A formal model of interpersonal inference.

Area of Science:

• Cognitive Science
• Social Psychology
• Computational Neuroscience

Background:

• Active Bayesian inference, a decision-making framework, is extended to interpersonal exchanges.
• Social cognition presents unique challenges that require novel formal modeling.
• This study demonstrates the psychological significance of applying active inference to social interactions.

Purpose of the Study:

• To formulate a novel mathematical model for active Bayesian inference in interpersonal exchanges.
• To integrate utility theory within a Bayesian inference framework for social decision-making.
• To explore how prior beliefs, incorporating self and other preferences, influence social inference.

Main Methods:

• Literature review focusing on interpersonal representations.
• Development of a mathematical model based on Bayesian inference and utility theory.
• Simulation study to illustrate model dynamics and generate predictions.

Main Results:

• Interpersonal representations are actively inferred beliefs, akin to mentalizing, based on interaction outcomes.
• Active interpersonal inference explains social-psychological phenomena like self-serving biases.
• Bayesian updating of social outcome valuations, including self-perception, is a key component of mentalizing.

Conclusions:

• A Bayes optimal framework is proposed for modeling individual differences in mentalizing during social interactions.
• Interpersonal representations are characterized by explicit functional and affective properties.
• The active inference framework may be valuable for studying psychiatric conditions involving distorted mentalizing.