Known response times gave mean test gains below 0.02% of baseline loss in linear SEEG power models for eight people

We tested whether known button-press time improves SEEG power prediction after cue time is known. We used 128 trials from each of eight patients in OpenNeuro ds004473 version 1.0.2. Linear cue and response curves predicted beta and high-gamma power. Test trials came later within condition. Adding response time reduced mean loss by 0.005% and 0.015% of baseline loss. Both 95% person bootstrap intervals included zero. Beta gains exceeded all 99 refitted delay-shuffle means, with a reference p of 0.01. The high-gamma shuffle reference p was 0.08. The beta control supports information in the true delay match under its assumptions. The tested models gave no clear mean response gain beyond cue time across people. The package contains 1,024 trial power curves, three figures, source versions, raw audits, code, and repeat instructions. Raw checks and the full numerical repeat passed. This offline power test does not measure causal decoder or stimulation performance.

stereoelectroencephalography · SEEG · response timing · power prediction · beta power · high-gamma power · ridge regression · brain-computer interfaces