Thank you Francois! Your answers were most helpful. I will follow your recommendations.
I would like to ask few more questions related to your answers.
Question 2)
Francois: "If you really want to use these deep structures, use them all, not only the ones were you want to observe activity, otherwise you introduce a strong bias in your analysis"
So just to be sure. If fore some reason I end up using mixed model, I should use also cerebellum to prevent bias (when selecting all subcortical sources)?
Francois: "The fact that you get all your activity at the center of the head might be an issue related with the dSPM computation and/or your noise covariance matrix."
There might be one huge mistake in my baseline. I marked some bad segments in the raw file of the baseline (in addition of cleaning up artifacts with ICA). Do I have to import the whole file to database before these bad segments are rejected, when the noise covariance matrix is calculated? This is the case when epoching, right?
Question 5)
I will contact my collaborator and ask exact list of details. This might take some time, but I post the list as soon as I get it.
Thank you in advance!