A computational model of cerebellar saccadic control

This dissertation describes the development of a computational model, based on current knowledge of the anatomy and physiology of the cerebellum, to investigate possible strategies for saccadic processing. The model employs a novel and efficient method of neural simulation to create a detailed model, down to the molecular level, of a population of Purkinje cells and their interneuron inhibition. The model of the Purkinje cell population has approximately 300,000 compartments. Simulations are run on a moderately sized workstation (SGI R5000, 180MHz, 256Mb RAM). A typical simulation, which includes initialization and simulation of 8 different eye movements, takes approximately 12 hours. The model demonstrates that the firing patterns from a population of Purkinje cells can accurately encode the direction of a saccade and that a state machine (i.e. determination of subsequent motor commands based on previous motor commands and the current state of the motor system) is a robust strategy for the storage and execution of saccadic motor programs. The firing patterns of the model, during intact and lesion experiments, qualitatively match electrophysiological recordings in the cerebellum during saccades. The model suggests plausible explanations for the dysmetria seen in cerebellar lesion and Wallenberg's syndrome patients. The model simulates adaptation paradigms and shows that error correction is possible when climbing fiber input is periodic and contains no error signal. The present results demonstrate the ability of this modeling methodology to efficiently simulate complex behaviors without sacrificing important details.