Visual capture and the experience of having two bodies – Evidence from two different virtual reality techniques

Imagine you're standing in a lab. There's a paintbrush on your back, and an experimenter is drawing slow, irregular strokes across your shoulder blades. It's a completely ordinary sensation. Now, imagine that at the exact same moment you feel each stroke, you're watching it happen to two different bodies standing a few feet in front of you. Both of them are wearing your clothes, and both are being touched in perfect synchrony with you. Hold that scene for a second. What Heydrich, Dodds, Aspell and colleagues found when they ran exactly this experiment is that participants started to feel like they existed in more than one place at once. The brain's sense of "where am I, and which body is mine?" began to fracture. That result opens a window into something fundamental about how selfhood is constructed. Let's start with the clinical backstory because it's what makes the experimental question urgent. There is a condition called heautoscopy, documented in neurology and psychiatry for over a century, in which patients report being reduplicated. They feel that they exist at two or even more locations simultaneously. It's not a hallucination in the ordinary sense. Patients often describe profound confusion about which body is really theirs, an experience that clinicians have linked to disruptions in the normal integration of bodily signals. Heydrich and colleagues point to these cases as evidence that the brain normally enforces what they call the singularity of the self — one body, one location, one owner — and that this enforcement can fail. The puzzle was that laboratory experiments had never been able to reproduce that failure. Previous work using virtual reality to manipulate bodily experience, including the famous rubber hand illusion scaled up to full-body setups, could reliably shift where people felt they were, and could make a virtual body feel like their own. But participants still reported one self, even a displaced or relocated one. The singularity held. Heydrich and colleagues wanted to know whether that barrier could be broken experimentally, and if so, how. The key mechanism they tested is called visuo-tactile synchrony, which is the moment-by-moment matching of seen touch and felt touch. When you watch a body being stroked at the exact instant you feel the stroke on your own skin, the brain has a tendency to merge those two streams of information and treat the visible body as the source of the felt sensation. This is the engine behind the rubber hand illusion and behind full-body illusions in virtual reality. The innovation here was to point that engine at two bodies at once. They ran two separate studies. In the first study, nineteen participants stood in front of a camera and viewed a split-screen video of their own body, duplicated and placed left and right in a head-mounted display. The video was live, with a latency under 16 milliseconds — essentially real time. In asynchronous control blocks, an artificial delay of 400 milliseconds was introduced. The experimenter stroked participants' backs irregularly, about once every two seconds, while participants watched the same strokes appear on both virtual versions of themselves. The second study used seventeen different participants and replaced the video feed with stereoscopic three-dimensional computer-generated avatars, rendered to match each participant's gender and scaled to their body width and height. The baseline latency here was about 40 milliseconds, still with a 400-millisecond delay added for asynchronous blocks. Both studies also included control conditions where the virtual bodies were replaced with human-sized rectangular objects. The main finding held across both studies: synchronous stroking produced stronger self-identification with the two seen items than asynchronous stroking. In the first study, the body-synchronous condition scored a median of 6 on a seven-point agreement scale for the question "I felt as if the virtual body was my body," compared to a median of 4 in the body-asynchronous condition. In the second study, the same comparison ran 5.15 versus 2.10. That's a real, statistically significant difference in both cases, and it appeared even in the object control conditions — synchrony matters regardless of what's being stroked. But beyond self-identification, the two studies diverged, and that divergence is where things get interesting. In the first study, participants in the synchronous body condition also reported a greater sensation of actually having more than one body — not just identifying with two bodies, but feeling owned by, or owning, both of them simultaneously. This was statistically significant compared to asynchronous conditions and to the object controls. The second study did not produce this effect. The self-location measure, where participants physically ended up after being guided blindfolded from their starting position, told the same story. In the first study, participants drifted an average of 11.7 centimeters forward toward the virtual bodies during synchronous body stroking, compared to just 1.3 centimeters in the asynchronous body condition. In the second study, the forward drift was about 10 centimeters in both the synchronous and asynchronous body conditions — meaning synchrony made no difference to where participants thought they were. To make sense of the subjective reports, Heydrich and colleagues ran a principal component analysis on seven questionnaire items. Two components emerged: one they called "illusory touch," which is the sense that the felt strokes were coming from the virtual bodies, and one called "illusory drift," which is the sense of being pulled toward the bodies in space. The item about having more than one body loaded meaningfully on both components. That tells you something: the experience of reduplication isn't cleanly separable from either the tactile illusion or the spatial one. It straddles both. Across both studies, synchrony drove the touch component strongly, while the drift component was shaped by both synchrony and whether participants were watching bodies or objects. So why did the two studies produce such different results on the stronger measures? The authors built the comparison into the design precisely to answer this question. Video-based and computer-generated virtual reality differ in three ways that matter: what's shown on the display, the latency of the visual feedback, and the overall visual richness. The video feed shows participants their own actual body — same face, same clothes, same morphology — while the avatars, though size-matched, are generic. The baseline latency was 16 milliseconds in the first study versus 40 milliseconds in the second study. The video environment preserves the rich visual texture of the real world, while the computer-generated setup used a plain background. The latency difference is probably doing significant work. Research by Vroomen and Keetels, which the authors cite, suggests that delays greater than 20 milliseconds can be perceptually noticeable. A 40-millisecond baseline in the second study means the synchronous condition was already slightly degraded before the 400-millisecond asynchronous delay was added, compressing the effective contrast between the two conditions. When the authors re-analyzed the second study excluding the asynchronous trials entirely, the body-synchronous versus object-synchronous comparison did produce a significant forward drift — a t-value of 2.37, with a p-value of 0.03. So the effect of body ownership on self-location is there in the second study, but the asynchronous baseline was too close to the synchronous condition to show a clean within-subjects contrast. Visual fidelity matters too. Seventy-six percent of participants in the second study who reported identifying with both avatars said this happened simultaneously, which is a striking finding in its own right. But the video setup, with its exact visual match to the participant's own body, appears necessary for the stronger effects: the drift toward the bodies, and particularly the sensation of actually having more than one body. What does this all add up to? Heydrich and colleagues showed that the brain's apparent insistence on owning exactly one body is not a fixed boundary. Under the right multisensory conditions — synchronous touch, matched visual identity, tight timing — that boundary can be experimentally probed and partially shifted. The self-location drifts in the first study, the double-body ownership ratings, and the consistent synchrony-driven self-identification across both studies are converging evidence that the singularity of the self is a construction, not a given. The brain builds it moment by moment from multisensory signals. And when those signals are rerouted — pointed at two bodies at once, perfectly timed — the construction begins to come apart. What heautoscopy patients experience pathologically, this research begins to reproduce in a laboratory, one brush stroke at a time. 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.