Dear Jack, > 1) Is this strictly unidirectional (say area A -> B but independent from B > -> A) or could it be that there is no directionality or no independance ? It is strictly uni-directional in the same sense as a regression is uni-directional. They are not independent in the same sense as a regression is not independent from the correlation. > 2) Is it possible that there is some effective connectivity that the PPI > cannot see and that structural equation modelers could see (and inversely) In a simple PPI one usually infers the modulation (by a psychological factor)of a contribution of a single region to another region. SEMs usually contain multiple regions. Mathematically both are related. Simple SEMs without latent variables are almost identical to a multiple linear regression, the only difference is that all path coefficients are estimated simultaneously. One obvious difference is that in a PPI analysis all voxels are possible "target" regions whereas in SEMs all regions are predefined. -Christian %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%