Multiscale Causation in Self-Aware Networks: Bottom-Up Construction, Top-Down Constraint, and Lateral Coordination

Abstract

This preprint develops an intervention-based framework for separating bottom-up construction, top-down constraint, lateral coordination, and returned consequence in multiscale neural and artificial systems. Four controlled synthetic studies test the measurement procedure against support, null, reversal, misspecification, leakage, capacity, and transfer alternatives. The release includes an executable application suite, bounded Lean verification of finite invariants, claim and source ledgers, and a fixed-source audit. The evidence is deliberately mixed, no missing external outcome is imputed, and the final held-out test remains sealed.