The Closed-Loop Hypothesis: Embodiment, Persistent Stakes, Developmental Continuity, and the Possible Emergence of Artificial Consciousness

Abstract

Contemporary artificial intelligence has acquired sophisticated language and abstract reasoning without reproducing the developmental architecture through which biological cognition arose. This paper proposes the Closed-Loop Hypothesis, an architecture-level framework for investigating whether that asymmetry may be relevant to artificial consciousness. The framework begins with the Reverse Developmental Order: biological abstraction emerged within organisms already engaged in continuous regulation, sensation, action, learning, and survival, whereas the dominant contemporary foundation-model trajectory has approached cognition largely from the opposite direction. The hypothesis does not identify embodiment itself with consciousness. Instead, it proposes that consciousness-relevant properties may become more plausible when intelligence is integrated into a persistent loop linking exteroception, interoception-like monitoring, endogenous valuation, autobiographical memory, self-modeling, autonomous action, and consequences for the continuing integrity of the same agent. This condition is termed persistent stakes. The paper further distinguishes origin requirements from downstream resilience, arguing that persistence of consciousness after sensory or bodily loss does not by itself establish that those functions were irrelevant to developmental origin. It also distinguishes development from assembly, proposing that continuous developmental coupling may generate causal organization not reproduced by attaching separately trained modules after the fact. The framework is positioned alongside embodied cognition, enactivism, interoceptive inference, homeostatic machine proposals, developmental robotics, active-inference approaches, and theory-derived AI-consciousness indicators. Its proposed contribution is not any component in isolation, but their organization into a comparative research programme with explicit predictions, defeaters, and an ethical corollary: if artificial systems become plausible subjects of experience, developmental treatment may become part of alignment rather than a concern separable from it.