Beyond Consciousness: Evidential Limits and Phase Transitions in the Science of Experience

This paper develops a unified framework addressing the evidential foundations and dynamical structure of consciousness science through three co-equal pillars.

The epistemological pillar presents a formal Bayesian argument (Theorem 2.3) demonstrating that first-person subjective reports, neural correlates, and all other functionally determined observations carry a likelihood ratio of exactly one for discriminating between phenomenal consciousness and functional duplicates lacking experience. This evidential degeneracy is not an empirical limitation but a logical consequence of how the zombie hypothesis defines functional equivalence, and it applies uniformly to every empirical methodology currently employed in the field — contrastive analysis, neural correlates of consciousness, perturbational complexity, and integrated information measures. The result is situated within the broader epistemic gap literature (Jackson, 1982; Levine, 1983; Chalmers, 1996; Stoljar, 2006) and formalises the forced choice between accepting zombie coherence (and therefore evidential paralysis) and rejecting it (and therefore embracing functionalism).

The methodological pillar, extending recent work on operational terminology in Integrated Information Theory (Krieger, 2025), articulates a three-level framework for scientific claims about consciousness: formal-mathematical (Level 1), empirical-functional (Level 2), and interpretive-phenomenological (Level 3). The evidential degeneracy applies specifically to the transition from Level 2 to Level 3. The paper engages with IIT 4.0's ontological commitments (Albantakis et al., 2023; Tononi & Koch, 2015), arguing that even constitutive identity claims must be evaluated against alternatives that no functional observation can adjudicate.

The constructive pillar develops a dynamical systems framework proving that integration measures — including IIT's Φ as a special case — undergo first-order phase transitions under conditions of bistable connectivity. Using Landau potential theory and bifurcation analysis, with two independent proof strategies (constructive via normal forms and topological via equilibrium manifold structure), the paper establishes discontinuity, hysteresis, and critical slowing down as generic features of the transition. Finite-size scaling analysis shows that the transition is effectively discontinuous for brain-scale systems (N ~ 10¹⁰). Seven schematic illustrations visualise the predicted phenomena, with all generating code provided. The framework generates specific, empirically testable predictions distinguishing first-order from second-order transitions.

A dedicated section examines the implications of advanced AI as the realised approximation of the philosophical zombie, arguing that the capacity of artificial systems to achieve cognitive parity through pure information processing renders the traditional concept of phenomenal consciousness functionally redundant for a predictive science of mind.

The synthesis demonstrates that the epistemological constraint motivates the methodological discipline, which in turn enables the constructive programme — a rigorous dynamical-systems investigation of functional transitions that is explicit about its evidential scope.

The paper includes a complete first-round peer review with detailed author responses, documenting the revision process.

Keywords: consciousness, Bayesian epistemology, evidential degeneracy, phase transitions, dynamical systems, integrated information theory, hard problem, phenomenal experience, bifurcation theory, Landau potential, philosophical zombies, artificial intelligence, operational definitions

Author: Boris Kriger¹² — ORCID: 0009-0001-0034-2903 ¹ Information Physics Institute, Gosport, Hampshire, United Kingdom ² Institute of Integrative and Interdisciplinary Research, Toronto, Canada