Toward a Functional Threshold for Conscious-Like Behavior: Self-Sustaining Relational Systems as a Unit of Analysis

Abstract

The question of whether artificial systems can be considered conscious remains unresolved, in part due to the absence of operational definitions that permit empirical evaluation. Existing approaches emphasize neural correlates, information integration, or subjective reports, but lack observable behavioral criteria that apply across biological and artificial systems. The present paper proposes a functional framework for identifying conscious-like behavior grounded in behavior analysis, Relational Frame Theory (RFT), Relational Density Theory (RDT), and nonlinear dynamic systems. Conscious-like behavior is defined as a system-level regime in which relational responding becomes self-sustaining, globally integrated, and stably regulated under perturbation. Four operational criteria are proposed: self-sustaining relational activation, constraint under perturbation, cross-domain integration, and self-referential modulation. This framework does not address subjective experience; instead, it defines a functional threshold at which systems transition from externally driven responding to self-sustaining relational organization within environmental constraints.