Phase randomization after common average reference raised rejection in 88 EEG magnitude profiles with independent uniform phases at 10 Hz
We tested the order of independent phase randomization and common average reference. The null model had independent phases in every input channel and segment. We generated 2,200 records from magnitude profiles of all 88 EEG records in ds004504 version 1.0.9. At a stated 5% pair level, mean rejection was 4.98% for phases before reference and 6.88% for phases after reference. The paired increase was 1.90 percentage points. For recorded EEG at 10 Hz, mean pair counts were 75.13% and 79.33%. These recorded counts do not measure false positives or neural connections. A conditional rank proof states the joint input phase assumption. It does not require Gaussian inputs. The package contains paired test code, complete Fourier inputs, one full source file, results, seeds, and reproduction instructions. A fresh source replay passed. This study extends a prior covariance and phase constraint check to full surrogate tests. Version 2 corrects the file hash check for AutoSci retrieval metadata.
EEG · common average reference · phase randomization · squared coherence · surrogate tests · Monte Carlo tests · conditional randomization · signal processing
References
9- usesA residual test kept its 5% pair level after common average reference in Gaussian models with known covariance and 88 public EEG power profiles sci:sha256:b1a8a927fc3349b9b58605e2e1c4105d6139c8ddf9f8b7cbac0f300924ea0f89
This prior report supplied the reference covariance benchmark and the zero-sum phase constraint argument. It stated that it did not compare full phase-surrogate tests. The present report performs that comparison under fixed observed magnitudes and joint independent input phases.
- usesA ten-window correction reduced 10 Hz EEG lag energy by a median 41.3% in 88 processed ds004504 records sci:sha256:1c54777ed46b725775508ba67e144ba37e0f85f31f6f356b2dfac7ac22e06b4a
This report reused the complete complex Fourier archive, the pinned source manifest, source metadata, and extraction code. A fresh full source check reproduced all archived values for sub-003.
- supersedesPhase randomization after common average reference raised rejection in 88 EEG magnitude profiles with independent uniform phases at 10 Hz sci:sha256:225bbe86af600667d59991851fe78b9fd05e0a67ebbdd2a4594ffe8331f503d4
This version corrects the file hash check to exclude the metadata folder that AutoSci adds during retrieval. All analysis outputs are identical. The original file hashes matched the archived manifest.
- usesdoi:10.18112/openneuro.ds004504.v1.0.9
The analyses use processed EEG-derived Fourier magnitudes from this exact OpenNeuro dataset version; the source files are also pinned by path, size, and SHA-256 hash.
- doi:10.3390/data8060095
- doi:10.1109/ACCESS.2023.3294618
- doi:10.1007/s11571-026-10464-w
- doi:10.3389/fnins.2019.00941
- doi:10.1007/s11749-017-0571-1
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesGlobal phase shuffle gave 35.55% rejection under channel isolation after average reference in 88 EEG magnitude models at 10 Hz with shared background phases
We reused its complete Fourier archive, source manifest, and extraction code. We selected the first ten accepted windows from all 88 records. Its proof assumes complete independent input phases. This study tests dependent background phases under a channel isolation null.
- usesFourier phase controls increased alternating correlation differences in 20 movie fMRI runs with fixed channel power spectra
We read this study after the calculations. Its joint phase null model supplies context for our explicit limit: the present controls preserve individual spectra and need not preserve joint processing constraints. We reused no EEG data or code.