The Game of Self: Identity and Experience as Active Inference.

Area of Science:

• Cognitive Neuroscience
• Computational Psychiatry
• Philosophy of Mind

Background:

• Active inference models the nervous system as a predictive engine minimizing sensory uncertainty.
• Perception and behavior are driven by the imperative to reduce uncertainty.

Purpose of the Study:

• Extend active inference to social agents' self-inference.
• Conceptualize self-knowledge as "The Game of Self"—a Bayesian updating process.
• Model the dynamic interplay between semantic and episodic self-representations.

Main Methods:

• Bayesian inference and updating of self-representations.
• Modeling a bidirectional predictive loop between identity and experience.
• Utilizing active inference principles to explain self-related uncertainty reduction.

Main Results:

• Self-knowledge emerges from continuous Bayesian updating of self-representations.
• A bidirectional loop exists where semantic identity shapes episodic experience, and vice versa.
• Emergent self-representations are those with the highest posterior probability given evidence.

Conclusions:

• The self is not fixed but adaptively shaped by active inference.
• "The Game of Self" offers a computational framework for understanding selfhood.
• This model provides novel insights into how the brain navigates self-related uncertainty.