Hi Anderson,
Thanks for your response.
I agree on not splitting the fit. However I still have difficulties understanding what is done in FSL by adding confound EVs and motion parameters to regress out from the signal. FSL wiki states: "If you select this option then only the components of the main EVs that are orthogonal to the motion confound EVs will be used in determining significance of the effects of interest". This sounds like the main EV is orthogonalized to the confounds. Is that right?
Thanks,
Alireza