Common-average and bipolar beta signs differ at 220 of 834 pairs in eight SEEG recordings

We tested reference sensitivity in public OpenNeuro ds004473 version 1.0.2. We used the first 80 eligible trials marked correct in each of eight people with epilepsy. Common-average anchor and adjacent-contact bipolar beta contrasts had opposite signs at 220 of 834 eligible pairs. The mean subject fraction was 26.3% (95% subject bootstrap interval 22.6% to 29.7%). Only 15 opposite-sign pairs reached 0.5 dB in both references. No pair reached 1 dB in both references. Only 55 opposite-sign pairs kept both signs in both interleaved trial halves. No opposite-sign pair had q < 0.05 in both references. Most sign differences were small or unstable. The signals have different spatial support. The result does not identify a preferred reference. A metadata audit found that all 2,059 synchronized trial samples point to fixation, and that sub-1 correctness flags disagree with the declared key and deadline in 104 trials. Code, source metadata, power estimates, range hashes, checks, figures, and repeat instructions are included. This study is exploratory and applies to this task sample.

stereoelectroencephalography · SEEG · beta power · signal reference · bipolar reference · common average · reference sensitivity · power contrast · OpenNeuro · exploratory study