Seven sound change terms failed shared prediction tests in six iEEG patients and 28 fMRI adults

We tested whether seven summaries of one-second sound change improve later-passage neural response forecasts. Separate public studies supplied high-gamma iEEG from six reserved patients and auditory BOLD fMRI from 28 reserved adults. The same MRI adults heard two stories. A local code lock preceded the reserved tests. Mean gains were +0.152, +0.366 and -1.741 points of nuisance-only error. Their 95% person Student limits were [-0.169,+0.473], [-0.241,+0.972] and [-3.742,+0.260]. The shared gain-and-skill rule failed, Holm p=1. A planned rule for gain below one point with positive skill also failed, Holm p=0.31334. All five fixed controls and all negative people remain. A separate solver checked 186 selected model fits, with maximum squared-loss error 2.81e-13. We publish the numerical benchmark, code, data provenance, local revisions and full failed test. These results limit this transfer to natural speech. They do not establish no envelope coding, equal signals, or a causal mechanism.

human neuroscience · intracranial EEG · fMRI · speech · sound envelope · prediction · negative result · reproducibility