A beautiful loop: An active inference theory of consciousness

Area of Science:

• Cognitive Science
• Neuroscience
• Artificial Intelligence

Background:

• Consciousness remains a significant challenge in understanding the mind.
• Active inference offers a potential framework for modeling cognitive processes.
• Existing models often struggle to capture the dynamic and integrated nature of conscious experience.

Purpose of the Study:

• To propose and validate a theoretical model where active inference can explain consciousness.
• To outline the necessary conditions for active inference to exhibit conscious properties.
• To introduce the 'Beautiful Loop Theory' as a formalization of conscious modeling.

Main Methods:

• Defining three conditions for active inference to model consciousness: world model simulation, inferential competition (Bayesian binding), and epistemic depth.
• Proposing a hyper-model for precision-control to manage inference layers.
• Describing the recursive nature of belief sharing in hierarchical systems.

Main Results:

• Active inference can simulate a world model (epistemic field) for knowledge and action.
• Bayesian binding, through inferential competition, selects inferences that reduce uncertainty, akin to conscious selection.
• Epistemic depth enables a world model to possess knowledge of its own existence (field-evidencing).

Conclusions:

• The proposed 'Beautiful Loop Theory' integrates active inference, Bayesian principles, and hierarchical processing to model consciousness.
• This framework provides insights into general intelligence, altered states of consciousness, and meditation.
• Active inference offers a viable computational approach to understanding the mechanisms of consciousness.