Technical Program

Paper Detail

Paper: PS-1B.26
Session: Poster Session 1B
Location: Symphony/Overture
Session Time: Thursday, September 6, 18:45 - 20:45
Presentation Time:Thursday, September 6, 18:45 - 20:45
Presentation: Poster
Paper Title: An Algorithm for Clustering Decision-Making Phenotypes from Behavioural Data
Manuscript:  Click here to view manuscript
Authors: Abraham Nunes, Alexander Rudiuk, Thomas Trappenberg, Dalhousie University, Canada
Abstract: Computational modeling of behavioural data is increasingly used for psychiatric applications. To this end, it is of interest to devise approaches for clustering behavioural phenotypes, where those phenotypes are reinforcement learning models fit to individual subjects. The present study highlights that clustering in the space of individual parameter estimates may not reveal the presence of decision-making phenotype groups. We also introduce a similarity measure that improves cluster detection using synthetic data generated on a task that measures goal-directed and habitual control. Our results motivate further investigation regarding unsupervised cluster detection in computational cognitive modeling of human behavioural data.