Consensus Paper: Models of Cerebellar Functions

For a long time, from the nineteenth century to most of the twentieth century, the cerebellum was thought to be an organ that regulates movement. Towards the end of the twentieth century, the brain functions associated with the cerebellum began to extend beyond motor control. Now, there is a consensus that the cerebellum is involved not only in motor functions but also in the most basic autonomic functions and the most complex cognitive and emotional functions, with a focus on predictions and internal models. A new functional model of the cerebellum is needed to explain all layers of brain functions by extending predictive computations in the cerebellum. On the other hand, the cerebellum and the basal ganglia were believed to be independent and complementary motor centers that lacked direct neural connections. For example, in neurophysiology classes in the 1980s, the characteristics of cerebellar ataxia were summarized as hyperkinetic and hypotonia, while the characteristics of Parkinson's disease (traditionally classified as "basal ganglia disorder") were summarized as hypokinetic and hypertonia, and therefore their functions were assumed at opposite poles, without interactions between the two main subcortical systems. The cerebellum and the basal ganglia were also assigned contrasting models regarding their learning mechanisms. Namely, the cerebellum was assumed to employ supervised learning with error signals, while the basal ganglia were assumed to employ reinforcement learning with reward prediction errors. However, recent neuroanatomical studies have demonstrated a number of novel connections between them, questioning their independence. Moreover, recent single-neuron recording and inactivation studies provided evidence that the cerebellum may also be involved in reinforcement learning. The cerebellum is neither independent of the basal ganglia nor exclusively specialized for supervised learning. We now need a new, general model to explain the contradiction between the known uniformity of the cerebellar cortex's structure and the newly added diversity of brain functions to which the cerebellum contributes. This consensus paper summarizes many of the seeds of such a new theory. The panel of experts (1) highlights the importance of the anatomical connectivity between cerebellar circuitry and basal ganglia, (2) points out that the anatomy of the cerebellum is unique and allows predictive computations in motor and extra-motor domains such as cognition, affect, social interactions and reward processes, (3) underlines the need to further elucidate the nature of interactions between cerebellar cortex and cerebellar nuclei to better understand cerebellar and psychiatric disorders and (4) suggests that common operations may underlie the motor and non-motor functions of the cerebellar circuitry. Cerebellar models remain a major topic of research to improve our understanding of the numerous cerebellar activities and to better understand the complexity of cerebellar disorders.