This paper proposes the Cognitive Circuit Hierarchy (CCH) model, a new theoretical framework that redefines intelligence as a property of the brain’s static neural architecture. Whereas previous research has focused on dynamic functional modes—such as the default mode network (DMN), central executive network (CEN), and the language network—these modes capture only momentary states and cannot explain why individuals consistently gravitate toward particular cognitive orientations across their lifespan. Building on a unified model in which understanding and memory form a single recurrent process, we show that static structural connectivity among Broca’s area, the prefrontal cortex (PFC), the parietal cortex, the frontopolar cortex, and the posterior cingulate cortex (PCC) determines the relative strength of the five hierarchical levels of understanding (Levels 1–5). From these structural differences, we derive a two‑dimensional model of intelligence consisting of five cognitive backbones (L/C/A/M/G) and three salience‑reactivity profiles (S/M/W), the latter reflecting the frequency of mode transitions among DMN, CEN, and the language network. An analysis of ninety historical figures demonstrates that these structural backbones are consistently reflected in their thought patterns, behaviors, and written works, providing external validity for the model. Extending the framework to artificial intelligence, we argue that current large language models (LLMs) correspond to an L0 type, possessing only the linguistic layer. We further show that adding causal (C), abstract (A), and model‑building (M) layers—along with an artificial salience network—would enable the construction of C/A/M‑type higher‑order intelligence. The CCH model thus offers a unified theory of intelligence applicable to both humans and AI, and provides foundational design principles for the development of next‑generation artificial cognitive architectures.
Yasumitsu Nakahama (Fri,) studied this question.