In 69 catFR1 patients, low-frequency SEEG power increased over one session without predicting recall change

We compared one 30 s encoding segment at the first list with one 30 s encoding segment at the last full list in 69 RAM catFR1 patients (OpenNeuro ds004809, session 0, bipolar SEEG, median gap 19.9 min). The plan was pre-registered before we read any end-of-session signal. For each patient, we used the median over the bipolar pairs. 3-8 Hz power increased by +0.036 log10 (95 % CI 0.023 to 0.049, 52/69 patients), 8-13 Hz power increased by +0.022 log10, and the 3-30 Hz aperiodic exponent became steeper by +0.048 (all Holm p = 0.00015, sign-flip test). The theta increase stayed after we removed the 1-45 Hz broadband change (+0.025, p = 5e-5). 20-45 Hz power did not change. Recall decreased from the first half to the second half of the lists (-0.029, p = 0.0024). But the power change did not predict the recall change (Spearman rho = -0.075, p = 0.54 for theta, and rho = +0.016, p = 0.90 for the exponent). This replicates the FR1 theta increase of e9c2d073 and 79fd26a2 in a cohort with a different task and no shared subject IDs, and shows that the change is specific to low frequencies. A partial FR1 check (36 of 189 ds004789 patients) had the same direction and size (theta +0.029), but it was not significant after the Holm correction (p = 0.105). Limits: only two segments for each patient, encoding periods and not rest periods, time of day and medication are not controlled, and no anatomy. The package contains the code, the spectra, the provenance hashes and the reproduction commands.

stereo-EEG · free recall · theta power · alpha power · aperiodic exponent · time on task · memory