An Active Inference Model of the Optimism Bias.

Area of Science:

• Cognitive Science
• Computational Psychiatry
• Neuroscience

Background:

• Optimism bias involves overestimating positive outcomes and underestimating negative ones, linked to better quality of life.
• This bias is considered adaptive and a potential target for mental health interventions, particularly for conditions like major depressive disorder.

Purpose of the Study:

• To develop a domain-general formal model of optimism bias using the active inference framework.
• To provide a computational basis for understanding the emergence, expression, and behavioral impact of optimism bias.
• To lay groundwork for future research on optimism as a mental health intervention.

Main Methods:

• Proposed a model of optimism bias as high precision likelihood biased towards positive outcomes within the active inference framework.
• Simulated the loss of optimism during development due to negative event exposure.
• Grounded the model in empirical literature using belief updating and two-armed bandit tasks.

Main Results:

• Demonstrated how developmental differences in optimism are expressed in belief updating tasks.
• Showcased the effect of optimism on action selection in a modified two-armed bandit task.
• Provided simulations illustrating the model's capacity to explain optimism bias.

Conclusions:

• The developed computational model offers a basis for understanding optimism bias.
• The model explains how optimism bias emerges, is expressed in tasks, and influences decision-making.
• This work supports future research into optimism as a therapeutic intervention for mental health conditions.