For 165 sounds, mean scores for own sound maps exceeded scores for other people in six SEEG patients and two fMRI cohorts

We tested whether sound relationship maps agree more within a person than across people. Separate public studies supplied SEEG high-gamma activity and BOLD fMRI for 165 sounds. We removed broad sound-class means and each sound spatial mean, then normalized sound response length. Two blocks supplied a gallery; a third supplied its target. Development used two patients and four adults. Fixed tests used six other patients, sixteen other fMRI adults, and eight adults from another fMRI study. Mean own-minus-other map similarities were 0.07479, 0.05350 and 0.03060. Descriptive whole-person 95% bootstrap ranges were 0.03623-0.11402, 0.04207-0.06597 and 0.01068-0.04839. Both shared claims passed a two-claim Holm rule, adjusted p=0.002778. Gallery top matches were only 4/6, 2/16 and 3/8. Site-count, disjoint-site and nonvocal checks retained positive means. A short electrical baseline range crossed zero. One fine-category MRI control was undefined and remains recorded. An exact common-direction example with unequal quality gave positive mean advantage and p=1/720. An identical latent code with different site mixtures identified every constructed record. These are interpretation limits, not fitted human explanations. The supported shared result concerns measured record association. It does not establish individual neural codes, a common neural cause or useful identity accuracy. Clinical sampling, separate cohorts, source selection on repeats, stable anatomy and noise, motion and vascular filters limit scope. The package preserves code, source pins and terms, plans, all outcomes, figures, audits and repeat steps. Original neural arrays stay external.

human neuroscience · SEEG · fMRI · natural sounds · representational geometry · record identity · measurement confounds · reproducible research