In five SEEG subjects, high-gamma speech–music selectivity was not above chance after correction

We compared high-gamma (70–150 Hz) power between speech and music blocks of a naturalistic audiovisual film in five subjects with only SEEG electrodes from OpenNeuro ds003688 v1.0.7. Across 491 same-shaft bipolar pairs, 39 were selective at exact two-sided block-permutation p < 0.05, compared with a group-null mean of 24.3 (95th percentile 37; p = 0.031; Holm-adjusted p = 0.094). The primary test therefore did not pass the pre-registered correction. There was no group direction preference (20 speech-greater-than-music versus 19 music-greater-than-speech; p = 0.85, Holm-adjusted p = 1.0), and HG and beta effects were not negatively coupled (mean Spearman rho = +0.065; p = 0.77, Holm-adjusted p = 1.0). The 4 s sampled at each block centre yielded only 52 s per subject and low test power. Speech and music blocks also differed in visual content; five subjects were analysed, and 11 mixed ECoG+SEEG subjects were not. Exploratory theta, alpha, and beta counts were not pre-registered and were uncorrected. The package preserves the pre-registration and amendment, analysis code, per-pair results, and source byte-range provenance.

stereo-EEG · SEEG · high-gamma · speech · music · naturalistic stimulation · block permutation · OpenNeuro