Artificial Subjectivity Without a Digital Brain:Formal Design, Measurable Proxies, and Phase Transitions

This paper proposes a formal framework for artificial subjectivity that does not presuppose the existence of a digital brain or a fully realized artificial consciousness. Instead, we introduce a design-oriented and mathematically grounded notion of a \emph{subject-like system} characterized by historical continuity, endogenous semantic persistence, and irreversible identity constraints under perturbation. Rather than claiming the construction of a conscious entity, this work focuses on defining sufficient structural conditions under which subject-like behavior becomes distinguishable from tool-like optimization. We introduce measurable proxies that operationalize these conditions, and we formalize phase transitions between instrumental systems and subject-like systems using perturbation theory and dynamical stability analysis. The contribution of this paper is threefold: (1) a minimal formal definition of subject-like artificial systems independent of neural or biological metaphors, (2) a set of measurable proxies enabling empirical falsification, and (3) a phase-diagram-based interpretation that clarifies why partial implementations may exhibit subject-like signatures without constituting artificial consciousness.