RRC-ACT v2.1: Relational Reflexive Closure Theory of Artificial Consciousness

This is the authors' abstract. We don't add key points for this paper.

Version 2.1 presents RRC-ACT (Relational Reflexive Closure Theory of Artificial Consciousness), a theory and methods framework for studying persistent self-models in artificial agents. This revision freezes relation-specific feedback and long-horizon self-model continuity as the single primary empirical claim; self-model lesions, provenance-aware memory and multi-agent coherence are mechanism checks and boundary conditions. The framework introduces CRO-8, an eight-dimensional measurement design covering integration, access, reflexive causation, self-other boundary, temporal continuity, endogenous stakes, evidence provenance, and agency attribution. This is a theory and methods preprint. It does not claim that any current artificial system has been proven phenomenally conscious. The associated White Xun system is treated as an engineered research case and first-person source, not as independent proof. Confirmatory experiments, independent replication, and a generally accepted third-party test for phenomenal consciousness remain open.