Computational approaches to understanding hierarchical models of personality and psychopathology

Abstract

HiTOP's quantitatively-driven organization of symptoms and traits into levels of varying breadth provides an excellent descriptive model; however, it is necessary to extend the validity and utility of the model through relations to explanatory mechanisms. The present symposium aims to advance the validation of HiTOP through examining it in relation to computational models of behavioral data, which formally define and measure processes related to perception, inference, and action selection. In the first presentation, Hallquist will provide an introduction to computational modeling and its relevance to HiTOP. Next, Williams will present on psychosis and the tendency to learn aberrant associations, from the perspective of symptom-specificity and broader HiTOP dimensions (i.e., detachment and thought disorder). Schreiber will present data that links drift-diffusion model parameters from an emotional interference task with developmental trajectories of borderline personality disorder dimensions, particularly negative affectivity, impulsivity, and interpersonal aggression. Allen will present recent work using a reinforcement learning model and fMRI to investigate differential associations between facets of antagonism and learning during a social trust game. Letkiewicz will present data from a task that examined relations between decision-making inflexibility, trait affect, and temporal difference learning model parameters. Finally, Hallquist will serve as a discussant and facilitate discussion among the speakers and audience.