Reward-Based Learning and Emotional Habit Formation in the Cerebellum

Abstract

There is growing evidence of the cerebellum’s contribution to emotion processing from neuroimaging studies of healthy function and clinical studies of cerebellar patients. As demonstrated initially in the motor domain, one of the cerebellum’s functions is to construct internal models of an individual’s state and make predictions about how future behaviors will impact that state. By utilizing widespread connections with neocortex and subcortical regions such as the basal ganglia, the cerebellum can monitor and modulate precisely timed patterns of events using prediction and reward-based error feedback in a diverse range of tasks including auditory emotion prosody recognition. In coordination with a broader affective network, the cerebellum helps to select and refine emotional responses that are the most rewarded in a particular context, strengthening neural activity in relevant regions to form a representational chunk. This chunked set of affective stimuli, cognitive evaluations, and physiological responses subsequently can be enacted as a unitary response (i.e., an emotional habit) more quickly and with less attentional control than for a novel stimulus or goal-oriented action. Such emotional habits can allow for efficient, automatic, stimulus-triggered responses while maintaining the flexibility to adapt output when prediction errors signal a renewed need for cerebellar modification of cortical activity, or, conversely, may lead to behavioral or mood disorders when habitual responses persist despite negative consequences.