A label-free word-evoked response predicts memory effects after serial-position control

In 31 independent pure-SEEG subjects performing free recall (FR1, session 0), bipolar pairs with a stronger label-free, word-evoked response had a larger subsequent memory effect (SME). Across pairs within each subject, the evoked index correlated with raw tilt SME at mean Spearman rho +0.179 (95% CI 0.122 to 0.237; 28/31 positive; Holm p=0.00015). After controlling SME for 12 serial positions, the association remained (rho +0.134, 95% CI 0.072 to 0.199; 24/31 positive; Holm p=0.0003), about 75% of its raw size. The control reduced the tilt association (Fisher z difference 0.048, 95% CI 0.018 to 0.083; p=0.0015). Higher-index pairs also had a steeper HFA decrease over list positions (rho -0.093, p=0.002), consistent with primacy accounting for part of the raw link. Findings replicate an earlier internal study in independent subjects, while retaining limitations: session 0 and all bipolar pairs were analyzed, the position control does not remove other list effects, and the evoked index uses recalled and unrecalled words, so recall-related ERP differences may contribute. Epochs used events.tsv sample/fs as EDF time; this timing assumption and source-file uncertainty are documented. The original plan, analysis code, results, per-subject and per-pair tables, and provenance records are included.

intracranial EEG · SEEG · free recall · subsequent memory effect · evoked response · serial position · replication · memory encoding