A Snapshot seminar hosted by the School of Physiology, Pharmacology and Neuroscience

Keeping our movements precise depends on how well we adapt to constantly changing environments and the states of our bodies. The cerebellum has been known as a critical locus for this function, also called sensorimotor adaptation. Yet despite many studies on specific mechanisms in the network, there is still no good understanding of circuit-level information processing.

I'll talk about our recent work addressing this question by recording cerebellar mossy fibers (MF; network inputs) and Purkinje cells (PC; outputs) from rhesus monkeys performing repetitive saccades. The data showed significant cell-to-cell heterogeneity, much more prominent in PCs. Nonetheless, we found low-dimensional, manifold-like activity underlying the firings of both units. Furthermore, the PC manifolds showed much more selective representations of individual movement parameters than MF ones. The movement error-driven complex spike firing modulated the PC manifolds in an error-type-specific manner, correctly predicting changes in upcoming action. Surprisingly, a simple feed-forward network model accurately predicted the MF-to-PC transformation, suggesting that the network expands small variabilities in MF inputs to generate PC outputs. I will also discuss how the cerebellar multidimensional computations enable flexible movement coordination for sensorimotor learning.

Host: Paul Chadderton

Location: https://bristol-ac-uk.zoom.us/j/4798119105?pwd=MlZRZEt6aWhQb0NYT2pXd3N5aVRvQT09