The regions of occipitotemporal cortex that process images of body parts evolved over millions of years in response to the bodies humans actually have. Virtual reality now allows people to inhabit body forms with no evolutionary precedent, which raises a question about how fixed those representations are. Researchers at the School of Psychological and Cognitive Sciences at Peking University tested it by training participants to fly using virtual wings.
The training protocol was deliberately minimal. Across four sessions over seven days, participants moved their upper limbs to flap wings and complete flight tasks, with arm movements tracked and translated into wing movements that changed altitude in virtual scenes. Initially they could see only the wings themselves, yet they learned to associate their physical movements with changes in flight altitude. No sensory feedback from the wings was provided.
Comparing functional magnetic resonance imaging before and after training, that learned association alone changed how the brain processed wing shapes. Viewing static wing images produced stronger category-specific activation in bilateral occipitotemporal regions. Multivariate pattern analysis showed wing representations in the right occipitotemporal cortex becoming significantly more similar to those of human upper limbs. Coupling between the right occipitotemporal cortex and the motor-related frontoparietal network also increased, specifically for wings.
The authors conclude that this cortical region can integrate non-human limbs into the body's representational system without evolutionary change or lifelong sensorimotor training. Published in Cell Reports, the finding bears on the design of prosthetics, wearable robotics and brain-computer interfaces, where the question of whether the brain will accept an unfamiliar effector has usually been treated as an empirical unknown.