Hello SPM-world,
I have a follow-up to a question I asked earlier about betas. I am
interested in determining whether individual differences in measures of
task performance, personality, etc., predict differences in activation of
ROIs that have been identified in a group-level contrast.
Could I extract betas in each of my subjects for the group-defined
functional ROIs (which would yield one beta for each trial type per
subject), and then treat the betas as a dependent measure - correlating
them with another measure such as task performance or individual
differences in personality? Would it be best to use the straight betas as
a dependent measure in this way? What about computing beta difference
scores between conditions and correlatiing those with behavioral
difference score? SImilarly, what about using proportional differences?
How does this compare to using SPM's basic models to perform a simple
regression using my individual difference measure to predict brain
activation? I have tried such regressions, but SPM doesn't produce a
bivariate scatterplot showing the relationship between activation in a
given cluster and my individual diff. measure.
Thanks much!
-KevO
__________________________
Kevin Ochsner, PhD
Department of Psychology
Building 420 (Room 430)
Stanford University
Stanford CA 94305-2130
__________________________
[log in to unmask]
lab: 650-725-0797
fax: 413-845-3695
__________________________
|