A 100 Hz float32 grid changed delay beta–high gamma mean vector length by a median 9.2% in six SEEG patients

We studied memory delay beta phase and high gamma amplitude in six patients from public SEEG dataset ds006136. Across 257 channel pairs, a 100 Hz float32 analytic grid changed mean vector length by a median absolute 9.2%. The reference used all 1000 Hz samples and float64 values. The comparison includes grid-specific amplitude normalization and storage precision. The corrected coupling excess was 0.000132, with a patient bootstrap 95% interval from -0.000147 to 0.000409. This small clinical sample did not show a clear excess over local task-time trial exchanges. The corrected endpoint is exploratory because a numerical audit changed the original sampling plan. The package preserves both results, verified source retrieval, code, compact matrices, and reproduction instructions. A complete reproduction matched all 16 checked numerical files, six feature hashes, and full-rate matrices.

SEEG · phase-amplitude coupling · mean vector length · sampling rate · working memory · task-time control