Neural computational model for visuo-tactile bilateral interactions

Multisensory integration is the brain’s ability to combine information from different sensory modalities, a process that is fundamental for accurate perception and adaptive behaviour. A novel computational model has been designed to simulate visuo-tactile integration within a biologically inspired neural network framework. It contains realistic synaptic architectures, including feedforward, lateral, inhibitory and cross-modal connections, and utilizes non-linear sigmoidal activation functions to approximate neuronal dynamics. Simulations are performed under various stimulus conditions, comprising unisensory tactile, unisensory visual, and multisensory visuo-tactile presentations, as well as switch and repetition paradigms with varying interstimulus intervals. Analysis of reaction times revealed that multisensory stimulation elicited faster responses compared to unisensory conditions, confirming the principle of multisensory facilitation. Moreover, the model successfully reproduced key phenomena such as sensory masking and neural interference, underlying mechanisms of cross-modal attention and integration.