For four hundred years, Western philosophy has been confined within the subject–object dichotomy, persistently asking how the subject accesses the object yet rarely tracing the genesis of this division itself. This paper addresses that fundamental question by developing a generative account of human cognition. Its core claim is that cognition does not passively represent a pre-given world, but participates in the generation and stabilization of reality as it appears. First, it establishes Quxiang–Bilei (image extraction and analogical correlation) as the meta-operation of all cognition, structured as a closed loop: Quxiang (extracting stable structural differences) → Bilei (analogical correlation) → verification → feedback. Second, it shows that xiang (image) evolves through four stages—primordial, symbolised, formalised, and algorithmic—each sustained by the same Quxiang–Bilei logic. Third, it reconstructs the genesis of the subject–object dichotomy in four phases: primordial differentiation → symbol fixation → formation of the self-image → theoretical reification. Fourth, it reinterprets Descartes, Kant, and Hegel: Descartes’ cogito is a structural displacement of the lived body by a symbolically constituted self-image; Kant’s a priori categories are operational paradigms sedimented through long-term survival verification and limit convergence; Hegel’s absolute spirit is a totalising expansion of cognition that neglects the unimageable field. This paper provides a generative grounding for the subject–object dichotomy in Western epistemology, clarifies the limits of artificial intelligence, and shows that reality manifests itself as a stable generative structure through the Quxiang–Bilei loop. Cognition constructs xiang, not reality as such.
Lijun Chen (Mon,) studied this question.