Dissecting the role of cortico-basal ganglia circuit diversity in action learning from reinforcement

Project Summary/Abstract To learn novel actions through reinforcement, a fundamental mechanism of motor learning, the brain needs to causally link previously performed movements to their resulting outcomes. But even single limb movements consist of multiple aspects, which poses what is known as the credit assignment problem: ‘What was it that I just did that led to the desired outcome?’ Through repeated reselection of the movement the brain converges on which aspects are relevant, and variability is reduced in those aspects, refining the action into a skill. Current theories suggest that this process is implemented in the cortico-basal ganglia-thalamo-cortical loop. Cortical information about planned and ongoing movements is conveyed to the striatum via corticostriatal projections. If the movement leads to a desired outcome, dopamine is released in striatum, strengthening the activated corticostriatal synapses. This plasticity, in turn, is thought to allow the reselection of the movement through the basal ganglia-thalamo-cortical loop. It is unclear however how different aspects of the action are distinguished such that the relevant ones are refined. The anatomical heterogeneity of the corticostriatal sensorimotor command may provide a circuit-level mechanism of this process. In the K99 mentored phase, I will dissect this mechanism. I hypothesize that distinct corticostriatal projections convey different aspects of the motor command and what is learned is determined by which projections are reinforced. I have developed a head-fixed behavior task in which mice get rewarded for moving a joystick into a defined circular target area. In this task, a specific movement direction, a position of the rewarded endpoint, or both may be reinforced and learned by mice. I use behavioral measurements and manipulations, and neural decoding of the different corticostriatal motor commands to probe what individual animals learn. I will be mentored by Dr. Rui Costa, Dr. Daniel Wolpert, and collaborator Dr. James Murray to hone my behavior analysis skills and train in cutting-edge methods for neural decoding, such as the use of neural networks. Then I use optogenetic manipulation to directly test if anatomically distinct corticostriatal commands determine what animals learn. Besides reselection of cortical motor commands through thalamus, the basal ganglia control movement by disinhibiting brainstem motor centers, providing a parallel pathway for action refinement. One such center, the red nucleus, is directly involved in forelimb control, as previously shown by me and others, and is also innervated by the cerebellum. In the R00 phase, I will start my independent research by investigating if action refinement also depends on basal ganglia control of brainstem motor centers, particularly the red nucleus. As I transition to independence, Dr. Megan Carey will advise me in questions of cerebellum-related motor control. All mentors will promote my career development. These research and training experiences will place me as a competitive candidate to become a successful independent PI and give me the needed support to achieve milestones of getting my first R01 and publishing the first independent paper from my lab.