Under patient independence and null sign symmetry, measured 20 to 40 Hz spectrum adjustment reduced late-maintenance theta–gamma power correlation beyond shuffled terms in 12 patients

This study tested whether measured spectrum terms change hippocampal theta and gamma power correlation. It used public OpenNeuro ds004752 version 1.0.1. The sample had 12 patients, 44 bipolar pairs, two sessions per patient, and 1,085 selected trials. The target bands were 4 to 7 Hz and 60 to 95 Hz during late verbal memory maintenance. All models adjusted for memory load and linear trial order. The full model also adjusted for measured 20 to 40 Hz spectrum level and slope. The comparison used 999 joint within-load shuffles of these terms per session and pair. The patient mean excess reduction was 0.194 Fisher z units. The 95% patient t interval was 0.054 to 0.333. The two-sided exact sign-flip p value was 0.00390625, under patient independence and null sign symmetry. Nine patients had a positive difference and three had a negative difference. The result shows sensitivity to the measured spectrum adjustment. It does not prove rhythm interaction, a shared neural cause, or a memory-specific effect. This is a new analysis of the same cohort used in the cited parent publication. The package preserves code, compact inputs, source pins, trial-level results, and offline instructions. Fresh checks matched all 2,334 source windows and 185 metadata files. Separate least squares fits checked all 87,912 primary shuffled fits. An offline replay gave identical analysis file hashes.

intracranial EEG · hippocampus · theta power · gamma power · spectrum adjustment · verbal working memory · power correlation · SEEG