Model-Based fMRI Analysis: Beyond Reinforcement Learning


Russell Poldrack

(University of Texas, Austin)

Play Video

Play Audio File

Date: 08/15/2012


The use of behavioral models to derive predictions for fMRI analysis has been particularly successful in the context of reinforcement learning models, but has been used relatively infrequently with other kinds of models. I will discuss three approaches that have used cognitive models to drive fMRI analyses. The first uses drift diffusion models of response times to understand the role of specific brain regions in simple decision processes. The second uses diffusion models to understand individual differences in executive function on the stop signal task. The third uses a model of dynamic decision making to better understand risk taking on the Balloon Analog Risk Task. Together, these studies illustrate new approaches to the integration of mathematical cognitive models into fMRI analysis.

Created: Wednesday, August 15th, 2012