Phase changes after a bipolar montage caused excess 8–30 Hz coherence rejections in independent contact models of two SEEG records

We tested a coherence null with power spectra from the first 32 seconds of two public SEEG records. The contact model had independent Gaussian spectral coefficients. In the primary 8–30 Hz test, independent phase changes after the bipolar montage rejected 100% of shared-contact model tests. A null that drew independent contacts before the montage rejected 4.93% and 5.08%. The stated rate was 5%. Edges without a shared contact stayed near 5% under both nulls. The main result used 64 model samples. Checks used 32 and 128 samples. The package contains source code, measured spectra, source versions and hashes, contact maps, model draws, ranks, figures, and an offline repeat method. The shared-contact mechanism is established in prior work. This contribution is a reproducible public-data calibration. It does not establish neural coupling or a patient population effect.

SEEG · coherence · bipolar montage · shared contact · null calibration · signal processing