Bayesian inferences about the self (and others): a review

Area of Science:

• Neuroscience
• Cognitive Science
• Psychology

Background:

• The brain's function can be viewed through the lens of approximate Bayesian inference.
• Understanding self-representation is crucial for cognitive and social neuroscience.

Purpose of the Study:

• To explore how the brain represents the self using Bayesian inference.
• To propose that self-representations serve to predict and optimize social interaction outcomes.

Main Methods:

• Conceptualizing prediction and optimization as active inference.
• Defining outcome utility as prior beliefs about final states.
• Analyzing interpersonal representations and their role in minimizing surprise.

Main Results:

• Interpersonal representations enhance the predictability of social interactions.
• Affective valence in interpersonal inference contributes to evaluative self-perception.
• Distorted self-representations are linked to psychiatric disorders like depression and paranoia.

Conclusions:

• Active inference provides a framework for understanding self-representation and social interaction.
• This approach may operationalize the study of interpersonal representations in pathological states.