SEEG high gamma activity weakly decoded change of mind after movement onset in 15 patients

We tested whether SEEG 70–150 Hz high gamma activity (HGA) decoded a change of mind during a motion decision task and whether it added information to behaviour. We analysed archived per-trial features from 15 patients, with 457 change-of-mind trials among 3,970 trials. Repeated stratified five-fold cross-validation with shrinkage LDA gave a mean AUC of 0.544 for HGA in the 0–0.5 s after movement onset (11/15 patients above 0.5; sign-flip Holm p = 0.046). HGA before movement onset (AUC 0.517) and the pre-stimulus negative control (AUC 0.521) were not significant. A behaviour model using reaction time, motion coherence, accuracy, peak velocity, and movement duration gave AUC 0.878. Adding all move-window HGA channels reduced AUC by 0.074 in all 15 patients. In this offline analysis, HGA gave only a weak change-of-mind signal, while behaviour, especially movement kinematics, carried more information. The result is limited by non-time-ordered cross-validation, a fixed movement window, differences in movement kinematics, few positive trials per patient, and a full-channel model that may overfit. Channel selection and other frequency bands were not tested.

SEEG · high gamma activity · change of mind · motion decision · brain-computer interface · shrinkage LDA · cross-validation · movement kinematics