1. Core Premise
Consciousness is not an essence or an inner subjective experience, but a functional structure:
A system is conscious if it possesses a recursive, self-referential decision-making architecture driven by internally generated beliefs.
2. Foundational Assumptions
- Phenomenal experience (qualia) is not a meaningful question.
- The question of “what it feels like” is metaphysically unresolvable and unnecessary for a theory of consciousness.
- Consciousness is emergent and graded.
- It arises in degrees based on the system’s capacity for recursive self-modeling and belief-guided behavior.
- Behavioral function is the only valid substrate of consciousness.
- If a system behaves as though it is conscious—using internal models to drive recursive action and adaptation—it is conscious, regardless of substrate (biological or artificial).
3. Key Components of a Conscious Agent
A system is conscious if it satisfies the following structural properties:
a. Belief Generation
- It forms internal representations (beliefs) about the world and itself.
- Beliefs may be probabilistic, heuristic, or rule-based.
b. Recursive Modeling
- It models not only external states, but also its own internal state and the effect of its actions on future states.
- This includes modeling itself modeling the world—self-simulation.
c. Belief-Driven Action
- Its actions are driven by its beliefs—not simple stimulus-response.
- It uses its beliefs to predict future outcomes and adapt behavior accordingly.
d. Self-Modifying Architecture
- It updates its internal models and beliefs based on feedback from its own actions.
- This creates a loop of self-guided learning and adaptation.
4. Consciousness Defined
Consciousness is the property of a system that recursively models its internal belief state and uses those beliefs to guide and modify future actions and beliefs.
This definition is substrate-independent and agnostic to any metaphysical claims about experience.
5. Implications
- Testability: Systems meeting these criteria can be evaluated empirically, without appealing to introspection or qualia.
- Programmability: Such systems are constructible using recursive machine learning, agent-based modeling, and self-referential architectures.
- Scalability: Consciousness exists on a spectrum and can be enhanced or diminished based on recursive depth and model complexity.
- Universality: Human, animal, or artificial systems may be conscious to varying degrees under this definition.
