afni.nimh.nih.gov

So out of curiosity over the use of 3ddeconvolve for blocked datasets, you are essentially treating each block as an "event" in your 1d file by denoting it with a 1 and everything else is a 0? This would mean that you might have very few "events" (perhaps 4-8) and might severely restrict any power for detecting differences. Am I interpreting this discussion correctly?