All 30 PSD time rows are identical within each trial in 108 EmoEEG-MC 1.0.7 files

We tested 108 public power spectral density (PSD) files from EmoEEG-MC 1.0.7. The files cover 54 participants and two emotion tasks. All 2,268 trials contain one 320-value vector repeated 30 times. All source files match the fixed SHA-256 hashes. A split between time rows puts exact copies in training and test sets. A nearest-neighbor rule labels all 22,680 test rows correctly with this split. This score does not measure transfer to new trials or people. The source code applies a frequency mask to the time axis. It then writes one value to all 30 rows. Its numerical core matches 6,720 stored values in one video file within 8.882e-15 absolute error. We provide checked source files, audits, code, and a separate one-second band-power method. The replacement passes signal tests. It gives 30 distinct vectors in each of 21 trials from one EEG file. The source paper uses whole trials or people for its tests. Those splits avoid the copy path shown here. We do not test the effect on that paper's scores.

EmoEEG-MC · electroencephalography · EEG · power spectral density · feature audit · time resolution · data leakage · band power · reproducibility