What kind of entity is a belief? A structural ontology

The contemporary philosophical debate on belief spans several traditions, including the propositional attitude model, dispositionalism, functionalism, predictive processing, and social epistemology. Although each captures important aspects of the phenomenon, none explicitly addresses a more basic question: what kind of entity a belief is in minimal ontological terms. This article argues that the dominant framework presupposes an unexamined assumption — that beliefs are discrete representational states with propositional content — and identifies three structural limits of this assumption: its difficulty accounting for non-propositional doxastic phenomena, its inability to ontologize the exclusionary dimension of believing, and its persistence in treating belief as an isolated representational unit. Hybrid approaches that combine propositional and subdoxastic states are shown to remain ontologically insufficient because they describe functional coexistence rather than structural unity. In response, the paper proposes a minimal structural ontology of belief. On this view, beliefs are best understood as networks of stabilized functional assignments that organize processing within a system by linking discriminable configurations with recurrent orientations of response. This infrastructure includes both positive assignments and exclusion functions that establish structural boundaries of admissibility within the system’s processing space. The article concludes by formulating four criteria that jointly define the threshold of a genuinely doxastic infrastructure.