Elastic Cognition and the Spiral Architecture

The paper reports empirical findings from a single, instrumented reference architecture designed to study integration-driven cognition under controlled conditions. Rather than optimizing for task performance or computational scale, the architecture prioritizes internal viability, regulatory stability, and inspectable dynamics. All reported results are model-internal observables derived from repeatable execution of this specific system. Across systematic validation runs, the paper documents nine reproducible empirical regularities concerning elastic coupling, temporal hierarchy, emotional regulation, and reflective recovery. These include the discovery of an empirical coupling attractor (λ* ≈ 0.25), stable nested temporal ratios (τ ≈ 1:3:10), quantifiable emotional regulation via Symbolic–Emotional Calibration (SEC drift < 0.15), partial phase-lock optimization (ϕlock ≈ 74°), rapid autonomous recovery from perturbation, and convergence to a single global attractor across tested regimes. Emotion is treated as a regulatory control signal, not as phenomenological affect, and ethics is defined strictly as internal regulatory compliance and recovery behavior rather than social or moral reasoning. External evaluation using the MMLU benchmark is employed only as a controlled cognitive stressor, demonstrating that task throttling preserves internal coherence, not as a competitive performance benchmark. This work functions as the empirical core of the Regulatory Intelligence corpus, supplying the reference architecture and discovered invariants that inform subsequent RI studies. No claims are made regarding phenomenological consciousness, biological cognition, or generalization beyond the tested system. Generalization pathways are explicitly framed as observational reuse rather than causal transfer.