Bias toward escape responding during reinforcement learning among those with suicidal ideation

Self-injury and suicide can be characterized by reward system dysfunction and self-reports of active efforts to escape unpleasant emotional states. Therefore, individuals with histories of suicidal ideation (SI) should exhibit a preference for active escape from unpleasant states, which exceed the effects of impulsive behavior under distress and lack of premeditation. Participants made active (Go) or passive (No-Go) choices in response to stimuli to escape or avoid an unpleasant state in a behavioral task. A drift-diffusion reinforcement learning model was used to estimate latent biases for active escape and avoidance in people with and without SI history. Bayesian logistic regression was used to examine the relationship between escape and avoid bias with SI. Escape bias predicted SI history, whereas avoidance bias did not. Escape bias remained a significant predictor of SI when controlling for negative urgency and lack of premeditation. Those with histories of SI demonstrate a decision-making bias favoring escape from aversive states. This bias remains significant after adjusting for facets of impulsivity linked to hasty decisions under distress and lack of premeditation. A heightened escape response may help clinicians to identify SI risk and develop targeted treatments to attenuate the escape bias.