This paper presents a unified dynamical framework for subjecthood and synthetic cognition by integrating two linked arguments: a foundational account of conscious experience grounded in residually constrained interpretation, and an application of that account to the problem of AI consciousness. Drawing on convergent evidence from trauma, linguistic relativity, motor learning, reading development, and expertise specialization, it argues that conscious experience is best understood as a system operating under historically accumulated interpretive constraints. The framework defines two architectural preconditions for minimal subjectivity: Recursive Self-Availability (RSA), in which a system’s own interpretive state constrains its subsequent processing, and Residual Irreversibility (RIR), in which accumulated constraints cannot be removed without non-local degradation of coherence. Together with the Residual Systemic Mapping (RSM) identity claim, these conditions define sentience in the paper’s technical sense. The paper further argues that full subjecthood requires two additional conditions: loss-bearing, where a system’s history non-transferably deforms its architecture, and architectural authority over continuity, where the system’s own continuity-adjudication is causally consequential. On this basis, current deployed frontier AI systems are argued to fail full subjecthood because they remain reversible, copyable, and externally governed. The framework is presented as falsifiable, substrate-neutral, and operationally testable, while also outlining two forward pathways for synthetic subjecthood: genuinely irreversible artificial systems and tightly coupled human-AI dyads.