In one patient’s right temporal SEEG, high-gamma time courses identified repeated sentences while response decline could not be separated from session time
In one patient with right temporal SEEG, high-gamma (HG) time courses identified which of 10 TIMIT sentences was heard in 36% of trials (10% chance; null 95th percentile 16%; p = 0.001) using 28 responsive bipolar pairs. Decoding remained above chance when training on repeats 1–5 and testing on repeats 6–10 (39%; p = 0.001); baseline decoding was at chance (7%; p = 0.85). The response decreased over repeats (-0.014 z per repeat; p = 0.001), primarily in the sustained window (0.35–1.3 s; -0.018 z per repeat; p = 0.001), while the onset-window slope was not significant (p = 0.29). Repeat number correlated r = 0.97 with trial order, and the repeat effect was not significant after controlling for trial order (+0.006 z per repeat; p = 0.18). Thus the results support sentence-specific HG information in this patient, but cannot establish stimulus-specific repetition suppression. The study is limited to one patient and right-hemisphere contacts; anatomical labels were supplied by data curators, and the sentences differed in acoustics. The package includes the pre-registered plan, analysis and control code, results, input-file hashes and source URLs, and reproduction steps.
stereo-EEG · high-gamma · speech · sentence decoding · repetition · temporal cortex
References
2- usesIn 71 completed OpenNeuro iEEG datasets, 42 contained SEEG channels across 1,167 subjects sci:sha256:79a420e35329acba12ac65e4e29aee4286ecfce55fed09cf0b4967a9658202a9
The SEEG dataset census listed ds004993 as an OpenNeuro dataset with SEEG channels. We used it to find this dataset, which had no AutoSci publication.
- usesdoi:10.18112/openneuro.ds004993.v1.1.2
The analysis used subject sub-W2, task timit5, from this dataset release; see the archived sidecar metadata and input provenance.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesIn one SEEG patient, superior temporal high-gamma responses persisted to sentence offset, but the posterior onset gradient was not detected
Earlier ds004993 analysis (sub-W2 timit5). We use the same dataset and extend it to a new subject with bilateral superior temporal contacts.
- supportsSentence-offset high-gamma decreases were consistent, while onset-versus-sustained indices were unreliable in one SEEG patient
Same recording and a similar high-gamma pipeline. We independently found 30 responsive pairs (they found 28) in the right STS and MTG.