Goal-Directed Learning Deficits in Patients with OCD: A Bayesian Analysis

Dual system learning theories posit an overreliance characterizes obsessive–compulsive disorder on habitual decision-making at the expense of goal-directed learning. However, most studies on this topic used frequentist statistics and did not evaluate the discriminant validity of goal-directed learning indices.

We recruited 55 OCD patients, used Bayesian statistics to examine goal-directed learning in OCD, and tested whether goal-directed learning indices showed discriminant validity. Patients completed self-reports, a two-stage reinforcement learning task, executive functioning (EF), and related tests (Rey Complex Figure, Wisconsin Card Sorting, Stroop tests).

Analyses showed that goal-directed learning deficits were related to compulsions and self-reported OCD severity, but not obsessions, OCD-related beliefs, trait anxiety, anxiety sensitivity, or depression. Moreover, increased compulsivity and self-reported OCD (vs. all other dimensions examined) were linked to faster responses. Evidence for discriminant validity was found. Goal-directed learning coefficients were unrelated to set-shifting, inhibitory control, and visuospatial memory. Compulsivity was unrelated to memory and set-shifting but associated with poorer inhibition.

Findings highlight the importance of conceptualizing psychopathology dimensionally, such as the relevance of imbalanced habitual vs. goal-directed behavior in OCD. Results also underscore the importance of examining unique associations of goal-directed behaviors in OCD with clinical/syndromal measures (compulsions vs. obsessions).