No clear memory-load effect in late-maintenance SEEG power across hippocampal and other regions

We tested whether SEEG band power during seconds 4–6 of a verbal Sternberg working-memory trial varied with memory load (4, 6, or 8 letters), and whether any effect was specific to the hippocampus. In 12 patients from OpenNeuro ds004752, we analyzed the first 20 artifact-free trials in session 1 (240 trials; 508 bipolar pairs, including 82 hippocampal pairs). Across the prespecified primary hippocampal tests, theta power showed mean Spearman rho +0.004 (95% CI −0.062 to +0.071; p=0.91), and high-gamma power showed rho −0.042 (95% CI −0.121 to +0.027; p=0.32; Holm-adjusted p=0.64). No band passed correction in hippocampal pairs or atlas regions; non-hippocampal effects were small, and no hippocampus-specific effect was detected. The confidence intervals include the weak high-gamma decrease reported by earlier analyses of all trials, so this study does not reject that effect. The result is limited by 20 trials per participant, a 2-s window, and 12 of 15 participants; statistical power is low. The acquisition used a live dataset URL without a recorded release, and the original fetched-byte hashes are preserved.

verbal working memory · SEEG · hippocampus · memory load · band power · maintenance period