Inverse dynamics model in the cerebellum

Although the input and output, and the functional roles of different regions of the cerebellum are vastly different, the neural circuit in the cerebellar cortex is rather uniform. Given this histological uniformity, Kawato and Gomi (1992) proposed a coherent model of cerebellar motor learning based on the feedback-error-learning scheme. In that model, we hypothesized that the cerebellum acquires internal models of controlled objects such as inverse dynamics model. Shidara, Kawano, Gomi and Kawato (1992,1993) reconstructed temporal firing pattern of Purkinje-cells of the ventral paraflocculus using inverse-dynamics representation based on ocular following responses in monkeys. This experimental data supports our computational model. Similar reconstruction of messy fiber firing patterns disproves local representation models for messy fiber inputs to Purkinje cells. Finally, adaptive significance of low firing frequency of climbing fibers is discussed.