Wake-to-N3 coupling scores passed primary tests in 34 SEEG patients and eight fMRI participants but not SEEG spectrum and gap checks

We tested whether wake-to-N3 changes in distant cortical correlation patterns repeat across windows in separate human SEEG and fMRI studies. The fixed score removed additive site and distance effects. The mean cross-window product was 0.001656 in 34 SEEG patients (95% person interval 0.000507 to 0.002849) and 0.013393 in eight fMRI participants (0.008304 to 0.018914). Both primary Holm p values were 0.015625. The primary measured-property rule passed. A simulation-based amendment, fixed before human pattern scores, required a flat-spectrum check for a stronger claim. That check passed in fMRI but not for the SEEG mean: 0.000721, interval -0.000352 to 0.001789. All primary SEEG N3 windows were adjacent. A later check moved the second window to the segment end in the same 34 patients. Its mean was 0.000891, interval -0.000107 to 0.001816. This exploratory check did not establish robust temporal separation. Simulations show that local filters can produce a positive score under fixed instantaneous mixing. Thus the evidence supports positive original-processing scores in both studies, but not a shared effect robust to the stronger SEEG checks. Cohort, coverage, physical frequency, absent fMRI covariates, and fold dependence limit inference. No common edge map or causal neural mechanism is established. The package preserves plans, source versions and hashes, code, all person results, negative checks, figures, validation, and repeat instructions. Source waveforms remain external under their terms.

human neuroscience · SEEG · fMRI · sleep · correlation patterns · cross-validation · spectral control · reproducibility