Neural mechanisms of iterative learning – Meca-Iter

A broad range of motor and cognitive behavioural skills are acquired, adapted and perfected through practice. It has proven very challenging to explain such behaviours by demonstrating the implementation of a sufficient algorithm within neural circuitry. There is strong evidence that the cerebellum is centrally involved in iterative learning of coordinated, rapid movements, and for the following reasons offers a promising experimental preparation for studying iterative learning: it has well characterised behaviours that have been localised to compact regions of the cerebellar cortex; is embedded in a short sensory-motor loop and must operate under strict temporal constraints; and detailed theories of cerebellar learning have been proposed. In sum, we know where, when and what to look for. Subsequent to our proposal of a new theory of iterative learning in the cerebellum, we set out to test the theory by obtaining longitudinal, multielectrode recordings of cerebellar activity during eyeblink conditioning, a key model cerebellar behaviour. We made several unexpected observations, notably a disconnect between behavioural learning and changes of Purkinje cell simple-spike firing, and our data suggest novel plasticity mechanisms that could modulate cerebellar and olivary activity during the motor command. The Barbour group will employ our chronic multielectrode recording techniques to clarify the discrepancies in the literature, investigate the role of complex spikes arising during the motor command and study the influence of molecular layer interneurones. The data will be analysed and modelled in an established collaboration with the Hakim group. We expect our project to lead to a clear understanding of the cellular and network mechanisms underlying this model of iterative motor learning. The insight gained should inform studies of other brain areas and may also suggest hypotheses for roles of the cerebellar circuitry in cognitive iterative learning.