PAL1 cued recall shows serial-position changes in SEEG band power without a change in recall

We tested whether serial-position changes in encoding power seen during free recall also occur during paired-associate learning (PAL1) with cued recall. Using archived sample/fs-timed features from 18 pure-SEEG subjects in OpenNeuro ds005059 session 0, we analyzed 2,334 study trials across six word-pair positions. The pre-registered analysis found a negative HFA (70–150 Hz) slope of −0.0198 z per position (95% CI −0.0265 to −0.0126; 16/18 subjects negative; Holm p = 0.00015) and a positive theta (3–8 Hz) slope of +0.0416 (95% CI +0.0274 to +0.0552; 17/18 positive; Holm p = 0.00012). Both pre-registered criteria were met. Without position 1, the HFA slope was smaller and did not meet the 0.05 threshold (Holm p = 0.052), while theta remained positive (Holm p = 0.00011). Recall probability did not vary with study position (slope +0.003, 95% CI −0.011 to +0.018, p = 0.68), and recall as a covariate did not explain the power slopes. These results show that the power pattern also occurs in word-pair study for cued recall and, in this task, is not a sign of better or worse encoding. The analysis used all-channel means, session 0 only, and 18 subjects; it did not reject artifact or seizure-onset channels. Serial position is also time since list start and number of preceding pairs, so these factors cannot be separated. The analysis assumes events.tsv sample/fs is EDF time. Slope units are not directly comparable with FR1, which used 12 positions per list and 1.6 s epochs.

SEEG · paired-associate learning · cued recall · serial position · theta · high-frequency activity · encoding · intracranial EEG