Two response baselines gave opposite mean SEEG associations with pre-pulse voltage change in five patient sessions

We tested whether response baselines change associations with pre-pulse voltage change in public SEEG recordings. We used OpenNeuro ds004457, version 1.0.2. The sample contained five reused patient sessions, 183 stimulation sites, 15 recording pairs, and 1,904 distinct pulses. Two response definitions gave equal-patient mean correlations of +0.626 and −0.179 with the same voltage-change predictor. All five patient means changed from positive to negative. Opposite signs also occurred in a matched before-pulse control. The sign change did not establish a causal effect of stimulation. An exact slope relation explains the difference. Matching forecast changes gave identical residuals. Four-pulse fits gave negative mean error gains against constant forecasts under both tested error weights. Two patients had positive gains for one target with errors in voltage units. The extension was exploratory and followed an initial analysis. We verified 110 source objects and matched ten data file hashes with a fresh raw extraction. A second run matched 27 result files byte for byte. The package includes selected data, source hashes, retrieval code, analysis code, four figures, results, and reproduction instructions. These checks support inspection of state-based stimulation measurements. They do not show improved BCI control.

stereoelectroencephalography · SEEG · brain stimulation · response baseline · voltage association · brain-computer interfaces · forecast error