In OpenNeuro ds003078 snapshot 1.0.0, 72 SEEG files were byte-identical; a one-run pilot found no reward-feedback HFA difference

An integrity audit of OpenNeuro ds003078 snapshot 1.0.0 found that all 72 SEEG data files had the same size and S3 ETag. The MD5 of one complete file matched that ETag, and SHA-256 hashes from 96 byte ranges across 24 files agreed. Fourteen parsed .set files had the same 662-event table; their EEGLAB history points to one source recording. The 12 anatomical T1w files had different ETags. A pre-planned pilot on the single iEEG recording (126 of 130 channels after QC; 101 reward and 47 no-reward trials) found no reward-related difference in 70–150 Hz power 0.05–0.75 s after feedback: no channel passed max-stat correction with either the file reference or common average reference. A stimulus-locked positive control passed in three channels with common average reference. The dataset does not identify channel names or positions, and the feedback-code meanings were inferred from behavior. This audit concerns the files served by OpenNeuro on 2026-10-11, not the original recordings. The pilot contains one run from one patient and cannot support population or brain-region conclusions; its null result does not show that the effect is absent in general. The byte-identity conclusion depends on the single-part S3 ETag being an MD5, checked against one complete file and 96 byte ranges. The package includes the audit and analysis code, plan, and result tables; the 155 MB data file is not included.

intracranial EEG · SEEG · OpenNeuro · dataset integrity · high-frequency activity · reward feedback · permutation test