UCTORI-EFCV: Empirical-Formal Extension with Neurophysiological Cross-Validation

UCTORI-EFCV is an empirical-formal extension of UCTORI, the Unified Consciousness Theory with an Operator for Rebalancing Information, designed to make the Informational Rebalancing Operator, ORI, mathematically explicit, computationally reproducible, and empirically testable through neurophysiological cross-validation. The document formalizes ORI as a corrective dynamic acting on coherence discrepancy, ΔC, and evaluates post-perturbation trajectories through finite correction time, T_c, clean return, R, and structural persistence, π. Instead of assessing models only by static integration or global access, UCTORI-EFCV asks a dynamic and falsifiable question: after a perturbation, does the system reduce coherence discrepancy, how long does correction take, does it return cleanly to baseline, and what structural pattern is preserved? The work includes a reproducible synthetic benchmark comparing three operational proxy architectures: IIT-like, GWT-like, and UCTORI-EFCV/ORI. It also incorporates a pilot neurophysiological cross-validation using real EEG/PSG data from the CAP Sleep Database, subject n16. The analysis processes 711 episodes, including 504 MCAP events and 207 matched NREM OFF controls, using stratified 5-fold cross-validation, Mahalanobis-distance coherence discrepancy, model-specific feature sets, and operational endpoints including ΔC(t), T_c, R, π, AUC, average precision, and null-test behavior. The pilot results are favorable but deliberately interpreted with caution. UCTORI-EFCV shows high finite closure in MCAP events, strong structural persistence, and a significant positive relationship between initial coherence discrepancy and correction time. It outperforms the IIT-like proxy in MCAP versus control classification, but does not outperform the GWT-like proxy in this specific detection task, suggesting that CAP events contain a strong arousal or broadcast component captured by the GWT-like model. The result is therefore not presented as definitive confirmation, but as a reproducible, partially supportive, and falsifiable empirical pilot. For transparency and auditability, the PDF is self-contained. It includes visible full code in the appendices, reproduction commands, complete scripts, generated tables, figures, CSV/JSON outputs, SHA-256 hashes, a manifest of all artifacts, and embedded internal attachments required to reproduce and verify the analysis. The document is intended for researchers in consciousness studies, computational neuroscience, philosophy of mind, dynamic systems modeling, reproducible science, and formal theory comparison. This publication is linked to the English UCTORI reference document: https://doi.org/10.5281/zenodo.18305907.