At 60 sounds, allocation transfer raised mean sampling risk in fixed tables from 18 iEEG patients and 16 fMRI adults

We tested whether a sound allocation learned from one neural measurement lowers class mean map error in the other measurement. Separate public studies supplied signed iEEG field means and BOLD fMRI. Four development people per modality set the allocations. A local code lock preceded new scores in 18 patients and 16 MRI adults. The two modalities used different sound clips. The risk is exact expected mean squared error under random sampling without replacement from each fixed released library. At 60 distinct sounds, the mean person risk increased by 4.540% in iEEG and 2.394% in fMRI relative to equal class sampling. Pointwise 95% person Student limits for the increases were [2.464,6.616]% and [1.897,2.891]%. Sixteen of 18 patients and all 16 MRI adults had higher risk. At 30 sounds, the electrical allocation increased risk in all 18 patients; reverse allocation was exactly equal sampling. Both planned benefit claims failed after Holm correction, p=1. Algebraic variance thresholds explain these fixed-table signs. Joint source-person and sound resampling gives a 60-sound change range of -9.385% to -0.420% in iEEG and -4.401% to +0.870% in fMRI. Thus the latter crosses zero. The additional MRI acquisition shares the MRI sound library. Cohorts, clips, tasks, tissue sampling, source screens and measurement noise differ. No positive cross-modal neural generalization, causal modality effect or prospective time saving was established. The package gives an exact-risk benchmark, code, permitted inputs, all outcomes, source pins and terms, figures and repeat instructions.

human neuroscience · intracranial EEG · fMRI · auditory cortex · sample allocation · finite population · negative result · reproducibility