If I'm doing an analysis of fMRI data, in which a single subject performed
condition 1 (and rested) in session 1,
condition 2 (and rested) in session 2,
and condition 3 (and rested) in session 3,
and I put them all in a single design matrix in that order,
if I want to see areas whose signal changes correlate with condition 1,
I would use the contrast
(1 0 0). No problem with that.
I was assuming that if I wanted to see areas which increase in activity
from condition 1 to 2 to 3, I would use the contrast
(-1 0 1).
What someone else asked and I couldn't answer is: would that contrast also
include areas whose signal correlation increases from condition 2 to 1 to
3, or 1 to 3 to 2? It seems like it would pick out all areas which show an
increase between conditions 1 and 3 regardless of what happens in
condition 2, in which case I need to do something different in the
analysis. Is that true, and if so what is the recommended solution?
(and yes, there's a confound between session and condition--that's ok for
the moment. If I did put all the conditions in a single session with their
independent rest conditions, how I would answer the question?)
Many thanks in advance,
Jessica Turner
Neuroimaging Research Lab
Long Beach VA
Long Beach, CA 90822
__________________________________________________
Do You Yahoo!?
Send FREE Valentine eCards with Yahoo! Greetings!
http://greetings.yahoo.com
|