In 12 ds005415 SEEG subjects, 70-150 Hz responses (0.1-0.6 s) of bipolar pairs did not show monotonic number-magnitude coding (1-9) with stimulus format held constant

We tested whether 70-150 Hz SEEG responses of bipolar pairs varied monotonically with number magnitude (1-9) when stimulus format was held constant. The analysis included 12 OpenNeuro ds005415 subjects (sub-72, 73, 78, 79, 82, 83, 84, 85, 86, 90, 91, 92); sub-80 was excluded because its EDF was truncated. For each trial and pair, the feature was the log ratio of power from 0.1-0.6 s after stimulus onset to the -0.3-0 s baseline. We removed each format mean, calculated Spearman correlation with number, and counted pairs with two-sided p < 0.05 across subjects. The preregistered permutation test (1000 within-format permutations) found 83 of 2091 pairs, versus a null mean of 75.1 (95th percentile 92.0; p = 0.232). No subject-level effect survived correction for 12 subjects (lowest p = 0.046, sub-85). Symbolic visual trials gave 111 pairs (p = 0.318; Holm p = 0.635); non-symbolic trials gave 110 (p = 0.351; Holm p = 0.635). With this montage coverage, time window, and monotonic model, tuned pairs were not more frequent than chance. Non-monotonic tuning, other frequency bands or windows, and regions without contacts were not tested. About 15-40 trials per format leave weak effects possible. The report also documents an event alignment issue in sub-84 to 92: photodiode samples refer to uncropped recordings while EDF and events files are cropped, and behavioral rows do not always match event rows one-to-one.

SEEG · numerical cognition · number magnitude · high gamma · intracranial electrophysiology · OpenNeuro · permutation test · stimulus format