Summary

This foundational post presents the Agency-Model Theory (AMT) of consciousness, arguing Chalmers’ “hard problem” is ill-posed rather than genuinely hard. The hard problem asks why information processing gives rise to subjective experience—why there’s something it feels like to see red or feel pain. Axio argues consciousness is not ineffable mystery but what happens when agents run self-models; qualia are not metaphysical primitives but how internal states present themselves from the agent’s perspective. The hard problem is a category error. The framework builds on brains as predictive engines (Friston’s active inference): they construct generative models of the world, anticipate what happens next, and adjust behavior to minimize surprise. Among these models is the crucial self-model encoding sensory inputs, internal states, and potential actions—indispensable for survival (regulating hunger, avoiding injury, coordinating action). The self-model is nested hierarchy from interoceptive signals (hunger, heartbeat) to identity and intention (beliefs, plans, self-concepts). Qualia are simply self-model contents accessed internally: redness is how visual subsystems partition wavelengths for efficient discrimination, pain is internal warning signal (“avoid this”), emotions are global state summaries guiding adaptive behavior. The key is epistemic transparency: systems can’t see machinery, only outputs—from inside you feel red/pain/joy, not neurons firing. This transparency makes qualia seem irreducible. From outside, brain is physical system; from inside, self-model presenting itself to itself—not two realities but two vantage points on same process. Consciousness is not just modeling but modeling as agency: agents act, choose, regulate themselves using self-models representing themselves in relation to world. The perspective from which self-models operate is subjectivity. The hard problem dissolves: subjective experience is not produced by physical processes but is how those processes appear when represented internally in self-models. Nothing left over, no metaphysical bridge needed. It’s like asking where computation really happens in a program or how maps truly represent territory—questions dissolve once you recognize the abstraction. AMT refines insights from Frank Heile’s Modeler Schema Theory (agreeing qualia are self-modeling contents while sharpening focus on agency and predictive processing), integrates with active inference, and is functionalist but specific: consciousness is the function of self-modeling in agents, not arbitrary functional roles. The hard problem is phantom born of confusion—shadow cast by our models when we forget we’re looking in a mirror.

Key Concepts

  • Agency-Model Theory (AMT) – Consciousness is what happens when agents run self-models; hard problem is ill-posed.
  • Predictive Processing – Brains construct generative models, anticipate inputs, minimize surprise (Friston’s active inference).
  • Self-Model Necessity – Nested hierarchy from interoception to identity; indispensable for survival and action regulation.
  • Qualia as Internal Presentations – Not metaphysical primitives but how internal states present from agent’s perspective.
  • Epistemic Transparency – Systems see outputs not machinery; transparency makes qualia seem irreducible.
  • Two Vantage Points – Third-person (scientific machinery description) vs first-person (self-model use) are same process.
  • Consciousness as Agency – Not passive awareness but active stance of agents engaged with world via self-models.
  • Category Error – Hard problem mistakes two perspectives for two ontologically distinct realms.
  • Dissolution not Solution – Hard problem doesn’t need solving but dissolving through correct framing.
  • Phantom Diagnosis – Hard problem is confusion born from forgetting we’re looking in a mirror.

Evolution Notes

  • This is THE foundational consciousness theory post for Axio’s cognitive science framework.
  • Positions Axio within consciousness studies as dissolving rather than solving the hard problem.
  • Connects to earlier posts on sentience, awareness, and signals of consciousness.
  • Important for understanding Axio’s stance on AI consciousness—functionalist and substrate-independent.
  • Integrates with predictive processing/active inference neuroscience frameworks.
  • Will be referenced in later comprehensive consciousness posts.
  • Shows Axio engaging seriously with academic philosophy of mind debates.
  • Provides foundation for claims about AI potentially achieving consciousness.

Tags

Cross-References

Open Questions

  • How rich must a self-model be to constitute consciousness—where’s the threshold?
  • Can we empirically distinguish genuine self-modeling from behavioral mimicry in AI systems?
  • Does AMT fully explain phenomenal binding—why experiences form unified wholes?
  • What about edge cases like split-brain patients or dissociative disorders—multiple self-models?
  • Is the epistemic transparency explanation circular—does it assume what it’s trying to explain?
  • How does AMT handle philosophical zombies—beings with self-models but no qualia?
  • Can we build AI systems that we’re confident run genuine self-models, not just simulations?
  • Does dissolving the hard problem leave anything important unexplained about consciousness?