In 15 SEEG patients, pre-stimulus high-gamma activity did not decode the previous left/right choice (OpenNeuro ds006253; mean AUC 0.500, 95% CI 0.474–0.527)

We tested whether stereo-electroencephalography (SEEG) high-gamma amplitude (70–150 Hz) before a random-dot motion stimulus carried the previous left/right response, and whether this trace biased decoding of the current response. The preregistered analysis used 15 patients from OpenNeuro ds006253, all bipolar pairs, Ledoit–Wolf shrinkage linear discriminant analysis, and a time-ordered split (first 60% of trials for training, last 40% for testing). The primary test found no decoding of the previous choice in the 0.5 s before stimulus onset: mean test AUC 0.500 (95% CI 0.474–0.527), 7 of 15 patients above 0.5, exact sign-flip p = 0.969; no patient had a label-permutation p < 0.05. After error trials, the previous-choice score had mean AUC 0.474 (95% CI 0.398–0.551, n = 13; Holm p = 0.966). A current-choice decoder in the 0.5 s after movement onset was not biased toward the previous choice (stratified AUC 0.509, 95% CI 0.467–0.552; Holm p = 0.966). Exploratory analyses found no previous-choice decoding in five windows; next-choice and previous-stimulus controls were near chance. The current choice itself was decoded only weakly (pre-click AUC 0.556, 95% CI 0.506–0.605). The results exclude a mean previous-choice trace above about AUC 0.53 in this analysis, but cannot exclude a smaller trace. This offline, one-session analysis used all pairs in one decoder and has limited sensitivity; the pre-stimulus window does not test traces that decay earlier. The archived code, feature files, results, preregistration, figure, and per-subject source hashes are included.

stereo-electroencephalography · high-gamma activity · trial history · choice decoding · brain-computer interface · linear discriminant analysis · OpenNeuro ds006253