Multisensory Integration in Action: a Multineuronal and Feedback-Control Approach

点击翻译按钮获取中文摘要

Project Summary: Multisensory processing is vital for daily activities such as walking and manipulating objects, yet much remains unknown about the neural mechanisms by which sensory information is integrated in the central nervous system to influence motor control. We address this knowledge gap by analyzing behavioral and multi-neuronal multi-area recordings in the cerebral cortex of Rhesus monkeys trained to perform a prolonged motor control task (the critical stability task (CST)) that cannot be performed without continuous sensory feedback (visual and/or tactile). Rhesus monkeys will perform the CST using hand movements or a brain-computer interface (BCI) to control a cursor, while we manipulate sensory feedback. Neural activity will be recorded from primary visual (V1), somatosensory (S1) and motor cortices (M1). Our motivating hypothesis is that cortical processing is highly flexible, and can be rapidly reconfigured based on the immediate sensory and motor context. Several specific predictions flow from this perspective. First, we predict that primary motor cortex (M1) will exhibit a strong sensory response during a motor task that requires ongoing sensory feedback. Second, we hypothesize that V1 neurons adopt tactile responses, and S1 adopts visual responses, when both are relevant for ongoing motor control. Third, we expect that altering the signal quality of one sensory modality will shift their relative contribution to neural responses, consistent with Bayesian estimation. Animals will perform the CST using BCI control as a more dramatic test of cortical flexibility. During BCI control, sensory responses should be reduced in M1, since the BCI decoder cannot distinguish sensory responses from motor commands, which would diminish the quality of control. If multisensory integration is reduced in M1 under BCI control, then it must occur elsewhere. We hypothesize that there will be an enhanced cross-modal sensory representation in the primary sensory cortices under BCI control, in comparison to hand control. We approach these questions through a collaboration that combines expertise in sensorimotor neurophysiology with expertise in computational modeling of multisensory integration. The findings of this research will improve the understanding of the neural mechanisms of multimodal sensory integration during continuous motor tasks, and will have clinical implications for BCIs and advanced prostheses design.