Hippocampal SEEG power did not distinguish later-recalled from forgotten words in a 14-patient free-recall sample

This study tested whether hippocampal SEEG power during word encoding differed between words later recalled and forgotten in delayed free recall. Competing predictions were a broad low-frequency decrease with a high-frequency activity (HFA) increase (Long, Burke & Kahana, 2014) and a slow-theta increase with a fast-theta decrease (Lega, Jacobs & Kahana, 2012). In OpenNeuro ds004789 session ses-0, 14 patients with 90 hippocampal bipolar pairs met the analysis criteria. The pre-specified method used 5-cycle Morlet power averaged over 0–1.6 s after word onset, per-pair Welch t statistics, subject means, exact sign-flip tests, and Holm correction over five bands. No band showed a significant subsequent memory effect: mean t values were −0.050 for slow theta, −0.249 for fast theta, −0.375 for alpha/beta, +0.157 for low gamma, and +0.088 for HFA; all Holm-adjusted p values were 1.00. The slow-minus-fast theta contrast was also not significant (mean +0.199, p = 0.191). Descriptive frequency-resolved values showed negative t values from 4–28 Hz and positive values from 33–44 Hz and 156–180 Hz, consistent in sign with a spectral tilt, but far from significant. This small, time-limited sample does not support the tested predictions; the analysis estimated that it could detect only group effects of about |mean t| > 0.5. Limitations include analysis of 14 of 123 eligible subjects, ses-0 only, atlas-based contact labels not verified on images, no interictal-discharge removal beyond a 1 mV rule, and no seizure-onset-zone exclusion.

hippocampus · SEEG · subsequent memory effect · free recall · theta · high-frequency activity · OpenNeuro · intracranial EEG