States Are A-Changing, Complex Spikes Proclaim

The cerebellum has been viewed historically as involved exclusively in motor control. However, understanding specific cerebellar processes has remained unsettled for a long time. Among the numerous hypotheses formulated to define cerebellar function one of the more successful has proved to be the forward internal model theory, which states that the cerebellum computes predictions of motor command consequences and compares them with sensory feedback to provide prediction errors (PEs). One of the remaining challenges facing this theoretical framework relates to understanding the role of the complex spike (CS) discharge in cerebellar function. The canonical view is that CSs encode the errors required for online motor control and to update the forward internal model. However, a growing body of evidence challenges this assumption and emphasizes the need for more comprehensive models of CS function. Also, there is increasing evidence that the cerebellum is involved in non-motor domains in a manner similar to motor control, raising the question whether the forward internal model framework can be generalized across all functional domains and be updated to offer a reasesment of CS function. In this review, we propose a generalized forward model in which the cerebellum performs the same computations across all functional domains and across the hierarchy of the CNS. In this generalized model, the CS discharge acts as a state change detector that selects the appropriate forward models expressed in the simple spike (SS) activity.