Recall was associated with smaller theta and HFA spread, but the HFA effect did not persist after high-amplitude word removal

In an archived sample of 30 patients with stereo-electroencephalography (SEEG) recordings during free recall (FR1), later-recalled words had smaller word-locked band-power spread across bipolar pairs after controlling for mean power and serial position. The planned group sign-flip tests found a mean theta statistic of -0.76 (95% CI -1.10 to -0.43; Holm-adjusted p = 0.0002) and a mean high-frequency activity (HFA, 70–150 Hz) statistic of -0.38 (95% CI -0.72 to -0.03; Holm-adjusted p = 0.038). Exploratory alpha, beta, and low-gamma statistics were also negative. In a post-hoc check that removed the 10% of words with the largest max-over-pairs robust raw-RMS z score, the theta result remained (mean t = -0.52, p = 0.0035), while the HFA result did not (mean t = -0.20, p = 0.27). This weak spike and artifact proxy therefore leaves the HFA finding uncertain. The study used one archived sample and had no independent replication set; spread depends on pair count. The analyzed data are archived features associated with OpenNeuro ds004789, session 0; the exact dataset release was not recorded.

stereo-electroencephalography · free recall · word encoding · band power · theta · high-frequency activity · bipolar pairs · memory