Probabilistic reinforcement learning abnormalities and their correlates in adolescent bipolar disorders

Area of Science:

• Neuroscience
• Psychiatry
• Developmental Psychology

Background:

• Bipolar disorder (BD) theoretical models suggest reward system dysfunction.
• Specific neurobehavioral deficits in the reward system in adolescent BD require clarification for targeted interventions.
• Research on probabilistic-reinforcement learning in adolescent BD is limited.

Purpose of the Study:

• To investigate probabilistic-reinforcement learning using a probabilistic selection task (PST) in adolescents with and without BD.
• To examine correlates of probabilistic-reinforcement learning, including reward/threat sensitivities and cognitive tasks.
• To determine if learning deficits incrementally predict BD diagnosis.

Main Methods:

• A probabilistic selection task (PST) was administered to 104 adolescents (with and without BD).
• Self-reported reward/threat sensitivities and cognitive tasks were used to assess correlates.
• Statistical analyses examined group differences, strategy efficiency, and predictive validity.

Main Results:

• Adolescents with BD exhibited lower win-stay rates in the PST acquisition phase compared to controls.
• Higher win-stay rates were associated with more efficient learning (fewer trials, better performance).
• Lower win-stay rates incrementally predicted BD diagnosis, independent of symptoms, sensitivities, learning, and IQ.

Conclusions:

• Adolescent BD involves distinct reward system dysfunctions, including impaired probabilistic-reinforcement learning.
• Win-stay behavior in learning tasks is a potential biomarker for adolescent BD.
• Further research is needed to understand the multifaceted reward system deficits in adolescent BD.