In eight SEEG patients, early high-gamma was higher before slower decisions

An accumulation-to-bound model predicts higher early stimulus-locked activity before faster decisions and no increase in pre-movement activity with decision time. We tested these signatures in archived SEEG high-gamma (70–150 Hz) window means from eight patients and 714 channels in the OpenNeuro ds006253 (MetaRDK) random-dot task. Per-channel OLS estimated the t-value for log decision time while controlling for correctness and coherence; group inference used exact sign-flip tests over subjects. The predicted negative early relation was not found: mean t was +0.67 (95% bootstrap CI +0.22 to +1.21; 7 of 8 patients positive; p = 0.031, Holm p = 0.0625). Early high-gamma was higher before slower decisions. The stimulus-minus-pre-movement dissociation was not found (p = 0.31), and pre-movement high-gamma did not decrease with decision time (mean t +0.36; CI includes zero). Results remained positive with a 0.8 s decision-time threshold and, in an exploratory analysis, after adding trial order. Limitations include eight of 22 patients, window means rather than time courses, undocumented normalisation of derivative high-gamma, and incomplete control of trial difficulty because the staircase links coherence to recent performance. The archived derivatives were not downloaded again or independently byte-matched to a dataset snapshot.

SEEG · high-gamma activity · decision time · evidence accumulation · random-dot motion · sign-flip test · pre-registration