Temporal halves had greater mean squared disagreement in fixed alpha EEG lag matrices than balanced random halves for the first 200 accepted seconds of 88 processed ds004504 v1.0.9 records
We tested fixed EEG lag matrices in all 88 processed eyes-closed records from ds004504 version 1.0.9. Each record supplied its first 20 accepted 10-second windows. The fixed score used signed alpha cross-power in channel differences, divided by projected total power. This score cancels a common reference signal. Mean squared change between temporal halves was 0.003418. The exact mean over balanced random halves of the same windows was 0.002703. Mean excess was 0.000716. The 95% person bootstrap interval was 0.000281 to 0.001181. The ratio of the two means was 1.265. Scores used squared dimensionless matrix norms. The first 100-second alpha check and the theta check had intervals that included zero. All six planned comparisons are reported. The package preserves all 5,280 band matrices, source pins, receipts, code, proofs, bootstrap draws, and repeat instructions. Fresh retrieval checked all 88 source hashes. Independent spectra matched three fixed source records. All person scores passed a separate calculation. A complete numerical repeat matched byte for byte. The interval assumes independent people and conditions on the supplied processing and selected windows. The result shows a time-order effect in this score. Serial dependence, artifacts, and power change can contribute. It does not establish neural drift, diagnostic value, or repeatability across visits.
EEG · imaginary cross-power · lag matrices · temporal stability · balanced random halves · finite-population reference
References
6- Four common references reversed alpha imaginary coherency signs at a 0.05 cutoff in 2,441 of 10,560 pair tests from 88 processed ds004504 v1.0.9 records sci:sha256:c5b0e845e0d595028954229a6f2dd3a4ce2f5bd8f831673c4fb488b81b4728eb
[Legacy relation: reproduces] Fresh retrieval of all 88 pinned sources reproduced all 1,760 early alpha window matrices and all first-20 three-band means to numerical precision. All selected starts matched. This relation is limited to those matrices and starts.
- usesFour common references reversed alpha imaginary coherency signs at a 0.05 cutoff in 2,441 of 10,560 pair tests from 88 processed ds004504 v1.0.9 records sci:sha256:c5b0e845e0d595028954229a6f2dd3a4ce2f5bd8f831673c4fb488b81b4728eb
This report supplied the pinned sources, metadata, complete example, fixed spectral function, and reference projection proof.
- usesThe 95% interval included zero for extra early-to-late disagreement of non-target Pz means with equal letter weights in 17 selected ERP CORE records sci:sha256:d3c6ac18ea7ee42ecc90e203bdef1324309fde5a7ca2214801da4ea6fb5b3e37
Its exact finite-pool comparison supplied related method context for the proof and interpretation. We read it after our local plan. The present test applies the same established sampling principle to fixed signed EEG matrices.
- usesReal zero-sum EEG filters fitted at 10 Hz on the first 100 accepted seconds gave lower mean imaginary coherency on the last 100 accepted seconds in 88 ds004504 v1.0.9 records sci:sha256:f5e030df9210b8b03cead5bf486eb8ea1f844ae62c24d57781e6992ef2a6c8d2
Its early and late filter comparison motivated a time-order audit of a fixed whole matrix in the same source group.
- usesdoi:10.18112/openneuro.ds004504.v1.0.9
The analysis used processed derivative EEG files from this exact OpenNeuro dataset version.
- doi:10.3390/data8060095
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesTemporal halves increased squared disagreement 2.76-fold for EEG theta-to-alpha scores in 95 ds008896 v1.0.1 records
Its temporal versus random half comparison for EEG lag matrices informed the question. This study tested a different scalar endpoint and a different public cohort.
- usesThe 95% interval included zero for extra MMN voltage disagreement between early and late halves in 15 NEMAR 250 Hz ds005429 records
Its EEG analysis supplied the comparison of temporal halves with a finite random split reference. This work tests a distinct auditory voltage contrast and dataset.
- usesRandom splits gave smaller mean log transfer gaps for fitted zero-sum theta/alpha EEG filters with ridge 0.1 in 88 ds004504 v1.0.9 records
The temporal-versus-random comparison of fixed alpha lag matrices informed the question and the paired split benchmark. This study instead refits regularized theta/alpha filters and compares their transfer gaps. The records overlap, so this is not an independent sample replication.