Testing Integrated Information Theory with ECLIPSE: A Population-Scale EEG Study

═══════════════════════════════════════════════════════════════DEPRECATION NOTICE - VERSION 1.0 SUPERSEDED═══════════════════════════════════════════════════════════════ This version (v1.0) contains critical methodological errors and should NOT be cited. A corrected version (v2.0) supersedes this publication entirely. ═══════════════════════════════════════════════════════════════ERRORS IDENTIFIED:═══════════════════════════════════════════════════════════════ 1. CRITICAL METHODOLOGICAL ERROR - SMOTE Artifacts The analysis employed SMOTE (Synthetic Minority Over-sampling Technique) to address class imbalance. However, SMOTE with single-feature data (Φ) creates artificial interpolated values that do not correspond to real neurophysiological states. IMPACT: The reported negative correlation (ρ = -0.45) is likely a methodological artifact, not a genuine empirical finding. SEVERITY: This invalidates the main empirical conclusions. 2. DOCUMENTATION ERROR - Channel Count The manuscript states analysis used "8-channel EEG recordings." Post-publication review revealed Sleep-EDF contains only 2-channel data (EEG Fpz-Cz and EEG Pz-Oz). IMPACT: Affects interpretation of spatial resolution and computational complexity estimates, but does not affect validity of computations performed. SEVERITY: Documentation error, not computational error. ═══════════════════════════════════════════════════════════════SCOPE OF INVALIDATION:═══════════════════════════════════════════════════════════════ INVALID FINDINGS:├─ Negative correlation (ρ = -0.45) - SMOTE artifact├─ Φ higher in sleep than wake - SMOTE artifact├─ Numerical values of Φ - Distorted by SMOTE└─ Falsification verdict - Based on artifactual data VALID COMPONENTS:├─ ECLIPSE framework design - Methodologically sound├─ Pre-registration protocol - Correctly implemented├─ Cryptographic verification - Functions as intended└─ Integrity metrics (EIS, STDS) - Correctly computed (Note: These metrics detected no data snooping, as none occurred. They cannot detect inappropriate preprocessing choices like SMOTE.) ═══════════════════════════════════════════════════════════════STATUS AND CITATION:═══════════════════════════════════════════════════════════════ This version (v1.0) is preserved for:├─ Transparency in error correction├─ Version control documentation└─ Demonstration of ECLIPSE's integrity mechanisms DO NOT CITE v1.0 for empirical findings regarding IIT. CITE v2.0 instead: [DOI will be added when published] ═══════════════════════════════════════════════════════════════LESSONS LEARNED:═══════════════════════════════════════════════════════════════ 1. SMOTE should not be used with single derived metrics2. Pre-registration prevents p-hacking but not preprocessing errors3. Integrity metrics (EIS, STDS) detect analytical flexibility, not inappropriate method choices4. Computational reproducibility requires complete method specification This error detection and correction exemplifies scientific self-correction. The ECLIPSE framework functioned as designed—the error was in preprocessing choice, not protocol adherence. ═══════════════════════════════════════════════════════════════CONTACT:═══════════════════════════════════════════════════════════════ Camilo Alejandro Sjöberg Tala, M.D.Email: [email protected]: 0009-0009-6052-0212 Questions regarding this deprecation: [email protected] ═══════════════════════════════════════════════════════════════