SEEG high-gamma number tuning was not similar across formats in a three-patient pilot

This pilot asked whether site-level SEEG high-gamma number tuning was similar for arabic digits, dot arrays, and spoken number words. We analyzed stimulus-locked data from three patients in the OpenNeuro ds005415 OHSU numbers task, using 476 bipolar pairs. The pre-specified Bonferroni rule identified few responsive pairs. The primary arabic-dots comparison could run only for sub-78 (6 pairs): r = 0.73, permutation p = 0.088. The arabic-spoken comparison in sub-78 gave r = -0.92 (p = 0.989). No primary cross-format result was significant. Exploratory tests using all 476 pairs found no cross-format similarity (group Stouffer p = 0.16–0.72). A nominal written-word number effect in sub-73 (1 pair) did not establish number coding and may reflect visual features; it was one of many uncorrected tests. These data do not show a shared number code, but low power, three patients, about 40 trials per format, and few responsive pairs mean they do not exclude one. The analyzed EDF files are identified by source URLs and SHA-256 hashes; correspondence to a versioned dataset snapshot was not established. The plan, code, provenance, per-pair values, and statistics are included.

SEEG · high-gamma · number tuning · numerical cognition · neuroscience · permutation test · OpenNeuro · pilot study