Differences in visual capture of connected body parts in the mirror box illusion reflect local variations in cross-modal congruence and unimodal variance
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
In the mirror box illusion, synchronous multisensory stimulation of a hidden hand alongside a reflected hand in the mirror can lead to 'visual capture' – where participants experience their hidden hand in the location of the reflected ‘mirror hand’. This effect is attributed to two main factors influencing multisensory integration: (1) cross-modal congruence between visual, tactile and/or motor signals, and (2) lower unisensory variance of vision compared to proprioception, weighting the integrated outcome toward the 'visual' input ('mirror hand'). Previous studies have primarily explored how these factors influence visual capture of a single, stimulated body part (e.g., visual capture of a viewed ‘mirror hand’ during synchronous bimanual movements), with limited investigations into how adjacent, unstimulated body parts are represented. Given the tendency to experience the body as a unitary whole composed of connected body parts, what happens when optimal integration of immediate multisensory information leads to ‘disconnected’ positions for connected body parts? We explored this question in two experiments, where participants’ hands were placed at unequal distances on either side of a mirror to create a spatial disparity between the ‘visual’ and ‘proprioceptive’ estimates of the hand and forearm behind the mirror. In Experiment 1, we manipulated the degree of cross-modal congruence between the two parts by synchronously stroking either the hands or the forearms, in isolation. In Experiment 2, we manipulated visual input variance by having participants fixate on either their hand or forearm while they were stroked synchronously across both parts. Analyses of participants’ perceived positions of their hand and forearm revealed stronger ‘local’ effects of visual capture at the site of stimulation (Experiment 1) or foveation (Experiment 2), as participants were significantly more likely to perceive either part of their hidden arm ‘in the mirror’ when it was the location of stimulation/foveation. Moreover, stimulation/foveation of a given part significantly raised the likelihood of ‘fragmented visual capture’ of that part in the mirror (e.g., simultaneously perceiving the hand but NOT the forearm in the ‘mirror’ location, when the hands were stroked). Meanwhile, we found distinct ‘global’ effects in both experiments, as visual capture of the stimulated (Experiment 1) or foveated (Experiment 2) part of the arm significantly raised the likelihood of visual capture for the adjacent unstimulated/un-foveated part. Taken together, these results indicate that multisensory integration for the body is driven by a weighted integration of bottom-up properties of ‘online’ sensory information (e.g., cross-modal congruence; unimodal variance), with top-down information from a stored, ‘offline’ body representation of how the different body parts are connected.
Description
Keywords
Causal inference, Cross-modal congruence, Mirror-box illusion, Multisensory integration, Visual capture, Visuoproprioceptive integration