Imaginary coherency reduced beta null rejection for shared-contact bipolar edges in an independent contact phase model of eight SEEG recordings

We tested phase controls on fixed Fourier amplitudes from eight public SEEG recordings. Each person had six sections and 16 selected shared-contact edges. We made 50 null observations from independent contact phases and used 99 controls per test. At p <= 0.05, output phases gave a mean rejection rate of 99.25% for coherency magnitude. The rate was 0.02% for absolute imaginary coherency. The latter test rejected 90.59% of a specified spectral phase-lag control. An exact proof shows that output phases add half the squared shared-contact power to the mean imaginary cross-product variance. Thus zero mean does not ensure the same null distribution. The result applies to this fixed-amplitude phase model. It does not establish neural communication. The package contains code, data, source versions, checks, and reproduction instructions.

imaginary coherency · bipolar SEEG · shared-contact edges · phase controls · null calibration · wPLI · spectral analysis · finite-sample inference