Person t tests placed mean gains from a fixed sound change model below one percentage point in six iEEG patients and 22 fMRI adults

We tested whether change in normalized sound spectra adds prediction beyond acoustic and word controls. Separate public studies supplied three iEEG development patients, six reserved patients and 22 fMRI listeners. Development selected a 0.3 s history. The shared positive gain test failed. Before reserved results, we added a test of mean gain below one percentage point. Holm correction covered both shared claims. Mean gains were -0.0217 and -0.2074 percentage points. Their 95% person Student intervals were [-0.0376 to -0.0058] and [-0.5932 to +0.1785]. The fixed mean limit test passed, adjusted p=1.90489e-06. Person bootstrap ranges and common time block ranges also had upper endpoints below the limit. All six electrical scores were negative. Five MRI scores were positive; three exceeded one point. The limit applies to this fitted residual reader under independent person scores and the Student model. The BOLD predictor had large values outside its training range. Exploratory checks capped these values and used longer time blocks. All results remain. The findings do not establish absent sound change coding or a shared neural cause. Clinical sampling, two fixed stories, short MRI records, source processing and missing nuisance traces limit scope. The package gives permitted inputs, code, plans, figures, source checks and repeat steps.

auditory encoding · spectral change · intracranial EEG · BOLD fMRI · cross-modal tests · prediction limits · public data