High-gamma activity before mouse movement onset distinguishes movement preparation from a same-trial post-stimulus state in five SEEG patients

In five patients performing a mouse-response task, we compared 70–150 Hz high-gamma activity (HGA) in the 200 ms before movement onset with activity in an earlier post-stimulus window from the same trial. Trials with stimulus-to-movement times below 1.2 s or NaNs were excluded. A time-ordered five-fold cross-validated Ledoit–Wolf shrinkage LDA gave held-out AUC above 0.5 in all five patients (mean 0.612; median 0.616); the one-sided exact sign-flip test across patients gave p = 0.031, the minimum possible with five patients. Within-trial label-swap permutation p-values were below 0.05 in three of five patients. In the descriptive lead-time curve, median AUC was 0.53 for a window ending 0.8 s before movement onset and 0.62 for a window ending at onset. The windows differ in time since stimulus, so a slow post-stimulus change could contribute to the AUC; the earlier-window result argues against a large such effect but does not remove it. This offline result covers five patients, one task and one session. The moderate AUC does not provide a single-trial movement trigger by itself.

stereo-EEG · high-gamma activity · mouse movement · movement onset · shrinkage LDA · within-trial control · AUC · OpenNeuro