Jeremy,
Wothout access to the data, it is hard to make judgment, but, first, do those huge % signal changes (also baseline coefficients around 10 and condition coefficients around 40) occur everywhere, or only around some area such as the border or outside of the brain?
In stead of converting the regressor coefficients into percent signal change, I prefer normalizing the original dataset before running 3dDeconvolve so that the regressor coefficients would be automatically the percent signal change. Check out Doug Ward's following message:
[
afni.nimh.nih.gov]
Gang