Circuit mechanisms to transform neural population dynamics for motor control

Abstract

We identify the input-output transformations performed by the cerebellar circuit and isolate the circuit properties that implement them. Extracellular recordings exploited new technology for neuron-type identification in part of the cerebellar cortex that is causal for an oculomotor behavior whose neural substrate is well-understood in monkeys. Data demonstrate how multiple classes of interneurons in the cerebellar circuit cooperate to transform incoming mossy fiber responses into Purkinje cell outputs. Computer models suggest that temporal decomposition in the granule cell layer enables flexible reconstruction of transformed output dynamics, regulated by cerebellar plasticity, to convert eye position-like inputs into predominantly velocity-driven Purkinje cell outputs. Convergence of mossy fiber inputs with diverse direction preferences would increase input layer dimensionality and enable learning of highly flexible direction tuning. The neural circuit mechanisms suggested by our data outline a cerebellar operating principle that enables flexible input-output transformations and could generalize across cerebellar and other brain areas.