Identifying the neural networks that support categorization using brain-informed drift diffusion modelling

Poster Session D - Monday, April 15, 2024, 8:00 – 10:00 am EDT, Sheraton Hall ABC

Victoria Liu¹ (tliu2@research.baycrest.org), Michael Mack¹; ¹University of Toronto

Categorizing elements of our experience is a cognitive shortcut to transform the overload of sensory information into something conceptually meaningful. Although many categories are defined by regularities (e.g., most birds fly), not all items conform to these simple rules (e.g., bats fly but are mammals not birds). Recent studies have focused on the roles of basal ganglia and hippocampus in category learning, but there remain key open questions about broader cortical networks that support category decision making. Here, we combined functional neuroimaging and drift diffusion modelling (DDM) to characterize the neural dynamics underlying category decisions. We leveraged a rule-plus-exception visual category learning task, in which most items are correctly categorized by a simple feature rule, but rare category exceptions violate the rule and are visually more similar to the other category. Given the opposing demands for learning rule-following items (i.e., generalizing across regularities) and exceptions (i.e., distinct encoding), we predicted that category evidence for item types may be represented by distinct neural networks. We linked trial-by-trial neural activation from atlas-defined ROIs across the whole brain to drift rate in the DDM and evaluated the degree that neural signal predicted category decision making. Results showed that default mode network regions tracked category regularities such that higher activation led to faster and more accurate responses to category prototypes. In contrast, regions within control and salience networks tracked decisions for category exceptions. These findings suggest that distinct cortical networks may represent key latent decision variables during category learning.

Topic Area: THINKING: Decision making