Date of Award

2026-05-01

Degree Name

Doctor of Philosophy

Department

Computational Science

Advisor(s)

Alexander Friedman

Abstract

Decision-making requires a complex integration of costs and rewards to determine a balance between pursuit and avoidance behaviors, a process mediated by cortico-striatal circuits. This dissertation presents two interrelated computational contributions to understanding how the brain encodes and controls decisions. In the first part, I propose a biologically realistic model of neural circuit activity during cost-benefit decision-making and its regulation by chronic stress. The model makes experimentally testable predictions about shifts in circuit balance that give rise to changes in decision-making observed following stress exposure in rodents and humans, advancing understanding of how the brain encodes choices and the mechanisms underlying risky behavior. In the second part, I address a foundational question about the neuromodulatory signal that drives this circuit: what does dopamine actually compute? I show that apparent reward prediction error (RPE) correlates in dopamine activity arise mathematically as a special case of a more general computation: the information gain of the policy of actions (policy-IG). Policy-IG quantifies how much newly arriving information refines the choice of action, and it accounts for dopamine responses that RPE cannot explain, including responses to aversive events, novelty, movement, and moment-to-moment decision control. Simulating impaired policy-IG replicates features of basal ganglia disorders, suggesting it as a target for future therapies. Together, these two works propose a unified computational framework for how the striosome-dopamine-striatum circuit constructs, regulates, and updates the decision-making process.

Language

en

Provenance

Received from ProQuest

File Size

609 p.

File Format

application/pdf

Rights Holder

Dirk Beck

Available for download on Friday, June 16, 2028

Share

COinS