Fixed voice direction tests did not confirm shared enrichment after gain removal in six SEEG patients and 16 fMRI adults

We tested how a learned speech/song response direction relates to repeated differences among nine nonspeech human voice clips. Separate public studies supplied SEEG high-gamma amplitude and BOLD fMRI. The fixed transform removed a uniform raw offset and positive gain in each spatial pattern. The direction and scales used 156 other clips and a separate response estimate. Twenty fixed panels of 24 sites gave one score per person. Two patients and four MRI adults were marked for development. After two patient scores, we added a narrower enrichment claim before other outcomes. Reserved tests used six other patients and 16 other adults. The enrichment contrast compared covariance along the direction with total covariance divided by 23. Its person means and 95% Student limits were +0.01389 [-0.00004, +0.02781] and +0.02267 [+0.01640, +0.02895]. The narrower mean enrichment test failed its shared rule. Its two-claim Holm p was 0.0504245. The original enriched-minority test failed, with Holm p=0.0504245. An eight-adult MRI acquisition check shared clips and could share people with the primary MRI study. All negative people, controls and sound sensitivity remain. The test concerns transformed source-selected recordings and a learned direction. Axis noise and stable nuisance can affect it. It does not identify neural rank, distinct cells, voice identity or a cause. Code, source pins and terms, results, figures and repeat instructions are supplied.

human neuroscience · SEEG · fMRI · voice · speech · repeat covariance · response geometry · gain invariance · negative results