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 cerebro-cerebellar task-specific representations over learning, task-specific benefits of cerebellar predictions and the differential impact of cerebellar and inferior olive lesions. Overall, our work offers a theoretical framework of cerebro-cerebellar networks as feedback decoupling machines.
| Original language | English |
|---|---|
| Article number | 51 |
| Journal | Nature Communications |
| Volume | 14 |
| Issue number | 1 |
| DOIs | |
| Publication status | Published - 4 Jan 2023 |
Bibliographical note
Funding Information:We would like to thank the Neural & Machine Learning group, Paul Anastasiades, Paul Dodson, Conor Houghton, Laurence Aitchison, Cian O’Donnell, James M. Shine, Max Jaderberg, Nadia Cerminara and Jasmine Pickford for useful feedback. We would also like to thank Samia Mohinta and Milton Llera Montero for their help with setting up model analysis and training. J.P. was funded by an EPSRC Doctoral Training Partnership award (EP/R513179/1), E.B. by the Wellcome Trust (220101/Z/20/Z), P.C. by the Wellcome Trust (209453/Z/17/Z) and R.P.C. by the Medical Research Council (MR/X006107/1). This work made use of the HPC system Blue Pebble at the University of Bristol, UK.
Funding Information:
We would like to thank the Neural & Machine Learning group, Paul Anastasiades, Paul Dodson, Conor Houghton, Laurence Aitchison, Cian O’Donnell, James M. Shine, Max Jaderberg, Nadia Cerminara and Jasmine Pickford for useful feedback. We would also like to thank Samia Mohinta and Milton Llera Montero for their help with setting up model analysis and training. J.P. was funded by an EPSRC Doctoral Training Partnership award (EP/R513179/1), E.B. by the Wellcome Trust (220101/Z/20/Z), P.C. by the Wellcome Trust (209453/Z/17/Z) and R.P.C. by the Medical Research Council (MR/X006107/1). This work made use of the HPC system Blue Pebble at the University of Bristol, UK.
Publisher Copyright:
© 2023, The Author(s).