Dear Group,
Bruno's question about how proportional scaling is implemented in
SPM brought to mind (by association) a question I've been
meaning to ask for some time: Is there any evidence at all that a
proportional scaling model for within a session is a good model
for fMRI data? I'll start things off by saying that I have seen
evidence which goes against a within session proportional scaling model.
However, I have seen very strong evidence for
between session scaling effects.
(I posted these latter results on http://cortex.med.upenn.edu/papers.html
" Studies of gain artifacts in fMRI data").
I think this is an important issue because proportional
scaling restricts inference to relative changes in the brain, which
might not always be the domain of interest. If there were no evidence for
within session scaling differences across brain volumes, then proportional
scaling within session could be justifiably avoided, and hence the
restriction to inference about relative change would be avoided.
Sincerely,
Eric
Eric Zarahn
Columbia University
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|