Hi, I have a set of stimuli with concurrent behavioral data on two dimensions, A and B, with a low correlation between the two. I'd like to see the effect of these behavioral measures on the brain data. In the old days of -stim_times, to solve this I'd create 3 regressors, the first one for the mean effects (just 1s at the onset of events); the second one with the normalized ratings for A (zero mean); the third one with the normalized ratings for B (zero mean).
With the -stim_times_AM2 option, I should be able to create just two .1D files: times*A and times*B. 3dDeconvolve should generate two regressors for each, one for the mean effect and one for the behavior-modulated effect. However, since the times are the same for the two .1D files (I have two behavioral measures per stimulus), I get a collinearity error. I THINK that 3dDeconvolve simply ignores one of the collinear regressors, in cases such as this one, but I'd like to make sure that using -GOFORIT is legitimate in this case.
Thanks.
-G
