Hi Gang-
So is it somewhat accurate to say that, if your regression model includes "brain" and BPRStot (which is a psychosis score), and your "brain" signals already reflect correspondence to a parametric regressor (reward prediction error, in this case), that the "BPRStot" coefficient that comes out of either 3dttest++ with a covariate or 3dMVM with a covariate reflects something like the MODULATORY EFFECT of BPRStot on the brain response to prediction errors (and thus a kind of, sort of interaction)?
Jim Waltz