I will present a progress report on a new approach to understand how the cerebellum works and learns. To provide functional context, I will briefly assert our current model of cerebellar motor learning for pursuit eye movements. Next, I will outline a strategy to define how the cerebellar circuit computes by identifying cell-types from extracellular recordings in the cerebellar cortex. I will tell you how far we have gotten with cell-type identification in monkeys, with a little help from our rodent friends. We have identified the discharge properties during smooth pursuit eye movements of “units” that we are quite confident originate from: Purkinje cells, basket cells, Golgi cells, unipolar brush cells, and mossy fibers. We do not record from granule cells, sadly. I will tease you with what we have learned about the operation of the “canonical” cerebellar circuit during smooth pursuit. Along the way, I will challenge current doctrine in multiple ways: (i) we don’t find evidence of Purkinje cell synchrony; (ii) we don’t find evidence of climbing fiber spillover to MLIs; (iii) we are disappointed that Golgi cells seem to be pretty useless for computations. There is still much to do, but I want to end with the old-fashioned idea that we should think of the original, canonical cerebellar circuit as the primary circuit and many (or all) of the “modern” emendations to the original cerebellar circuit as secondary.
Chris de Zeeuw, Martijn Schonewille