White matter stimulation advantages appear with referential but not bipolar recording in ds004696

Archived pair tables from two prior analyses of OpenNeuro ds004696 disagree on whether white matter (WM) stimulation sites yield larger responses than gray matter (GM) sites. We applied a common distance-adjusted, site-label permutation analysis to 36 variants. In four shared subjects, all 18 referential variants had positive WM-minus-GM effects (+0.06 to +0.46 SD), with 12 at p < 0.05. All 18 bipolar variants had effects from -0.07 to 0.00 SD, with none at p < 0.05. This pattern held for detection and response size, alternative distance ranges, GM-only recording contacts, and inclusion of subject 01. Removing WM recording contacts reduced the referential effect. The disagreement therefore tracks recording montage rather than outcome type: the WM advantage is present in the shared broad signal, but the bipolar results do not show a local advantage. The analysis reuses archived tables and does not check responses by an expert or model artifacts. The tables differ in channel subset and time window, so montage is not fully separated from those factors; a direct test requires all montages computed from one channel set in the raw MEF3 data.

white matter stimulation · stereo-EEG · recording montage · bipolar recording · distance adjustment · permutation analysis