Hallucinations Skew Perception of Crowd Size

Summary: Presence hallucinations—reports of an unseen person nearby—are linked to an increased tendency to overestimate the number of people in a scene. Researchers at EPFL show this overcounting is specific to social stimuli (people) and does not occur for inanimate objects such as boxes. Using a combined virtual reality and robotics platform, the team developed an objective method to induce and measure hallucination susceptibility and produced an online version of the test that patients can complete at home. These advances offer a scalable way to monitor hallucination risk and could help identify early signs of cognitive decline in Parkinson’s disease.

Key Facts:

  1. Presence Hallucinations and Social Overcounting: People who experience presence hallucinations, including many with Parkinson’s disease, tend to overestimate how many people are present in a scene. This effect does not extend to counting inanimate objects, highlighting a socially specific perceptual bias.
  2. Technodelics for Objective Assessment: The study introduces a controlled method that combines virtual reality and robotics—referred to as “technodelics”—to induce presence hallucinations and obtain objective measures of susceptibility.
  3. Online Numerosity Test for Wider Reach: A simplified, web-based version of the numerosity experiment enables patients to assess their susceptibility to hallucinations at home, expanding access to low-cost, objective screening tools.

From an evolutionary viewpoint, humans and many animals are biased to over-detect potential social threats: it is safer to assume there are more agents nearby than to underestimate them. EPFL neuroscientists have now shown that this built-in tendency is amplified in people who experience presence hallucinations—particularly those with Parkinson’s disease—so that they overcount people to a greater degree than individuals without such hallucinations.

The research clarifies that this added overcounting is specific to social perception. When participants were asked to estimate how many boxes appeared in a scene, those experiencing presence hallucinations did not overcount more than others. That selective effect points to disrupted social-perceptual processes rather than a general counting impairment.

Published in Nature Communications, the study challenges the narrow view of Parkinson’s disease as solely a movement disorder and highlights perceptual changes—especially for social stimuli—that can accompany the condition. Olaf Blanke, head of EPFL’s Laboratory of Cognitive Neuroscience (Neuro X), emphasizes that presence hallucinations appear to alter how the brain represents other people, and that this can produce measurable changes in behavior such as exaggerated numerosity judgments for people.

Presence hallucinations describe the subjective sense that someone is nearby when no one is present. They are often milder than vivid visual hallucinations but can occur early in Parkinson’s disease, sometimes before a formal diagnosis, and are recognized as an early marker of increased risk for cognitive decline in this population.

The authors propose that when an invisible presence is experienced—whether arising from disease mechanisms or induced experimentally—that implicit extra “presence” becomes incorporated into the perceptual and decision processes used when estimating social numbers. In other words, an internally generated sense of another person is effectively added to what the observer believes they saw, but this addition is limited to social agents.

The numerosity experiment and technodelics

To test their hypothesis, the team developed a “technodelics” platform that merges immersive virtual reality scenes with a robotic device capable of subtly perturbing bodily feedback. In the human numerosity experiment, participants were shown brief (200 ms) 3D scenes containing five to eight virtual people—too many and too fast for reliable serial counting. Separately, the robotic device induced presence hallucinations by delivering tactile feedback that was out of sync with participants’ own movements, creating a mismatch in bodily sensations known to elicit a felt presence.

By combining these elements, the researchers could induce and quantify hallucinatory states and read out their impact on social counting. Healthy individuals exposed to the induced presence also showed increased overcounting, demonstrating that the method can produce measurable changes in perception. As lead author Louis Albert explains, this approach turns subjective experiences into objective, repeatable measurements that can assess hallucination susceptibility at a given time.

Compared with conventional screening—relying on self-report, clinical interviews, or questionnaires—this technodelics platform offers a more automated and implicit route to determine who is prone to presence hallucinations.

Monitoring hallucinations at home

To make this work widely accessible, the team also created a simplified online version of the numerosity test that patients can complete on a computer or tablet without specialized equipment or trained staff. This web-based test can be performed independently at home, lowering barriers for patients who live far from clinics or in regions with limited resources.

In an online sample of about 170 Parkinson’s patients, 69 reported presence hallucinations. The online task reproduced the lab findings: patients with presence hallucinations exhibited larger overestimations of people, with some reporting as many as 11 or more people when only eight were shown. These results demonstrate the online test’s potential as an inexpensive screening and monitoring tool.

Co-author Fosco Bernasconi notes that reliable strategies to detect presence hallucinations could enable clinicians to identify patients at higher risk for cognitive decline and offer earlier interventions or closer monitoring.

EPFL’s Neuro X Institute, founded in 2022, brings together neuroscience, neural engineering, and neurocomputation to accelerate translational research. The institute’s multimodal approach—combining behavioral experiments, robotics, virtual reality, and computational models—was key to developing the technodelics platform described in this study.

From body ownership to technodelics

The robotic task at the heart of this discovery arose from earlier work on body ownership and embodiment. A decade ago, researchers using robotic perturbations to study how sensory signals create a sense of one’s own body noticed participants sometimes reported eerie feelings of a companion presence. Rather than dismissing these reports, the team refined the setup and harnessed it as a reproducible means to provoke presence hallucinations and probe their perceptual consequences.

About this social perception and hallucinations research news

Author: Hillary Sanctuary
Source: EPFL
Contact: Hillary Sanctuary – EPFL
Image: The image is credited to Neuroscience News

Original Research: The findings will appear in Nature Communications