When
Noon – 1:30 p.m., Feb. 13, 2026
Image
Ari Kahn
Assistant Professor, Psychology
Cognitive Science
Cognition & Neural Systems
Predictive Models in Planning and Decision Making
Abstract: Humans are capable of an array of complex tasks. We can navigate through cities, play games like chess, and plan our life choices to achieve our goals. All of these require that we can understand the long-term outcomes of our actions. To support such planning, the human brain must construct detailed predictive maps of the world. While the role of cognitive maps has been extensively studied in spatial navigation, evidence has shown that people also routinely plan using simplified mental models such as the Successor Representation (SR) that allow them to skip multiple steps into the future. When do we rely on these simplified models of the world, and how do we learn them?
My research addresses these questions through a combination of computational modeling, behavioral, and neuroimaging studies. To understand the role of predictive models in healthy cognition, my work draws on reinforcement learning as a basis for conceptual models of planning, and network science to characterize relevant structure of our environments. Behaviorally, my work examines how trial-by-trial choices allow us to contrast different proposed planning mechanisms, while measures such as reaction time allow us to infer internal predictions and ask questions about how those predictions are constructed. I couple this with fMRI to investigate neural mechanisms of these processes. Through these approaches, I seek to characterize both how behavioral demands affect our choice of world model, and the cognitive mechanisms that drive such arbitration.
https://arizona.zoom.us/j/84883278914
Contacts
Jonathan Tullis