BIVIP: a competitive neural network for bilateral visuo-haptic processing.

Visuo-haptic interactions are gaining interest in perception research because of their relevance to robotics and rehabilitation. While the benefits of integrating concurrent visuo-tactile cues are well-documented, the temporal dynamics of competitive interactions between vision and touch remain less understood. Building on extant empirical evidence of unilateral intersensory switching, we ask whether bilateral detection of visual and tactile stimuli can be explained by cross-sensory competition and explore the underlying sensorineural dynamics via a multi-level neural network model. Tested against results from a speeded reaction time task to unilateral and bilateral visual, tactile, and visuo-tactile targets, the network successfully predicts higher switch costs for sequential identification of unisensory stimuli of different modalities on the same hand rather than across hands. Our network also captures significant effects of inter-stimulus interval, trial type (i.e. repeat vs. switch), and modality type (i.e. visual, tactile, or visuo-tactile). Additionally, a sensitivity analysis on specific network connections differentiated between potential mechanisms underlying unisensory and multisensory processing of bilateral targets. Both network predictions and behavioral data reveal a slowdown in response to the repeated presentation of simultaneous visuo-tactile targets, suggesting that cross-sensory inhibition may contribute to offline influences of vision on bilateral tactile perception.