A fixed sleep test did not confirm more nearby coupling in 21 SEEG patients and 15–16 fMRI units

We tested whether deep N3 sleep raises nearby coupling relative to distant coupling. Separate public studies supplied human SEEG and fMRI. A local plan fixed distance bins, two-block products, person splits and three shared tests. Discovery used 12 SEEG patients. Confirmation used 21 other patients and 16 BOLD records, treated as 15 or 16 units because two codes may name one person. The mean N3-minus-wake change was -0.000002 in SEEG, with a 95% person interval from -0.002502 to +0.002051. The merged fMRI estimate was -0.007667, with interval from -0.021941 to +0.005393. The shared rise test failed. A planned secondary N3 preference had positive signs in 16/21 SEEG patients and all fMRI units. Its shared sign p after correction was 0.03991. However, the SEEG mean interval crossed zero, so the full fixed rule failed. Site-mean removal also reversed that SEEG preference. Model tests show why a squared sample correlation need not show repeatable coupling. The contribution is a fixed spatial sleep benchmark, a limited secondary sign result and full negative results. Short windows, sparse clinical sampling, different cohorts, source processing and missing fMRI nuisance traces limit scope. No shared causal process is established.

SEEG · fMRI · human sleep · spatial correlation · independent confirmation · negative result