I'm a little confused about how we're supposed to treat time intervals in our data which we want to ignore. This includes semi-rest periods, where we know there will be activation, but don't care about it.
If I don't introduce regressors for these areas, they'll contribute to the error term, making intersubject analysis harder for no good reason. If I do, it's not clear what sorts of regressors I need (perhaps one for each interval or something).
On a somewhat related note, do you have any suggestions on how I might go about picking gamma functions for my various task parts? I've read this:
[
afni.nimh.nih.gov]
And it seems like I might want to stick with GAM(b, c) (after figuring out appropriate values for b and c).
Thanks!
(BTW, if you read this Rick, thanks for your help on the other threads).