From Collective Computation to Developmental Inquiry: A Cross-Substrate Test of Structural Admissibility

This paper traces an argument across four levels of analysis: theoretical foundation (Hopfield, 1982), developmental evidence (Vargas-Barroso et al., 2026), formal specification (Geometric Inquiry Theory), and empirical test (MARL Protocols 2-6). Hopfield established that computation can emerge from network organization without component-level understanding. Vargas-Barroso et al. demonstrated that functionally useful structure develops through pruning from dense/random toward sparse/structured connectivity. GIT proposes a candidate minimal inquiry structure. Six preregistered MARL protocols show patterns consistent with degraded governance behavior under varying constraint architectures. Whether these patterns are substrate-independent is addressed by a preregistered cross-substrate test (Protocol 8). Whether the proposed operational structure improves AI reasoning is the subject of a separate preregistered evaluation study. This paper contributes a falsifiable argument with locked empirical tests, not proof.