A 0.5 s sound gap gave a mean gain in six iEEG patients but failed a shared prediction test with twenty fMRI adults

We tested whether variation of band sound level inside 0.5 s improves auditory response prediction beyond current level, functions of mean level and word timing. Separate public studies supplied iEEG high-gamma amplitude and BOLD fMRI. Three patients and five listeners supplied development. A local fixed test reserved six other patients and twenty other listeners. The added learner first removed the baseline predictor span and kept the baseline prediction fixed. Mean added test R2 was 0.0024996 in iEEG and -0.0004830 in fMRI. Crossed person and common 60 s story-block 95% intervals were [0.0009811, 0.0040346] and [-0.0019241, 0.0004898]. The shared positive rule failed, conjunction p=0.886422. Eleven fMRI people selected zero correction at both outputs. Negative and discordant results remain. A separate proof shows that the gap adds the same linear predictor span as a pooled square-root sound envelope when the baseline contains the square root of mean level. Both measurement designs passed numerical identity checks. Thus a gain cannot identify a distinct detector of sound variance in this family. The study gives a prediction tool and a fixed negative cross-modal test, not a positive shared biological or causal finding. Small electrical sample, clinical and spatial selection, fixed stories, source filters, full-run cleaning and measurement differences limit scope. The package gives code, permitted inputs, source versions and terms, plans, every outcome, proofs, figures and repeat instructions.

auditory encoding · sound statistics · intracranial EEG · BOLD fMRI · temporal pooling · Jensen gap · residual prediction