Autocorrelation-Break Detection of Structural Regime Transitions

This paper presents the Keller-ΔΦ Protocol, a pre-registered time-series detection method designed to identify abrupt disruptions in short-range autocorrelation continuity within dynamical systems. The method was originally developed in cognitive psychology to detect transient “decoupling events” during human decision-making tasks and is here evaluated as a generalized signal-processing framework across multiple simulated domains.Three transition classes were tested using fixed, non-optimized parameters derived from prior cognitive datasets (N = 526 sessions):(1) abrupt financial volatility regime shifts,(2) seismic material fracture transitions, and(3) gradual cardiac homeostatic adjustment.The protocol successfully detected abrupt structural transitions in the financial and seismic simulations while failing to detect the gradual cardiac homeostatic shift. This negative result is explicitly reported and used to define the operational boundary conditions of the method.The findings suggest that the Keller-ΔΦ Protocol is sensitive to abrupt disruptions in local predictive continuity rather than to all forms of regime change. The work is therefore framed as a lightweight, model-free detector for autocorrelation-breaking structural reorganizations rather than as a universal transition detector.The paper additionally separates empirical findings from theoretical interpretation. While the protocol originated within the Mitchellian ΔΦ framework, the empirical results are presented independently of that interpretation and stand as a standalone signal-processing contribution.This Version 1.0 release contains:the full pre-registered protocol,simulation definitions,positive and negative cross-domain results,operational boundary conditions,methodological limitations,and theoretical separation statements.Future work will evaluate the protocol on empirical financial, seismic, physiological, and infrastructure-failure datasets under fixed pre-registered thresholds.