Consciousness as Constraint: A Critical Engagement with Integrated Information Theory

Integrated Information Theory (IIT) treats consciousness as a quantifiable property (Φ) instantiated by physical systems with specific causal architecture, but its bearer-commitments generate stable structural tensions: the exclusion problem, the grain problem, the small-Φ problem, and temporal instability. The Triaxial Existential Field (TEF) retypes consciousness as integrated reflexive availability under constraint, articulated through three irreducible roles: Coherence, Reflexivity, and Participation. This retyping yields three differential payoffs: a clinically testable typology of consciousness disorders classifying by axis-failure rather than scalar degree; cross-domain convergence with constraint-based treatments of time and quantum measurement; and implementable architectural criteria for artificial consciousness where IIT's Φ-maximisation remains computationally intractable. Explicit falsification conditions are stated.

Study at a glance

Abstract

Abstract This paper applies the bearer/constraint audit methodology developed in “From Substance to Constraint” to Integrated Information Theory (IIT); the most rigorous contemporary attempt to treat consciousness as a quantifiable property (Φ) instantiated by physical systems with specific causal architecture. IIT’s bearer-commitments generate stable structural tensions: the exclusion problem, the grain problem, the small-Φ problem, and temporal instability. These persist across theory versions because they are downstream consequences of treating consciousness as a bearer-property rather than a constraint-role. The Triaxial Existential Field (TEF) retypes consciousness as integrated reflexive availability under constraint, articulated through three irreducible roles; Coherence, Reflexivity, and Participation. This retyping yields three differential payoffs: (1) a clinically testable typology of consciousness disorders classifying by axis-failure rather than scalar degree; (2) cross-domain convergence with constraint-based treatments of time and quantum measurement; and (3) implementable architectural criteria for artificial consciousness where IIT’s Φ-maximisation remains computationally intractable. Explicit falsification conditions are stated. TEF is positioned as a methodological bet with concrete stakes in clinical classification and AI architecture.