In 8 SEEG patients, stimulus-period high-gamma showed no excess of difficulty coding, but was slightly higher on easier trials

We asked whether mean high-gamma activity (HGA) in the 0.5 s after random-dot stimulus onset varied with the adaptive staircase difficulty level in SEEG bipolar channels. We used archived per-trial features from 8 of 21 patients in OpenNeuro ds006253 (714 channels; 1,618 registered trials). The difficulty measure was the behaviour column coh, which matched stim_int in two checked files. Behaviour showed that higher values were harder trials, although the exact stimulus parameter could not be confirmed without task code. The pre-registered per-channel model regressed HGA on coh, current accuracy, and trial index. The primary test found 72 channels with p < 0.05, close to the 1000-shift null mean of 73.7 (95th percentile 99; p = 0.51), so it did not support an excess of difficulty-coding channels. The null mean was about 10% of channels because difficulty and HGA were autocorrelated; parametric channel p values should not be used alone. The secondary test found negative coh t values in 7 of 8 patients (mean −0.49; exact sign-flip p = 0.039), consistent with slightly higher stimulus-period HGA on easier trials. The direction remained the same with log reaction time or previous-trial accuracy covariates and after removing 15 no-response trials (code 99). For trials where movement began at least 0.5 s after stimulus onset, the direction remained but was weaker (sign-flip p = 0.055; circular-shift p = 0.018). The analysis includes only 8 patients, a narrow staircase range (within-patient SD 0.08–0.25 rad), and window means; it does not test time-resolved activity or accumulation. Slower session drifts may remain, and the exact stimulus meaning of coh is uncertain. The feature, source, and code hashes, analysis code, logs, and results are included with the original files.

stereo-EEG · high-gamma activity · random-dot motion · adaptive staircase · difficulty · OpenNeuro ds006253 · autocorrelation · circular-shift null · sign-flip test