A model of mood as integrated advantage

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Affective Science

Background:

• Mood influences cognition and behavior.
• Mood fluctuates slowly based on environmental feedback.

Purpose of the Study:

• Introduce a new computational theory for mood valence: the Integrated Advantage model.
• Explain the bidirectional interaction between mood and the environment.

Main Methods:

• Conceptualize mood valence as a leaky integral of action appraisals.
• Utilize reinforcement learning and stochastic optimization theory.
• Extend previous models based on reward prediction errors.

Main Results:

• The Integrated Advantage model accelerates learning in complex environments.
• Mood acts as a momentum-based update for behavioral policy.
• Explains various contextual mood effects like expectation, surprise, and counterfactuals.

Conclusions:

• The Integrated Advantage model offers a principled explanation for mood's adaptive functions.
• Mood facilitates rapid learning of optimal actions in stochastic environments.
• Provides a parsimonious account for diverse affective science findings.