Computational models of psychological processes, such as reinforcement learning, have the potential to reshape how we conceptualize and treat psychiatric dysfunction. As a new field, however, what methods are appropriate for within- and between- person studies of dysfunction and treatment response in computational psychiatry are unclear. My talk will first cover empirical findings on reinforcement learning dysfunctions and response to cognitive-behavioral therapy (CBT) in depression. In this work, we found valence- and symptom-specific learning disruptions in depression. These disruptions were remediated with successful CBT treatment, suggesting the therapeutic potential of targeting learning dysfunctions in depression and related disorders. Second, I???ll talk about some of the methodological issues we encountered in these analyses and discuss approaches to maximize reliability to ensure clinical translation of computational psychiatry work, with a particular focus on advantages & appropriate use of hierarchical models.

Vanessa Brown
Assistant Professor
Department of Psychiatry
University of Pittsburgh
Pittsburgh PA USA


Code for simulations in the talk is here:

Vanessa Brown – Reinforcement learning dysfunction and treatment in depression: Individual differences, treatment effects, and the reliability “paradox” in computational psychiatry