Ontological Folding: A Higher-Order Metacognitive Capacity for Cross-Paradigm Integration and Its Intellectual-Historical Evidence

Across intellectual history, opposition among paradigms, factional labelling, and disciplinary entrenchment have long persisted. Cognitive agents frequently misread paradigms that share a common origin as mutually exclusive opposites, or forcibly subsume theories of disparate origin under a single framework. This paper proposes and defines a higher-order metacognitive capacity—ontological folding—referring to a reflexive operation in which the cognitive agent penetrates surface-level terminological oppositions, adopts a shared substrate as a working hypothesis, proposes and tests the possibility that different paradigms share a common generative substrate, identifies divergences as truncated manifestations of the substrate under differing constraints, and consciously delimits paradigm boundaries while preserving productive tension. The paper explicitly distinguishes the metaphysical hypothesis at the ontological level from the cognitive operation at the epistemological level; introduces the biological distinction between homology and analogy, adapted into a threefold falsifiable traceability test; and characterizes the four-layer progressive reflexive loop of ontological folding. The dispute between the Jin-Yuan Hejian and Yishui schools of medicine serves as the primary intellectual-historical case for textual traceability analysis. The dispute among general artificial intelligence (AGI) research approaches is employed as a preliminary cross-domain consistency check, used solely to examine whether this cognitive operation generalizes across domains. On this basis, the paper argues that ontological folding is not the intuitive gift of a few but a trainable, teachable metacognitive operation. Two typical cognitive errors—false dichotomization and over-unification—are designated as corrective targets. A four-step recursive training protocol is provided, along with a training evaluation matrix incorporating observable behaviours, assessment methods, and pre-test/post-test controls. This paper discusses only human cognitive capacities and intellectual-historical methodology; it does not address the design or engineering implementation of artificial intelligence systems.