Two averaging orders gave a mean hippocampal theta effect difference of −0.232 dB in 12 SEEG patients

This study compared two trial averaging orders in public hippocampal SEEG spectra from OpenNeuro ds004752, version 1.0.1. It used 12 patients, 24 sessions, 44 matched pairs, and 1,085 trials. Both methods used the same fixed theta contrast and late-minus-fixation windows. The mean effect was 0.374 dB from trial log spectra and 0.142 dB from log mean spectra. The paired difference, arithmetic minus geometric, was −0.232 dB. Its 95% t interval was −0.566 to 0.102 dB. The exact patient sign flip p value was 0.114746. This did not show a clear mean signed difference or establish equivalence. A known Jensen-gap identity explained the signed frequency contributions. Theta and flank contributions partly canceled. A reusable NumPy tool and a fixed-shape cancellation proof permit inspection of this choice. All ten numerical outputs matched in a separate offline run. Fresh public checks matched 185 metadata files, three task source files, and 144 raw phase records. The fresh records reproduced 21,648 saved spectral values with zero numerical difference. The planned full replay was reduced because downloads were slow. The package records this validation change and preserves the full retrieval recipe. It also preserves the exact source archive, plans, code, proofs, results, figures, and provenance. These results apply to this clinical sample and these fixed analysis rules.

hippocampus · stereoelectroencephalography · theta power · spectral analysis · trial averaging · Jensen gap · working memory