Angela Radulescu
Assistant Professor
Department of Psychiatry
Icahn School of Medicine at Mt. Sinai
Bipolar disorder (BD) is marked by striking fluctuations in motivation and goal-directed behavior, yet the cognitive mechanisms underlying these changes remain unclear. In this talk, I will present a reinforcement learning (RL) perspective on BD that links learning and motivation within a unified computational framework. This approach aims to identify cognitive markers of BD, highlighting positive overgeneralization — the excessive spread of learned value across contexts — as a potential marker present outside of acute episodes. Modeling results show that heightened sensitivity in self-efficacy belief updating can produce overgeneralized value representations and goal-directed behaviors resembling (hypo)manic symptoms. Finally, this perspective helps reconcile puzzling findings from ongoing work on value-based attention in BD, offering a cohesive account of how learning mechanisms contribute to cognitive vulnerability across the bipolar spectrum.
View a recording of this session here.