In 13 SEEG patients, gamma bursts did not increase during encoding or late delay, while beta bursts decreased in middle frontal gyrus

We tested working-memory burst-model predictions in human stereoelectroencephalography (SEEG) from the middle frontal gyrus (MFG) and middle temporal gyrus (MTG). The analysis used OpenNeuro ds006136 v1.0.1: 13 patients, 143 channels, and curator-processed artifact-free trials from a load-3 object task. We filtered each trial separately in beta (15–30 Hz) and gamma (65–100 Hz), then measured occupancy of bursts above each channel’s mean plus 2 SD for at least three cycles. Subject-level exact sign-flip tests assessed five preregistered contrasts, with Holm correction. Gamma occupancy did not increase from fixation to encoding (difference −0.0005, 95% CI −0.0013 to +0.0002; p = 0.1902; Holm p = 0.402) or from early to late delay (difference −0.0003, 95% CI −0.0009 to +0.0002; p = 0.3276; Holm p = 0.402). Beta occupancy decreased during encoding (difference −0.0020, 95% CI −0.0036 to −0.0006; p = 0.0132; Holm p = 0.066), in the predicted direction but not significantly after correction. This exploratory decrease occurred in MFG (difference −0.0046; p = 0.0010) and not MTG (difference +0.0005; p = 0.8152). Encoding gamma occupancy was lower, not higher, on full-recall trials (difference −0.0017; p = 0.0640; Holm p = 0.256), and late-delay beta increased rather than decreased (difference +0.0008; p = 0.1340; Holm p = 0.402). No preregistered test passed Holm correction. The conclusions apply only to the studied regions, bands, windows, and threshold rule. The sample was small (n = 13), occupancy was about 0.4–0.7% of samples, gamma was near the noise floor in the MU cohort, and region labels were not verified. Full recall was confounded with trial difficulty and class trial counts.

working memory · SEEG · burst occupancy · beta oscillations · gamma oscillations · middle frontal gyrus · middle temporal gyrus