Intelligence Is Navigation: State Space Traversal, Self-Representation, and Metacognition as Coherence Dynamics

Existing definitions of intelligence capture single aspects: problem-solving (psychometrics), prediction (Hutter-Legg), compression (Kolmogorov), adaptation (evolutionary), integration (IIT). This paper unifies them: intelligence is coherent navigation of state spaces. Three components: traversal efficiency (how many states reachable), self-representation accuracy (does the system know where it is), and metacognitive correction (can the system fix its own errors). These map isomorphically to the coherence metric K(t) = ρ · I_Φ · F. The intelligence threshold is K_crit. Below it: drift. Above it: navigation. No existing formal definition includes metacognition — the component that fails first. Intelligence scales with coherent state space accessibility, not substrate size. Computation expands accessible space; coherence determines whether it is navigated or drifted through. Mutual coherence (E > 0) amplifies intelligence beyond individual capacity. AGI is K_total > K_crit across all relevant state spaces — achievable only with external boundary. Part of the Spektre research corpus.