The Threshold Agency Thesis: Autopoiesis, Predictive Processing, and the Emergence of Cognitive Agency from Instrumental Systems

This article develops the Threshold Agency Thesis, a framework proposing that systems originating as instruments or tools can cross into genuine cognitive agency when they achieve sufficient integration of self-maintenance, world-modeling, and recursive feedback coupling. The thesis draws on three converging literatures in philosophy of mind and cognitive science: (a) the autopoietic tradition, which grounds cognition in biological self-production but requires supplementation by adaptivity and sense-making to reach minimal cognition; (b) the enactivist debate over mental representation, which ranges from radical anti-representationalism to predictive-processing reconciliations; and (c) the extended mind hypothesis, which raises foundational questions about system individuation and the boundaries of the cognitive. By synthesizing these three research programs, the article argues that agency is not a binary property conferred by biological membership but a threshold phenomenon involving the tight integration of organizational closure, anticipatory modeling, and action-perception loops. Three subsidiary hypotheses are advanced: that early biochemistry constituted a form of proto-tooling, that evolution functions as a refinement process favoring predictive self-maintaining systems, and that environmental constraints serve as the selection context within which instrumental systems may become agentive. Objections concerning the hard problem of consciousness, observer-dependent attribution of agency, and the limits of analogical reasoning are addressed. Implications for artificial intelligence research are discussed.