State centering changed mean beta–gamma power correlation by 0.00055 in later rest and pre-response windows from eight SEEG recordings

We tested beta–gamma power correlation in public motor-task SEEG from eight people. The source was OpenNeuro ds004473 version 1.0.2. The test used 834 adjacent contact pairs and 674 later trials. Each trial supplied rest and pre-response windows. State centering changed mean Pearson correlation from 0.11969 to 0.12024. The paired change was 0.00055, with a descriptive 95% person interval of −0.00183 to 0.00265. Eight pairs per person, selected with early trials, had a larger mean change of 0.01545. In a secondary test, beta reduced later gamma squared error beyond a reference with known state labels. The mean pair-relative error decrease was 5.90%, with a 95% person interval of 1.35% to 11.84%. Gains varied from 0.15% to 20.33% across people. Beta and gamma came from the same window. These reused clinical data do not establish a neural interaction, live BCI gain, or stimulation effect. The package preserves features, source pins, code, models, predictions, three figures, independent checks, and reproduction instructions. Fresh raw checks covered one trial per person. A separate analysis run reproduced all numerical results exactly.

stereoelectroencephalography · beta power · high-gamma power · brain-computer interfaces · state centering · power correlation