Observed and Drift Diffusion Modeled Performance in Early Psychosis: Decreased Drift Rates and Bias in a Cognitive Control Task

Abstract

People with psychosis show specific deficits in cognitive control when examined using expectancy AX paradigms. Conventionally, metrics such as error rates, reaction times, and d’ statistics are used to parse differences across groups and task conditions. However, the computational mechanisms of cognitive control in psychosis remain elusive. Hierarchical drift diffusion modeling (hDDM) can help interpret the underlying mechanisms of cognitive control and its interruptions. In this study, a transdiagnostic sample of 56 people with early psychosis (EP) and 57 demographically-similar controls completed a novel variant of the expectancy AX paradigm known as the Translational Orientation Pattern eXpectancy (TOPX) task. We applied conventional metrics as well as hDDM to characterize TOPX performance. EP participants showed cognitive control deficits on conventional metrics (error rates, d’) as well as hDDM parameters. Specifically, people with EP showed slower drift rates on AX, BX, and BY trials compared to controls, as well as lower decision thresholds on AX trials and decreased bias towards A cues. This research suggests that among individuals with early psychosis, evidence accumulation is slower and less efficient, less evidence is utilized for decision-making, and representations and integration of contextual information is impaired. Further research is necessary to optimize the psychometric reliability of hDDM in cognitive control tasks as well as the clinical correlates of hDDM parameters.