fMRIprep documentation does a really good job when it comes to the CompCorr regressors – Thank you!

There is only one thing I can’t find – when using a subset of a_compcorr, t_compcorr (let’s say 5 with highest var explained), but not all, how many cosine variables should I include with a subset?

I think I know the answer, but I would be very grateful for a second, third, etc. opinion if anyone has a second!

My intuition would say to use all of the cosine regressors (since they may relate more to the act of high-pass filtering than they do to the compcor components themselves), but I am not sure.

Hi, sorry I didn’t see this. Yes, use all the cosine regressors, because they are all used before calculating the compcor regressors. We use a pretty standard threshold (128s) for determining how many regressors to compute. If you have true cognitive signals of interest that oscillate more slowly than that, then fMRI might not be the right modality.