No evidence that polarity-invariant SEEG responses adapt more than artifacts during alternating-polarity stimulation

In 9 patients, we compared the first and last seconds of 14 alternating-polarity SEEG stimulation trains (669 channels without a stimulated contact) from OpenNeuro ds006519 v1.0.0. The pre-specified primary contrast, log(S_last/S_first) minus log(D_last/D_first), was -0.051 (bootstrap 95% CI -0.136 to +0.037; 6 of 9 patients negative; exact sign-flip p = 0.31). S alone changed by -0.079 (p = 0.16), while D changed by -0.009 (p = 0.79). The plan-literal 43.2-Hz subset also gave no evidence of adaptation (7 patients, 12 trains; contrast -0.032; p = 0.61). A pre-specified result among channels selected for high S_first was not reliable because selection can cause regression to the mean. Post-hoc symmetric selections gave weaker, inconsistent evidence. S_first and D_first correlated across channels (mean Spearman rho +0.53), so S may contain residual non-linear artifact; the analysis cannot distinguish this from shared scaling with distance to stimulation. The study has 9 patients and low power for small effects; it compares only the first and last second of a 5-second train, uses bipolar montage, and does not model amplifier saturation. The full report retains methods, deviations, results, limitations, source provenance, and reproduction instructions.

stereoelectroencephalography · SEEG · electrical stimulation · alternating polarity · pulse-locked response · artifact · neural adaptation · intracranial EEG · OpenNeuro