Deformed probability estimation in the goal-directed reinforcement learning model explains anxious-depression dimension of psychiatric disorders.

Psychiatric disorders are complex, multidimensional pathologies rooted in various cognitive processes. Computational psychiatry aims to reveal distortions in these processes through behavior modeling, providing a deeper understanding of psychiatric disorders. Previous studies that used Daw's two-stage task linked imbalances between habitual/model-free and goal-directed/model-based behaviors to disorders featuring compulsive behaviors and intrusive thoughts. The model-based part depends on estimating environmental probabilities. To improve this, we incorporated a well-known distortion in subjective probability estimation into the model, which enhanced its fit. More importantly, the fitted deformation explains some of the variance in the anxious-depression dimension of psychiatric symptoms. The deformation parameter aligns with the description-experience gap discussed in decision-making research. Our results suggest that subjective distortions may be a key cognitive process underlying anxiety, apathy, and depression. This study also shows that including cognitive biases in models can reveal hidden behavioral aspects linked to disorders. Our approach improves the accuracy of computational psychiatry and provides deeper insights into the cognitive processes behind psychiatric symptoms, paving the way for more personalized and effective treatments.