Characterizing Body Image Distortion and Bodily Self-Plasticity in Anorexia Nervosa via Visuo-Tactile Stimulation in Virtual Reality

Abstract

We combined virtual reality and multisensory bodily illusion with the aim to characterize and reduce the perceptual (body overestimation) and the cognitive-emotional (body dissatisfaction) components of body image distortion (BID) in anorexia nervosa (AN). For each participant (20 anorexics, 20 healthy controls) we built personalized avatars that reproduced their own body size, shape, and verisimilar increases and losses of their original weight. Body overestimation and dissatisfaction were measured by asking participants to choose the avatar that best resembled their real and ideal body. Results show higher body dissatisfaction in AN, caused by the desire of a thinner body, and no body-size overestimation. Interpersonal multisensory stimulation (IMS) was then applied on the avatar reproducing participant's perceived body, and on the two avatars which reproduced increases and losses of 15% of it, all presented with a first-person perspective (1PP). Embodiment was stronger after synchronous IMS in both groups, but did not reduce BID in participants with AN. Interestingly, anorexics reported more negative emotions after embodying the fattest avatar, which scaled with symptoms severity. Overall, our findings suggest that the cognitive-emotional, more than the perceptual component of BID is severely altered in AN and that perspective (1PP vs. 3PP) from which a body is evaluated may play a crucial role. Future research and clinical trials might take advantage of virtual reality to reduce the emotional distress related to body dissatisfaction.

Keywords: anorexia nervosa; body dissatisfaction; body image distortion; embodiment; interpersonal multisensory stimulation; virtual reality.

Figure 1

Creation of 3D Stimuli. (A) Example of three customized avatars built according to the participant’s body measures and pictures: an avatar that reproduced participant’s real body (avatar 0%), a thinner avatar (avatar −30%), and fatter avatar (+50%). (B) Example of avatars selection extracted from the continuum of avatars lying on a deck chair and increasing in size in steps of 3%, starting from the thinnest (−30%) to the fattest avatar (+50%).

Figure 2

Experimental procedure. After selecting the avatar most similar to their perceived body and the one most resembling their ideal body, participants were enrolled in embodiment blocks in which synchronous and asynchronous interpersonal multisensory stimulation (IMS) were applied to three different bodies (the perceived body, −15% thinner body, +15% fatter body). After each embodiment block participants repeated the perceived and ideal body tasks to measure the effects of the embodiment of different sized avatars on body dissatisfaction. Explicit and implicit measures of the embodiment illusion, as well as the emotional response after being exposed to a/synchronous touching of different sized avatars were recorded after each embodiment block. At the end of the experiment we asked participants to rate from a first-person perspective the three avatars in terms of similarity to their own body and overall attractiveness.

Figure 3

(A) Perceived/ideal body task. In separate blocks, participants choose the avatar which was the most similar to their own body (perceived body task) and the avatar which best resembled their ideal body (ideal body task) along a continuous of avatars presented from a third-person perspective (3PP). Each task comprised two trials presented in counterbalanced order: in one trial participants started the selection from the thinnest avatar (upper part of panel A), in the other from the fattest one (lower part of panel A). (B) Embodiment procedure. During the embodiment procedure a three minute of a/synchronous visuo-tactile stimulation was delivered. During the embodiment participants observed one of three different avatars from a first-person perspective (1PP). A virtual ball was programmed to touch the avatar on three different spots around the belly button in eight different ways (single touches and stroking movements).

Figure 4

Emotional response after embodiment. Graph showing the effect of the interaction between avatar size (−15%; 0%; +15%) and group (healthy controls—HC; patients with anorexia nervosa—AN) on the emotional scale ranging from 0 (very negative emotions) to 100 (very positive emotions). Error bars represent standard error of mean. * = p < 0.05, § = marginally significant (p = 0.057).

Figure 5

(A) Avatars’ similarity ratings. Graph showing the main effect of Avatar size (−15%; 0%; +15%) on similarity ratings given during the observation of the avatars from a 1PP. (B) Avatars’ attractiveness ratings. Graph showing the effect of the interaction between Avatar size (−15%; 0%; +15%) and Group (healthy controls, HC; patients with anorexia nervosa, AN) on attractiveness ratings given during the observation of the avatars from a 1PP. Error bars represent standard error of mean. ** = p < 0.001, * = p < 0.05, § = marginally significant, i.e., p = 0.058.

Figure 6

Correlations between +15% Avatar emotional response and symptoms severity. Scatterplots showing correlations between emotional changes after the exposure to the +15% Avatar and severity symptoms scores in the two groups. In the group of patient with anorexia nervosa (AN), the +15% Avatar emotional index correlated significantly with scores at the Body Shape Questionnaire (BSQ) and with the global severity index of the Body Uneasiness Test (BUT-GSI), while correlations were not significant in the healthy control (HC) group.