Valuation and Decision-Making in Cortical-Striatal Circuits.

Abstract

Adaptive decision-making relies on a distributed network of neural substrates that learn associations between behaviors and outcomes, to ultimately guide future behavior. These substrates are organized in a system of cortical-striatal loops that offer unique contributions to goal-directed behavior and receive prominent inputs from the midbrain dopamine system. However, the consequences of dopamine fluctuations at these targets remain largely unresolved, despite aggressive interrogation. Some experiments have highlighted dopamine’s role in learning via reward prediction errors, while others have noted the importance of dopamine in motivated behavior. Here, we explored the precise role of dopamine in shaping decision-making in cortex and striatum. First, we measure dopamine in ventral striatum during a trial-and-error task and show that it uniformly encodes a moment-by-moment estimate of value across multiple timescales. Our optogenetic manipulations demonstrate that changes in this value signal can be used to immediately enhance vigor, consistent with a motivational signal, and alter subsequent choice behavior, consistent with a learning signal. Next, I measured dopamine in multiple cortical-striatal loops to examine the uniformity of the value signal. I report that dopamine is non-uniform across circuits, but is consistent within them, implying that dopamine may offer unique contributions to the information processed in each loop. Finally, I performed single-unit recordings in the dorsal striatum, a major recipient of dopamine, to examine whether distinct its subcompartments—the patch and matrix—carry distinct value signals used in the selection of actions. I report preliminary data and summarize improvements in my electrode localization technique.