Task dependent cortico-cerebellar responses to delayed visual movement feedback

AbstractForward models in the brain issue predictions of sensory movement consequences that can establish self-other distinction through comparison with actual feedback. But forward models can also be updated by sensory prediction errors, as in sensorimotor adaptation. Disentangling the neuronal correlates of processes related to sensorimotor incongruence detection from those related to adaptation can be challenging. Here, we approached this challenge with a novel, virtual reality based hand-target tracking task that allowed us to manipulate the behavioral relevance of delayed visual movement feedback, while recording hemodynamic brain activity with functional magnetic resonance imaging (fMRI). Participants performed continuous right hand grasping movements to track a target oscillation with either their real, unseen hand or with a glove-controlled virtual hand. The movements of the virtual hand were delayed in a roving oddball fashion. Therefore, tracking with the virtual hand (VH task) required visuomotor adaptation, whereas tracking with the real hand (RH task) required ignoring visuomotor delays. VH > RH task execution produced stronger activity in the left posterior parietal cortex (PPC) and the bilateral extrastriate visual cortex. Delays correlated with activity in several regions, prominently including right temporoparietal regions, in both tasks. Crucially, the cerebellum showed a stronger delay dependent activation, and increased functional connectivity with the PPC, in the VH > RH task. Delay changes and errors correlated with activity in the anterior insulae (AI), and more strongly so in the VH>RH task. Thus, the instructed behavioral relevance of delayed visual movement feedback enhanced responses of the cerebellum and PPC, and their communication, likely for visuomotor adaptation. In contrast, temporoparietal regions compared predicted and actual visual movement feedback irrespective of its behavioral relevance, while the AI signaled visual mismatches particularly when these were behaviorally relevant.