Constant signal (e.g. a dipole) added to a given noise pattern does not change the covariance.
We consider a matrix of noise: 100 channels x 1000 time samples (first column, the two panels are strictly identical)
add constant signal (zero or a constant "dipolar" source, whatever, constant in time but not in space). (Second column)
Compute covariance of the sum of the two first columns (Third column) --> strictly identical
Add Comment