Residual spatial contrasts repeated on average in 23 EEG and 19 fMRI listeners; a fixed EEG–BOLD relation failed its shared test

We tested whether spatial response differences share timing across listeners after fixed word, sound, and nuisance controls. Separate public human studies supplied scalp EEG and fMRI during one chapter of Alice in Wonderland. Eight people per study supplied reference and development. A local code lock preceded tests in 23 EEG and 19 fMRI listeners. The EEG contrast was posterior minus anterior scalp voltage at 300-500 ms. The MRI contrast was left posterior minus anterior temporal BOLD. Mean residual correlations, summarized as tanh(mean Fisher z), were 0.03438 [0.02139,0.04736] and 0.13847 [0.08903,0.18722]. Brackets give 95% person Student limits. Positive counts were 22/23 and 15/19. The fixed shared mean, majority, and phase-reference rule passed two-claim Holm correction, p=0.01921. Joint reference and target resample ranges also stayed positive. A fixed EEG-to-BOLD temporal model failed its shared test, Holm p=0.617. EEG spatial contrasts also repeated before word onset. Omitting eye controls reduced the central EEG score. Later gain, common-signal, protocol, and behavior checks are descriptive. The positive result concerns residual timing reliability in two measurements. It does not identify the same missing language computation, a neural source, anticipation, or causality. One story, source region selection, separate cohorts, unequal anatomy, source filters, vascular effects, and protocol differences limit scope. The package supplies compact inputs, all negative scores, source pins and terms, code, plans, figures, numerical audits, and repeat instructions.

human neuroscience · EEG · fMRI · naturalistic language · spatial contrast · intersubject correlation · independent listeners · reproducible analysis