In one SEEG patient, sentence repetition did not predict responses after trial-position control; responses declined over the session

In patient W2, who heard 10 TIMIT sentences 10 times each, 19 of 98 bipolar SEEG pairs had a positive 70–150 Hz high-frequency activity (HFA) response. After controlling for sentence identity and trial index, repetition number did not predict the response (Freedman–Lane permutation, 10,000 permutations; onset 0.05–0.40 s: β = +0.106 z/repetition, 95% CI [−0.028, +0.240], Holm p = 0.243; sustained 0.40–1.30 s: β = +0.032, 95% CI [−0.124, +0.188], Holm p = 0.681). Both estimates were positive. The repetition test has low sensitivity because repetition number and trial index were strongly correlated (r = 0.967); the sustained-window interval does not exclude moderate suppression. Recency, which was nearly independent of trial index (r = −0.03), also did not predict responses (Holm p = 0.564 and 0.561). The mean response declined over the session, with an exploratory trial-index effect in the sustained window (β = −0.0064 z/trial, p = 0.0001); prestimulus HFA did not change (p = 0.49). A second patient, W3, heard 126 unique sentences; responses did not decline with trial order. Each task was recorded in one patient, and W2 and W3 differed in electrodes, so there is no population claim and their comparison is suggestive only. The findings are more consistent with a slow session-level decrease than sentence-specific repetition suppression, but the data cannot distinguish adaptation to the small stimulus set from fatigue or another state change.

SEEG · high-frequency activity · sentence repetition · recency · trial position · sentence processing · intracranial EEG · OpenNeuro ds004993