Intentionality as Architectural Self-Organization: Generative Closure, Ontological Expansion, and Cognitive Agency

This paper develops the framework of Evolving Cognitive Architectures (ECA), a theoretical account of intentional cognition that identifies both the organizational conditions required for cognitive agency and the developmental mechanisms through which cognitive systems overcome their own generative limitations. Rather than treating intentionality as a consequence of consciousness, interpretation, evolutionary history, or predictive adaptation, the framework argues that intentionality emerges from the ongoing architectural organization and developmental dynamics of cognitive systems. ECA specifies five constitutive organizational conditions—organizational continuity, dynamic circulatory memory, representational plasticity, non-arbitrary internal semantics, and developmental plasticity—that enable directed, context-sensitive, self-modifying, and developmentally coherent cognition. Beyond specifying these conditions, the framework explains how intentional systems avoid generative closure through representational reorganization, ontological expansion, imaginative exploration, and meta-representational control, thereby sustaining open-ended cognitive development rather than merely optimizing within fixed representational structures. The paper makes four contributions. First, it provides a diagnostic critique of influential approaches associated with John Searle, Daniel Dennett, and Ruth Millikan, arguing that they identify correlates, diagnostics, or stabilizing mechanisms rather than the constitutive organizational conditions of intentionality. Second, it advances a positive architectural account that relocates intentionality to the level of organized cognitive dynamics. Third, it develops a theory of representational transformation that explains how cognitive systems recognize generative limitations, reorganize their representational architecture, expand their ontological resources, and construct novel representational organizations. Fourth, it demonstrates how the ECA framework provides principled criteria for evaluating cognitive architectures by shifting assessment from behavioral performance to architectural capacity, thereby clarifying the organizational requirements for creativity, conceptual innovation, and AGI-level cognition. Contemporary artificial intelligence systems motivate the need for such an architectural framework. Although they increasingly exhibit sophisticated behavioral performance, contextual adaptation, and coherent output generation, they generally remain constrained by fixed representational organizations that limit sustained self-modification, ontological expansion, and open-ended cognitive development. Within the ECA framework, these limitations are interpreted not primarily as failures of performance but as manifestations of generative closure arising from insufficient architectural plasticity. By reframing intentionality as an achievement of architectural self-organization rather than a property of substrate, interpretation, or historical origin, and by identifying the mechanisms through which cognitive systems expand beyond existing generative constraints, the paper provides a structural framework through which intentionality, cognitive agency, and representational development can be comparatively analyzed and, in principle, empirically investigated across both biological and artificial systems.