Phase randomization after bipolar subtraction gave excess rejections in independent contact controls from two SEEG recordings

We tested phase null models before and after bipolar subtraction. Controls used spectra from two public SEEG recordings. The original recorded contact signals were independent in these controls. Bipolar channels could still share a contact. In the 13-30 Hz band, phases after subtraction gave rejection rates of 100% and 99.996%. Phases before subtraction gave rates of 5.073% and 4.979%. The test level was 5%. Pairs with separate contacts had rates near 5% for both models. The covariance formula explains the dependence due to a contact shared by bipolar channels. The result concerns the specified control model. It does not identify false neural connections in the original data. The package preserves input spectra, source hashes, code, all 310720 control tests, a mask for contact reuse, and reproduction instructions.

SEEG · bipolar referencing · phase randomization · contact reuse · negative control · connectivity · Monte Carlo testing · covariance