Encoding memory effects included shared and local components across sampled SEEG sites in 52 FR1 subjects

We tested whether the subsequent memory effect (SME) in word-encoding SEEG features was shared across implanted bipolar pairs or remained local to pairs after removing a shared component. We used archived 0–1.6 s features from 54 subjects; 52 met the inclusion rule, contributing 4,946 pairs. For tilt (z-scored HFA minus theta), the shared all-pair mean showed an SME (mean subject t = +1.18; 41/52 positive; Holm p = 0.0001). After removal of the leave-one-pair-out mean, shuffle-corrected local SME energy remained (E_res = 0.46; E_raw = 0.92; 44/52 positive; Holm p = 0.0001). The shared fraction was 0.50 (95% CI 0.13–0.73), so the prespecified rule did not decide whether tilt SME energy was mostly shared or local. Removal reduced the mean signed pair t from +0.43 to +0.03. Exploratory results showed a local HFA SME energy estimate (shared fraction −0.03, CI −0.32 to 0.18) and partly shared theta SME energy (0.42, CI 0.02–0.66). A serial-position control gave a similar tilt result. The shared component is defined over sampled implanted pairs; results concern one task, session, and time window.

SEEG · subsequent memory effect · free recall · word encoding · theta · high-frequency activity · shared signal · local signal