In ds004703 SEEG, high-gamma activity did not track English phone timing at the group level

We tested whether the SEEG high-gamma (70–150 Hz) envelope follows the timing of phones in English passages in OpenNeuro ds004703 (9 subjects, 1162 channels). We used 20-Hz high-gamma envelopes from AutoSci result c6e54f80 and phone onsets from dataset event files. For each channel, we calculated the correlation between the high-gamma envelope and phone-onset counts in 50-ms bins at lags of 0–300 ms, then compared the maximum correlation with a null from 1000 circular shifts of the phone regressor. The analysis plan was written before the statistics were calculated. The primary group test was null: the mean of subject mean z values was 0.06 and the exact two-sided sign-flip p was 0.80 (4 of 9 subjects positive). Only subject SD019 had channels that followed phone timing: 65 of 232 channels had Benjamini–Hochberg q < 0.05, and most peak lags were 50–150 ms. The other 8 subjects had no channels with q < 0.05. Many subjects heard only 1–2 English passages of 20 s, phone rate is only a proxy for the speech envelope, and electrode locations were not checked. A missing effect in a subject may reflect low power or poor auditory coverage; it does not show that the cortex does not track speech.

stereo-EEG · high-gamma activity · speech tracking · phone timing · passive listening · neuroscience