With fixed references, two Hurst measures shared 64% of selected cortical regions in ten iEEG patients and 32% in 34 fMRI controls

We tested whether two released Hurst measures select the same high-deviation cortical regions. Each person had a separate reference-normalized map for FIP-MLE and WPLM. The selected quarter had to retain more than half of its regions under estimator replacement. Fixed shared stability tests failed in 21 iEEG patients and 34 different fMRI patients. Mean retention was 62.55% and 34.16%. A later fixed follow-up used 33 MNI and 18 fMRI calibration people. Ten eligible MNI test patients and 34 fMRI test controls gave 64.17% and 31.51% retention. Their 95% person intervals were 59.17–68.33% and 27.83–35.20%, conditional on fixed references. The opposite pattern passed its numerical rule, with nominal three-claim Holm p=0.01171875. It is not independent confirmation: the first clinical reference summaries included the later test people. Reference resampling widened the electrical follow-up range across one half, and most electrical reference draws lacked full coverage. A wider electrical contact check was also uncertain. Matched-location and processing checks remain in the package. The result limits one shared map-stability claim. It does not identify a neural cause or validate a treatment target. The contribution is a reproducible benchmark with public source tables, code, hashes, fixed plans, a dependence correction, all exclusions, negative results, figures and repeat instructions.

Hurst exponent · intracranial EEG · fMRI · normative mapping · estimator robustness · epilepsy · reference uncertainty