During a source estimation exercise using the MNE-Python package, the localization error (which is calculated using Euclidean distances between the pairs of ground truth and estimated locations) was found to be exactly the same for both sLORETA and eLORETA, despite using different SNR. Can you explain why this might be the case?
if the noise you are adding is zero-mean white noise, then the estimated source location is likely still correct even when adding a little more noise. This is why in simulation studies, you should add some noise with a more interesting covariance structure, for example the noise covariance from our sample dataset.