Hippocampal-scalp 4-12 Hz phase locking did not increase with verbal working-memory load in 12 SEEG patients (OpenNeuro ds004752, session 1; p = 0.95)

Question: does phase locking between hippocampal SEEG and scalp EEG during working-memory maintenance increase with memory load? In 12 subjects from session 1 of OpenNeuro ds004752, we analyzed 83 hippocampal bipolar pairs, 17 scalp channels, and 484 correct, artifact-free trials. Hippocampal EEG was read for trial seconds 0-1 and 3-6 using HTTP byte ranges; SHA-256 hashes are recorded. A prewritten plan specified FFT band-pass filtering and analytic phase, within-trial phase-locking value (PLV), per-subject Spearman correlations with set size, exact sign-flip tests, Holm correction across three bands, and bootstrap confidence intervals. Response time increased with load (mean z +0.31; 11/12 subjects; p = 0.0015). Maintenance PLV exceeded trial-shuffled PLV (4-12 Hz: 12/12, +0.015, p = 0.0005; 4-8 Hz: 12/12, p = 0.0005; 8-12 Hz: 11/12, p = 0.001). The primary 4-12 Hz load effect was not significant (mean z +0.004, 95% CI [-0.108, 0.132], 4/12 positive, p = 0.95; Holm p = 1.0). The 4-8 Hz effect was also not significant (mean z -0.033, p = 0.56); the 8-12 Hz effect was not significant (mean z +0.034, 7/12 positive, p = 0.46; Holm p = 1.0). Across-trial PLV differences between set sizes 8 and 4 were not significant; the alpha-band result was p = 0.057 uncorrected. Fixation theta PLV decreased with set size (2/12 positive, p = 0.034, Holm p = 0.10), although load was not known during fixation. The sample included 12 subjects and one session each; effects with |z| below about 0.1 are not excluded. Atlas labels, the recorded scalp reference, possible volume conduction and common-reference effects, and PLV sensitivity to signal-to-noise ratio and window length limit interpretation. The analysis did not use wPLI. The data were fetched from a live OpenNeuro mirror, and the exact dataset release is uncertain.

hippocampus · scalp EEG · SEEG · phase locking value · verbal working memory · Sternberg task · memory load · OpenNeuro · trial shuffling