A Virtual Reality Full Body Illusion Improves Body Image Disturbance in Anorexia Nervosa

Anorexia nervosa patients overestimate the size of their own body by a measurable and consistent margin. This is not just a figure of speech — it is a number that can be recorded. A team in the Netherlands put thirty patients inside a virtual reality headset and, in under an hour, changed that number. Body image disturbance in anorexia nervosa is not simply a matter of low self-esteem or a distorted opinion. It represents a measurable mismatch between the physical body and the brain's internal model of it — what researchers call a multisensory body representation. Patients who are clinically underweight still experience their bodies as too large, and this distorted representation is linked not only to the onset of the illness but also to prognosis, maintenance, and relapse. Keizer and colleagues frame the distortion as genuinely perceptual: overestimation occurs across multiple senses simultaneously. Studies show altered haptic perception, disturbed visual-proprioceptive integration, abnormal body-scaled action, and decreased interoceptive awareness — the brain's sense of the body is incorrect in several channels at once, not just one. This raises a clinical puzzle at the center of this research. If the distortion is so entrenched, can it even be changed? The team had a reason to think it might be possible. They had previously conducted the Rubber Hand Illusion in anorexia nervosa patients — a classic experiment where synchronous stroking of a visible fake hand and the participant's hidden real hand produces a vivid sense of ownership over the fake limb. That earlier study found the illusion normalized hand-width overestimation in patients with anorexia nervosa. However, the hand is not where the emotional weight resides. Shoulders, abdomen, and hips are the body parts that drive distress. So, Keizer and colleagues asked whether they could scale the rubber hand effect up to an entire virtual body. The Full Body Illusion operates on the same logic as the rubber hand version but fully immerses the participant. Participants wear an Oculus Rift headset, look down, and see a first-person view of a neutral female avatar — age set to 25, waist-to-hip ratio of 0.75, waist circumference of seventy-one point eighty-three centimeters. The experimenter then strokes the participant's real abdomen with a soft brush for ninety seconds. The brush is connected to a motion sensor that feeds the movement into the virtual reality scene in real time, so the participant simultaneously feels the stroke on their body and sees it happen on the avatar's body. When those two signals arrive in sync, the brain begins to assign ownership of the virtual body to itself. The latency from brush movement to visible change was about one hundred milliseconds — fast enough that the synchrony feels genuine. Each participant experienced two blocks. In the synchronous condition, the real and virtual stroking happened at the same time. In the asynchronous control, they were decoupled: the experimenter froze the virtual reality image while stroking the participant, then replayed the recorded movement afterward, so the two sensations never overlapped. This control was designed to test whether synchrony specifically — not just the experience of being stroked near a virtual body — was doing the work. Block order was counterbalanced across participants. The sample included thirty women with anorexia nervosa and twenty-nine healthy control women. Body size estimates were collected at three points: before the illusion, immediately after each condition, and at a follow-up roughly two hours and forty-five minutes later. The estimates themselves were concrete and body-centered — participants placed adhesive markers on a wall to indicate perceived width and formed a piece of string on the floor to represent perceived circumference for shoulders, abdomen, and hips. They were instructed to estimate based on how they subjectively feel about their body size, not on stored factual knowledge. After each illusion block, they also completed the Embodiment Questionnaire, a twenty-item scale yielding scores on ownership, location, and agency — how much they felt the virtual body belonged to them, where they felt themselves to be, and how much they felt they controlled it. The results are striking. Before the illusion, anorexia nervosa patients were overestimating abdomen circumference by more than sixty percent — sixty point forty-six percent, on average. Hip circumference overestimation was just over forty percent. These are not small errors. They reflect a body map that is calibrated substantially larger than the physical body it represents. After the Full Body Illusion, both figures dropped. Abdomen circumference overestimation fell from sixty point forty-six percent to forty-eight point ninety-four percent. Hip circumference fell from forty point thirty-three percent to thirty point twenty-five percent. The effects were strongest for circumference estimates — the three-dimensional measure of body volume — rather than width alone. That pattern is significant, because circumference is arguably closer to how people experience bodily presence than a flat width measurement. Healthy controls showed changes too, but with a different pattern. Their baseline overestimation was much smaller to begin with — abdomen circumference at twenty-one percent versus sixty in the anorexia nervosa group — and their changes appeared more at follow-up than immediately after the illusion. Effect sizes tell the story clearly: in the anorexia nervosa group, effect sizes for significant follow-up reductions all exceeded a Cohen's d of one point zero seven. In healthy controls, all analogous effects were below zero point six nine. The two groups are not doing the same thing. Now here is the complication. The reductions in anorexia nervosa patients appeared not only after the synchronous condition but also after the asynchronous control. Post-synchronous and post-asynchronous size estimates did not significantly differ from one another for most body parts. That is unexpected. If the mechanism were specifically the embodiment produced by synchronous visuo-tactile matching, you would predict the asynchronous condition to show no effect. Instead, it did show an effect. Keizer and colleagues flag this as an open question rather than explaining it away — and it is worth contemplating, because it suggests the mechanism might involve something broader than synchrony alone. Perhaps being in virtual reality and attending to a body at all is part of what recalibrates the brain's estimate. The data do not yet resolve this uncertainty. What the data do clarify is the durability question. In the subset who returned for follow-up — nine anorexia nervosa patients and twenty-six healthy controls, assessed an average of about one hundred fifty-six minutes after the illusion — the reductions in the anorexia nervosa group held. Shoulder circumference at follow-up was eleven point ninety-four percent overestimation. Hip circumference was nineteen point ninety-four percent. Both were significantly lower than baseline. For shoulder width, the effect size at follow-up was a Cohen's d of one point two zero. The brain map did not snap back. That durability is the finding that changes the conversation. Body image disturbance in anorexia nervosa has long been treated as one of the most resistant features of the disorder — clinically entrenched, emotionally fortified, and slow to respond to treatment. What Keizer and colleagues show is that the underlying representation is, in their words, flexible. It can be changed in a single experimental session, and the change persists for hours. However, this does not mean virtual reality is a treatment. Keizer and colleagues are careful here. This is a proof-of-concept study, not a clinical trial. The follow-up sample in the anorexia nervosa group was just nine people. The illusion induced changes in a controlled lab setting. The questions of how these effects transfer to clinical reality, and how to make them durable over days and weeks rather than hours, remain entirely open. The mechanistic questions are equally unresolved — whether what is driving the change is visual recalibration, proprioceptive updating, or the specific experience of feeling ownership over a body of a different size, the study cannot yet determine. The result from the asynchronous condition needs an explanation before the mechanism can be confidently identified. But here is what this research conclusively establishes. The body map in anorexia nervosa is not locked. Even for emotionally charged body parts — the abdomen and hips — the brain's model of size can be updated through a multisensory experience. That reframes the distortion from an immovable symptom to a targetable process. The next questions focus on mechanism and translation. But the foundational question — can you change the number? — now has an answer. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.