PhD Thesis Defense
Title: Value-based decision making and learning as algorithms computed by the nervous system
"How do we do what we do? Casting light on this essential question, the blossoming perspective of computational cognitive neuroscience gives rise to the present exposition of the nervous system and its phenomena of value-based decision making and learning. As justified herein by not only theory but also simulation against empirical data, human decision making and learning are framed mathematically in the explicit terms of two fundamental classes of algorithms--namely, sequential sampling and reinforcement learning. These counterparts are complementary in their coverage of the dynamics of unified neural, mental, and behavioral processes at different temporal scales. Novel variants of models based on such algorithms are introduced here to account for findings from experiments including measurements of both behavior and the brain in human participants. The thesis is divided into three parts:
1) Value-based decision making via sequential sampling with hierarchical competition and attentional modulation
2) Learning where to look for high value improves decision making asymmetrically
3) Prediction errors in mesostriatal circuits of the human brain mediate learning about the values of both states and actions: evidence from high-resolution fMRI"