Hello!

I have a question about 3dDeconvolve. I have a functional dataset of
an event-related experiment. I only look at the beta-weight value in
the output of 3dDeconvolve, which reflect the overall activation at
each voxel for each experiment condition.
I specify the trial index number of each experiment condition by the
'-stim_file' option. Now the problem comes up when I divide
some of the experiment conditions to sub-conditions. The trial numbers
are of course smaller for the divided conditions, but remain the same for the
unchanged conditions. Except the '-stim_file' options, the other
inputs to 3dDeconvolve remain the same. I compare the 3dDeconvolve
outputs before and after dividing experiment conditions, and find the
beta-weight values for those unchanged conditions are quite different.
The difference is so big that it changes the relative difference between
these unchanged conditions in the ROI statistics. I was wondering if it is
supposed to be the case according to the implementation of the
deconvolution method.

Thanks,
Zhuang