A diversity rule for eight sites failed a shared sound matching gain test in six SEEG patients and 24 fMRI adults

We tested whether eight sites with complementary sound response profiles give better identity matching than eight sites with the largest individual profiles. Separate public studies supplied SEEG high-gamma responses and fMRI BOLD estimates. Calibration used 56 sounds. Matching used 109 other sounds within three broad groups. Two patients and four adults supplied development. A local code lock preceded tests in six other patients, 16 B2021 adults and eight NH2015 adults. Mean AUC gains were 0.002040, 0.018398 and 0.026257. Their 95% person Student intervals were [-0.002651,0.006731], [0.009681,0.027116] and [0.011867,0.040648]. The shared positive-gain test failed, Holm p=0.375. A planned shared test for means within plus or minus 0.02 also failed, Holm p=0.830972. Joint person and test-sound ranges kept this pattern. A later check used 24 candidates in every person. Its MRI mean intervals crossed zero; its SEEG mean was negative. These checks do not replace the primary result. An unfitted one-source model gave a 0.1020 mean matching gain from correlated measurement error. Thus selection gains need not imply several neural signal sources. Independent calculations and exact reconstruction of 10,449,945 source-derived cache values checked the work. The contribution is a repeated-sound benchmark, a scoped negative cross-modal test and an interpretation limit. Separate cohorts, sparse clinical sampling, unequal sensing volumes, earlier full-panel selection and different filters limit scope. Code, source pins and terms, plans, all results, figures and repeat steps are supplied.

SEEG · fMRI · auditory cortex · site selection · sound matching · cross-modal test · negative result