The project entitled "Cortical-subcortical information transfer underlying skill learning" (CSIT) was aimed at understanding the brain's stability-plasticity dilemma. The stability-plasticity dilemma is a critical constraint in brain networks underlying learning and memory. The dilemma is how the brain can acquire new information (plasticity) without overriding older knowledge (stability). In this project, we studied this dilemma by examining how brain activity changes as a new behavior is learned: are there specific brain areas involved only in learning, while brain areas store learned behaviors? Or do networks of brain areas interact to mediate both processes? We specifically studied two brain areas associated with learning and storage of behaviors, the motor cortex and the subcortical basal ganglia. Neural signals were recorded across both regions in the rat brain as rats learned a specific motor task that allowed for the precise behavioral quantification of learning. We then measured the behaviorally relevant information carried by neural signals in each brain area, as well as the information flow between areas. We found that while both motor cortex and the basal ganglia contained behaviorally relevant information during the initial learning of the task and the production of the stored, learned behavior, the flow of information changed dramatically. During learning, information flow was "top-down", originating from motor cortex and flowing subcortically, however, when the stored behavior was performed after learning, information flow was "bottom-up", originating from the basal ganglia and flowing to cortex. This revealed to us that there are not separate regions in the brain that underlie the learning and storage of a behavior. Rather, distributed networks of cortical and subcortical brain regions mediate both processes, with changes in the dynamics of communication within the networks allowing them to switch functions. To achieve the results of this project, we developed a new computational tool that allowed us to isolate the dynamics of behaviorally relevant information flow in the brain. Our tool will help advance our understanding of brain function, both within the context of the brain's stability-plasticity dilemma, as well as other complex functions of the brain. The results of this project shed light on how distributed networks in the brain, spanning both cortical and subcortical areas, work together to generate behavior. This understanding is fundamental for understanding of how brain areas communicate, as well as our understanding of how brain injuries impact both the ability to learn new behaviors and perform stored behaviors.
