Grant: Project Grant

Every time we reach for a coffee cup or swing a tennis racket, the brain faces a core computational problem: Which motor memory should I retrieve? Does it need refining? Or is this situation so new that I need to build a memory from scratch? For decades, these decisions were assumed to be governed primarily within sensorimotor circuits. Our recent work challenges this view. Using advanced neuroimaging methods that characterize the evolving geometry of large-scale brain network activity, we found that the Default Mode Network (DMN)-classically implicated in autobiographical memory and internally oriented cognition-plays a central and previously underappreciated role in motor learning. This result suggests that learning and retrieving a motor skill may share fundamental neural machinery with encoding and recalling a personal experience. Our proposed project will determine how, and through which circuits, these computations are implemented. We will integrate our neuroimaging approach with the recently developed Contextual Inference (COIN) computational framework, which formalizes learning as the interaction of three separate processes: memory expression (retrieval), updating (error-driven refinement), and creation (formation of new memories). We will test the hypothesis that these processes map onto dissociable neural systems: that the DMN preferentially supports memory expression, sensorimotor circuits implement error-based updating, and novel memory formation recruits domain-general mechanisms for segmenting experience into discrete events. This work will deliver a unified model of skilled behaviour with direct implications for rehabilitation-suggesting that many motor deficits may reflect a failure to retrieve the right context, not a failure to move.