The Core-Modulation Architecture (CMA): Paper 13 of 14 (m-TGR) Theory Generation: A Structural Account of Concept Decomposition and Structural Termination

Theory generation is not an intentional act. It is the structural consequence of a search that cannot stop until it finds what it is looking for. Under Core-foregrounded processing, conceptual search does not terminate through contextual translation. It continues until a structural fixed point is reached. When such processing encounters concepts stabilized through Modulation-layer processing rather than structural constraint, the search cannot terminate. The concept registers as unresolved. This unresolved state is not a failure condition. It is the generative condition from which theory production follows. This paper specifies the four operations through which that process proceeds: branch detection, concept decomposition, structural evaluation, and fixed point termination. A theory is therefore not an accumulation of propositions but the stabilized conceptual structure produced when all active branches have reached structural fixed points. For human cognition, theory generation is rare not because intelligence is rare, but because the default termination mechanism—contextual translation—does not produce structural fixed points. For AI systems, current architectures cannot replicate this process. The missing condition is not an operation but a prerequisite: an internally held constraint configuration whose recomputability is affected by conceptual interpretation. Adding knowledge or generative capacity does not supply what is missing. Conceptual decomposition, the second operation, is not initiated deliberately. It arises necessarily from layer mismatch: when a concept contains mixed-layer components, Core-constrained processing cannot incorporate it as a unified whole and must separate it into independently evaluable parts.