Homeostatic Self-World Boundary as an Admissible Task: A cECT Benchmark Toward Minimal Consciousness

This revised version of Homeostatic Self-World Boundary as an Admissible Task presents a cECT-based computational benchmark for the admission of a homeostatic self-world-boundary task as a precursor to minimal consciousness research. The paper does not claim to generate consciousness, qualia, or a complete phenomenal self-model. Instead, it tests whether a cECT-like agent can retain and exploit an operational distinction between world-driven and internally driven contributions to control under homeostatic pressure. The benchmark frames the self-world-boundary task as an admissible task: a transformation that can be retained, repeated, and causally exploited under relevant constraints. The reported validation compares a world-only baseline, a decorative-boundary control, hand-tuned and oracle boundary controls, and an emergent SWB agent. The central result is that the emergent agent admits and retains a moderate self-world-boundary weight, w_SWB, in the recruitable band, suppresses the boundary in trivial regimes, and blocks admission in collapsed regimes where conjunctive sustainability cannot be maintained. Version v1.1 strengthens the original release by adding three technical refinements. First, it introduces the notion of algorithmic friction, making rejection visible through the temporal decay of the eligibility trace Λ_SWB when an internal boundary signal fails to causally improve the Horizon Score S_H. Second, it revises operational valence, V_t, from a monolithic distance measure into a multidimensional critical-minimum structure. In this revised interpretation, V_t is constrained by the weakest indispensable viability dimension, so that failure in energy, damage, fatigue, or another essential variable cannot be masked by favorable values in other dimensions. Third, it adds an analytical autopsy of collapse, distinguishing whether admission failure in hard regimes is primarily energy-limited, cycle-retention-limited, or substrate-stability-limited through the conjunctive sustainability gate Σ_H. The revised paper also clarifies the role of the update gains η+ and η− in the eligibility-trace dynamics. The positive gain η+ strengthens Λ_SWB when the candidate self-world-boundary transformation improves U_control, while the negative gain η− penalizes or suppresses Λ_SWB when the boundary fails to provide sustained causal utility. This makes the rejection of decorative boundaries explicit rather than merely reporting a final zero-weight outcome. This version also includes new schematic figures illustrating eligibility-trace rejection dynamics and the phase-transition topology of SWB admission failure. The paper is intended as a narrow, falsifiable contribution to computational biophysics, constructor-theoretic modeling, artificial consciousness research, and cECT-based task-admission architectures. The associated software artifact is cECT-SWB v2.6.0: Homeostatic Self-World Boundary Benchmark, published separately as a Zenodo software record. This preprint should be read together with that frozen code artifact and its validation outputs.