Hippocampal high gamma was slightly higher for in-list probes in 14 patients, with no pre-probe effect

This analysis tested whether hippocampal bipolar SEEG power in the first second after a verbal Sternberg probe differed between in-list (IN) and new (OUT) probes. It used OpenNeuro ds004752 v1.0.1, sessions 1 and 2 from 14 of 15 subjects (sub-13 had no hippocampal or amygdala contacts): 28 sessions and 1,161 correct, non-artifact trials. A plan was written before probe-period data were fetched. For each pair, 1-s Hann log band power was z-scored within session and set size; IN-versus-OUT AUCs were averaged by subject and tested with exact two-sided sign flips, with Holm correction across five bands. In the hippocampus (14 subjects, 96 pairs), high gamma (70–150 Hz) was slightly higher for IN probes (mean AUC 0.527; 12/14 subjects above 0.5; p = 0.0096, Holm p = 0.048). Theta (AUC 0.510, p = 0.30), alpha (0.483, p = 0.098), beta (0.492, p = 0.34), and gamma (30–70 Hz; 0.505, p = 0.68) did not differ significantly. In the pre-probe second, high gamma AUC was 0.508 (9/14, p = 0.47), with no band p < 0.09. With log response time as a covariate, the high-gamma IN effect remained positive (mean t = 0.34; 11/14; p = 0.031) but was not significant after Holm correction (0.155). Probe-period alpha and beta power decreased from pre-probe values (0/14 and 1/14 subjects positive; Holm p = 0.0006 and 0.0024); high gamma increased by 0.017 log10 (10/14, p = 0.073). In the secondary amygdala analysis (10 subjects, 39 pairs), high-gamma AUC was 0.542 (8/10, p = 0.020, Holm p = 0.098); pre-probe AUC was 0.489 (p = 0.33), and RT-adjusted p was 0.014 (Holm 0.068). Behaviour did not differ significantly: median RT was 0.10 s shorter for IN (9/14, p = 0.18), and accuracy was 2.8 percentage points lower (p = 0.34). The effect was small (AUC about 0.53) and the primary corrected p value was just below 0.05. The analysis used one 1-s window without time-frequency resolution. Button press differences for IN and OUT could contribute motor or eye-movement signals; response usually followed the analysed second (median RT 1.1–2.0 s). ROI labels were atlas labels only, interictal discharges were not removed beyond the dataset artifact flag and an amplitude rule, and only 2 of up to 8 sessions per subject were used.

verbal working memory · Sternberg task · hippocampus · amygdala · SEEG · intracranial EEG · high gamma · probe match · OpenNeuro