For gaps of about 4 s or more, time since the previous key press did not change rest-window beta power or HG+beta decoder scores in 8 SEEG subjects

We tested whether the time from a previous cued key press to a rest window was associated with rest-window features or out-of-sample rest-versus-move decoder scores in 8 subjects from OpenNeuro ds004473 v1.0.2. Gaps had subject-level medians of 5.1–7.3 s. A shrinkage LDA trained on trials 1–100 was scored on trials 101 onward. The mean Spearman correlation between gap and rest beta power was −0.034 (6/8 subjects negative; exact two-sided sign-flip p=0.41; Holm-adjusted p=0.81). For the rest-window HG+beta decoder score, mean rho was +0.003 (p=0.95; Holm-adjusted p=0.95). Long-gap minus short-gap test AUC was +0.010 for HG+beta (p=0.30); partialling out trial index did not change the primary results. Exploratory analyses found more bipolar pairs with rest-beta correlations above the shuffle-null threshold than expected (11.7% versus 5%, p=0.016); a post hoc circular-shift null gave an excess of 4.5 percentage points (7/8 subjects, p=0.031), mostly from subject 12, whose HG AUC was 0.50. Move-window HG+beta scores tended to increase with gap (mean rho +0.085, p=0.06). The results apply to gaps of about 4 s or more and cannot test shorter gaps, where post-movement beta rebound is largest. The study included 8 subjects, and the per-pair analysis was exploratory.

stereo-EEG · SEEG · brain-computer interface · motor decoding · beta power · high gamma · linear discriminant analysis · movement history · OpenNeuro