Research Projects

The neural basis of trial-and-error learning has been studied primarily using simple choice tasks, but little is known about how the brain learns in complex, real-world environments. The brain must learn efficient strategies to explore available options and determine the optimal choice. Theoretical studies suggest that agents can take advantage of statistical structure in a task's reward landscape to explore more efficiently. However, the brain's ability to learn and the neural circuits supporting this ability are unknown. The objective is to uncover exploration strategies used by the brain to explore large action spaces and solve complex decision-making tasks. The study will also investigate the role of basal ganglia in directing efficient exploration strategies, specifically determining the relative contribution of basal ganglia pathways through the dorsolateral (DLS) and dorsomedial (DMS) striatum to efficient exploration of large action spaces. The working hypothesis is that the brain learns efficiently by taking advantage of statistical structure in a task or environment, allowing it to generalize reward values of actions it has not yet taken from those it has already sampled. Experiments will be conducted using a new automated foraging paradigm for rats, analyzing their performance and trial-by-trial choices using reinforcement learning theory. The experiments will causally reveal the neural circuitry responsible for learning and exploiting statistical structure in a complex task for improved performance.