BACKGROUND

Behavioral flexibility -the ability to dynamically readjust our behavior in response to reward contingency changes- is often investigated using probabilistic reversal learning tasks (PRLT). Poor PRLT performance has been proposed as a proxy for compulsivity, and theorized to be related to perseverative gambling. Previous attempts to measure inflexibility with the PRLT in patients with gambling disorder have, however, used a variety of indices that may conflate inflexibility with more general aspects of performance in the task.

METHODS

Trial-by-trial PRLT acquisition and reacquisition curves in 84 treatment-seeking patients with gambling disorder and 64 controls (non-gamblers and non-problem recreational gamblers) were analyzed to distinguish between (a) variability in acquisition learning, and (b) reacquisition learning in reversed contingency phases. Complementarily, stay/switch responses throughout the task were analyzed to identify (c) premature switching, and (d) sensitivity to accumulated negative feedback.

RESULTS AND INTERPRETATION

Even after controlling for differences in acquisition learning, patients were slower to readjust their behavior in reversed contingency phases, and were more prone to maintain their decisions despite accumulated negative feedback. Inflexibility in patients with gambling disorder is thus a robust phenomenon that could predate gambling escalation, or result from massive exposure to gambling activities.