Using each template once improved sound matching in six SEEG patients and 24 fMRI adults

We tested how a known sound inventory changes neural identity matching. Separate public studies supplied SEEG high-gamma responses and two fMRI cohorts. Each test used the same 16-site samples and held-out response distances. One reader chose the nearest template for each response. The other used every template once to reduce total distance. 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. Mean accuracy gains were 2.22, 2.37 and 1.44 percentage points. The 95% person intervals were [1.22,3.23], [1.89,2.84] and [0.99,1.89]. All 30 reserved people improved. The fixed three-cohort rule passed, p=0.015625. Copying responses to remove 19.09% of true identities reduced mean gains to 0.71, 0.88 and 0.69 points. This was a model stress, not a new repeat experiment. A sharp counting bound limits complete assignment accuracy to u/n when n responses have u distinct true identities. Perfect measurements can then favor the independent reader. Thus a known inventory can improve this offline task without new neural observations. The result does not establish more neural information, online benefit, a common neural mechanism or a cause. Source selection, fixed sounds, separate cohorts, unequal measurement scales and repeat dependence limit scope. Code, source pins and terms, all outcomes, figures, source checks and a second solver audit are supplied. Original neural arrays stay external.

human neuroscience · SEEG · fMRI · auditory cortex · sound identification · assignment · decoding · stimulus inventory · reproducibility