I see your point; I'm curious if anybody's empirically tested this. I don't think it should make a big difference either, and it's unlikely that anyone is able to sample precisely on the TR using an event-related design, since jittering will probably throw you off. Exactly how FIR bins are interpolated between time points, I'm not sure, but I trust the software packages to do a good job (although I have heard that AFNI's TENT function is more flexible for estimating time points that are not on the TR time grid).
If you come across anything else, let me know!
-Andy
|