Conference on Cognitive Computational Neuroscience

During task execution cortical dynamics must bridge sensory cues with future behavioural outcomes. However, how cortical networks acquire such task-specific dynamics remains unclear. Here we propose that the cerebellum drives cortical dynamics to enable rapid task acquisition via cortico-cerebellar loops. We model cerebellar networks that are tuned through timing rules to provide cortical networks with task-outcome predictions. The model readily generalises to cognitive-based tasks and supports experimental evidence for a cerebellar role in working memory. In particular, we show that ablation to the cerebellar network in our model leads to leaky cortical dynamics and a resultant loss of task selectivity in delayed association and evidence accumulation tasks, consistent with recent optogenetic studies. Our results also suggest that local cortical connectivity is relatively stable and that learning occurs primarily in the cerebellum to enable task acquisition. Finally, we use our model to introduce a systems consolidation theory in which task information is gradually transferred from the cerebellum to the cortex. In summary, our results suggest that the cerebellum plays a critical role in enabling cognitive task-specific dynamics in the cortex.