ePoster, COSYNE 2023
Cerebro-cerebellar networks facilitate learning through feedback decoupling
Ellen Boven, Joseph Pemberton, Paul Chadderton, Richard Apps and Rui Ponte Costa
University of Bristol; Biomedical sciences; University of Bristol
- Montreal, Canada
- Neuroscience, Computational Neuroscience and
Abstract
Behavioural feedback is critical for learning in the cerebral cortex. However, such feedback is often not readily available. How the cerebral cortex learns efficiently despite the sparse nature of feedback remains unclear. Inspired by recent deep learning algorithms, we introduce a systems-level computational model of cerebro-cerebellar interactions. In this model a cerebral recurrent network receives feedback predictions from a cerebellar network, thereby decoupling learning in cerebral networks from future feedback. When trained in a simple sensorimotor task the model shows faster learning and reduced dysmetria-like behaviours, in line with the widely observed functional impact of the cerebellum. Next, we demonstrate that these results generalise to more complex motor and cognitive tasks. Finally, the model makes several experimentally testable predictions regarding (1) cerebro-cerebellar task-specific representations over learning, (2) task-specific benefits of cerebellar predictions and (3) the differential impact of cerebellar and inferior olive lesions. Overall, our work offers a theoretical framework of cerebro-cerebellar networks as feedback decoupling machines.
Topics
- cerebral cortex, search World Wide
- cerebellum, search World Wide
- cerebro-cerebellar interactions, search World Wide
- Behavioural feedback, search World Wide
- recurrent network, search World Wide
- feedback predictions, search World Wide
- sensorimotor task, search World Wide
- dysmetria-like behaviours, search World Wide
- inferior olive lesions, search World Wide
- deep learning algorithms, search World Wide
Details
- Session
- Poster Session II