Chapter 10 - Computational neuroimaging: monitoring reward learning with blood flow

Abstract

This chapter discusses the strengths and limitations of functional MRI (fMRI) in the study of reward processing and reviews recent advances in using fMRI to map, and in some instances extend, reinforcement learning models of human brain function. The reward prediction error theory of dopamine function is one of the great recent advances in neuroscience. It has spurred research on reinforcement learning at all levels of investigation that has now localized many components of the associated computational algorithms to specific neural processes. With these advances, interest has expanded to include the neural basis of human reward learning and decision-making. Functional MRI (fMRI) has become the method of choice for these experiments since it offers the desired combination of spatial and temporal resolution. Furthermore, the chapter discusses the strengths and limitations of fMRI in the study of reward processing and review recent advances in using fMRI to map, and, in some instances extend, reinforcement learning models of human brain function.