A neural network model of peripersonal space representation around different body parts

Abstract

The Peripersonal Space (PPS), the space immediately surrounding the body, is coded in a multisensory, body part-centered (e.g hand-centered, trunk-centered), modular fashion. This coding is ascribed to multisensory neurons that integrate tactile stimuli on a specific body part (e.g. hand, trunk) with visual/auditory information occurring near the same body part. A recent behavioral study, using an audiotactile psychophysical paradigm, evidenced that different body parts (hand and trunk) have distinct but not independent PPS representations. The hand-PPS exhibited properties different from the trunk-PPS when the hand was placed far from the trunk, while it assumed the same properties as the trunk-PPS when the hand was placed near the trunk. Here, we propose a neural network model to help unrevealing the underlying neurocomputational mechanisms. The model includes two subnetworks, devoted to PPS representations around the hand and around the trunk. Each subnetwork contains two areas of unisensory (tactile and auditory) neurons communicating, via feedforward and feedback synapses, with a pool of audiotactile multisensory neurons. The two subnetworks are characterized by different properties of the multisensory neurons. An interaction mechanism is postulated between the two subnetworks, controlled by proprioceptive neurons coding the hand position. Results show that the network is able to reproduce the behavioral data. Network mechanisms are commented and novel predictions provided.