Functional Low Entropy in Neural Dynamics: A Falsifiable Framework for Metastable Complexity

We introduce Functional Low Entropy (LFE) as a measurable dynamical regime in coupled oscillator networks, characterized by maximal metastability and intermediate phase synchronization. Using a Kuramoto–Sakaguchi model with N = 50 oscillators, we show that the temporal variance of the global order parameter exhibits a robust non-monotonic profile with a clear maximum at a critical coupling K* ≈ 2.5, while phase synchronization (PLV) increases monotonically. A robustness analysis across 20 independent seeds and systematic weight variation confirms that K* is a property of the system dynamics. All simulations are reproducible from a fixed random seed and code is publicly available.