larger relative to that observed when they moved without a specific goal and merely followed cued timings, indicating that goal-directed movement significantly enhanced body representation updating. In addition, when goal-attainment feedback was not provided, the instruction to touch a virtual object did not affect proprioceptive drift, indicating that congruence between goal and feedback was important in body representation updating.

The present study provided the first evidence indicating that goal-directed movement led to greater enhancement of body representation updating relative to that observed for movement without a specific goal. In the no-goal condition, the participants showed larger proprioceptive drifts relative to those in the random condition, indicating that changes in body representation occurred within a short period during the experimental task. Moreover, in the goal-with-feedback condition, proprioceptive drifts were significantly larger than those in the no-goal condition. When there was a specific goal, the expectation of goal achievement was probably generated on a high cognitive level, and when it was congruent with the goal-attainment feedback, this enhanced body representation updating, into which incorrect visual information was integrated to a greater extent.

Why would congruence between goals and goal-attainment feedback enhance interactions between motor control, vision, and proprioception in the rubber hand illusion? It is possible that sense of agency played an important role in this phenomenon. Prior research has shown that congruence between goals and goal-attainment feedback, besides that between motor commands and sensory feedback, exerted a strong influence on sense of agency (Metcalfe and Greene, 2007; Metcalfe et al., 2013; Wen et al., 2015b,c). Further, Tsakiris et al. (2006) found that sense of agency was important in integrating distinct body parts into coherent and unified bodily awareness. In addition, recent studies have shown that intention of action (without actual movement) can also produce a weaker form of the rubber hand illusion with a brain-computer-interface (Perez-Marcos et al., 2009; Slater et al., 2009), revealing the potential role of agency in construction of body representation. Therefore, goals and goal-attainment feedback probably enhanced body representation updating via sense of agency in our experiment. We tested this hypothesis further in Experiment 2, comparing the effects of goal-directed movement between the active and passive conditions. Prior research has shown that intention regarding action plays a critical role in sense of agency, and people experience a strong sense of agency only during active movement (Tsakiris et al., 2006; Kalckert and Ehrsson, 2012). If sense of agency plays an important role in the effects of goal-directed movement on body representation updating during active movement, this effect should be greater relative to that observed with passive movement.