Tests of two sound groups did not confirm order crossings in six SEEG patients and 24 fMRI adults

We tested sound order crossings within two fixed source sound groups. Separate public studies supplied SEEG high gamma responses and BOLD fMRI. One repeat selected four required contrasts at two sites; two other repeats tested their minimum. If validation errors have zero conditional means, a monotone scalar response table has a nonpositive expected score. Half the expected score also bounds the best uniform fit error from below. Development used two patients and four MRI adults. A local code lock preceded tests in six other patients, 16 other B2021 adults and eight NH2015 adults. Nonvocal mean scores were -0.4884, -0.2719 and -0.4085, with 95% person intervals [-0.7889,-0.1879], [-0.4113,-0.1325] and [-0.6414,-0.1756]. Positive counts were 0/6, 2/16 and 0/8. Speech-and-song scores were also negative in mean. Both planned shared positive claims failed, with Holm p=1. All positive and negative outcomes remain. A Gaussian model used four contrasts with equal true margins and independent errors of equal variance. Its expected minimum was the true margin minus 1.0294 times the noise standard deviation. Thus, true crossings can have negative scores. All observed means exceeded 199 sound label mismatch references. A later scalar example passed that reference despite a zero crossing bound. Using that reference for scalar rejection gave false rejections in both training outcomes. Constructed nonmonotone scalars and stable artifacts can also give positive scores. The assay does not count neural sources or prove causality. Earlier full-panel site screens, sparse clinical sampling, separate cohorts, source filters, shared schedules and only three repeats limit inference. No positive shared neural finding is established. The contribution is a conditional method bound and a failed human benchmark. The package supplies code, provenance, statistics, audits, figures and repeat steps.

human auditory cortex · SEEG · fMRI · sound order · monotone response · independent repeats · negative results · model error bound