Odd and even list estimates of the SEEG word-encoding memory effect correlate in 27 FR1 patients

We tested whether the subsequent memory effect (SME) in word-locked intracranial EEG is reproducible across odd and even free-recall lists within a patient. The analysis used archived 0–1.6 s band-power features from 32 pure-SEEG participants in session 0 of FR1 in OpenNeuro ds004789; 27 met pre-registered criteria of at least four lists and at least 10 recalled and 10 not-recalled words in each half. After removing list and serial-position means, we calculated Cohen's d (recalled minus not recalled) per bipolar pair and averaged across pairs. Odd- and even-list estimates correlated across patients for both high-frequency activity (70–150 Hz; Spearman rho 0.47, 95% bootstrap CI 0.15–0.70, Holm-adjusted p 0.033) and theta (3–8 Hz; rho 0.42, CI 0.00–0.72, adjusted p 0.045). Only 2 of 27 patients had a significant high-frequency SME in individual permutation tests (both positive; adjusted binomial p 0.39); 5 had a significant theta decrease (adjusted p 0.040). Between-patient heterogeneity was significant for theta (I² 0.49, adjusted p 0.014), but not high-frequency activity (I² 0.30, adjusted p 0.15). The mean high-frequency effect was near zero. The estimates are stable across list halves in this sample, but patient and implant-location effects cannot be separated. The analysis includes one session, averages over all electrode pairs, and does not test region-specific effects.

subsequent memory effect · intracranial EEG · stereo-EEG · free recall · split-half reliability · theta · high-frequency activity · individual differences