Fixed models gave negative adult age slopes for iEEG knee time and BOLD correlation area in separate cohorts

We tested whether older adult age marks shorter local signal persistence in two separate human studies. We used public iEEG knee fits from 79 patients and resting fMRI series from 122 typical adults. A local plan and code lock preceded the new age coefficients. Models gave each person equal weight. Electrical inference used patient clusters and controlled region, location, sex and electrode type. MRI inference controlled site, sex, motion and record length. Estimated differences per decade were -12.56% and -6.72%. Their 95% person intervals were -20.97% to -3.24% and -12.98% to -0.00321%. Two-sided Holm probabilities were 0.02008 and 0.04990. The fixed shared-direction rule passed. The MRI result was close to its threshold. Electrical exponent adjustment and source MRI bandpass filtering gave intervals that included zero. Added MRI eye-state and source-group terms also gave an interval that crossed zero. An exploratory paired MRI slope difference was uncertain. An exact model example gave a 10% shorter knee time and a 35.81% larger positive correlation area. This example was not fit to people. Direct calculations checked all 18,544 retained MRI series and both model standard errors. The electrical cohort was already used for an age result. The MRI study provides separate-cohort evidence for the selected measures. The available model choices do not establish a robust common age decrease. These cross-sectional results do not identify a common neural clock or a cause. Cohort, clinical, sampling, vascular and processing differences limit scope. The package gives code, source pins and terms, all aggregate outcomes, figures and repeat steps. Source arrays and person tables remain external under their own terms.

human neuroscience · intracranial EEG · fMRI · adult age · signal persistence · aperiodic spectrum · autocorrelation · measurement scope · cross-modal generalization · reproducibility