For 165 clips, person t tests kept mean error gains below 1% in maps between one iEEG panel and 20 fMRI adults

We tested whether electrical sound profiles after offset add BOLD prediction beyond during-sound profiles, categories and supplied acoustics. We also tested whether BOLD adds prediction of the late electrical profiles. Public studies supplied one fixed panel of 32 distinct selected iEEG profiles and separate fMRI cohorts. Eight BOLD people supplied development. The unchanged test used 20 other BOLD adults and the same 165 clips. Adding late electrical features increased mean BOLD test error by 3.895%, with conditional 95% person interval 3.441 to 4.349%. Adding BOLD increased late electrical error by 0.841% (0.660 to 1.023%). The shared positive test failed. A development-led test placed both mean gains below 1% of baseline error under person t assumptions; two-claim Holm p was 1.07e-14. A sound-resampling diagnostic did not extend that bound to new clips. Missing electrical patient labels prevent electrical person inference. All intervals condition on the one panel and fixed sounds. Clinical coverage, category selection, source filters, unequal measurement, noise and the fixed ridge model limit scope. A synthetic same-code model gave positive gains without a distinct late code. No shared positive neural effect, absent offset activity or causality is claimed. The package preserves code, source hashes and terms, plans, all results, figures and repeat instructions.

auditory cortex · intracranial EEG · fMRI · sound offset · cross-modal prediction · negative result · ridge regression