Poster D106, Monday, March 26, 8:00-10:00 am, Exhibit Hall C
Drift-Diffusion Modeling of Reward Learning in Depression
Victoria Lawlor1, Christian Webb1, Madhukar Trivedi2, Maurizio Fava3, Patrick McGrath4, Myrna Weissman4, Ramin Parsey5, Maria Oquendo6, Patricia Deldin7, Gerard Bruder4, Diego Pizzagalli1, Daniel Dillon1; 1McLean Hospital, 2University of Texas Southwestern Medical Center, 3Massachusetts General Hospital, 4Columbia University Medical Center, 5Stony Brook School of Medicine, 6University of Pennsylvania Perelman School of Medicine, 7University of Michigan
Major Depressive Disorder (MDD) has been associated with disrupted reward learning, but the underlying neurocognitive mechanisms are poorly understood. For example, relative to healthy controls, adults with MDD typically show poorer performance in the probabilistic reward task (PRT), but the reason for this group difference remains unclear. Therefore, we applied the Hierarchical Drift Diffusion Model (HDDM) to three PRT datasets. The HDDM decomposes behavioral data into component cognitive processes, and we sought to identify which processes are affected by MDD. PRT data from 104 healthy controls and 302 depressed participants were analyzed. The HDDM was used to extract three decision-making parameters: drift rate, decision threshold, and prepotent bias. The HDDM revealed slower drift rates and higher decision thresholds in depressed versus healthy adults. In all three samples, HDDM parameters mapped onto standard PRT outcome variables: drift rate and threshold explained discriminability, while prepotent bias explained response bias. Discriminability, the ability to differentiate between task stimuli, predicted the number of rewards received better than response bias.These findings indicate that the PRT is readily modeled as a perceptual decision-making task, and they highlight key roles for discriminability and drift rate (in addition to response bias) in task performance. Conceptualizing the PRT in this way may forge a link between studies of reward learning in depression and extensive work on evidence accumulation in non-human primates. Most importantly, these results provide insight into aberrant decision-making in depression, by linking MDD to slow evidence accumulation and conservative threshold settings.
Topic Area: THINKING: Decision making