afni.nimh.nih.gov

I still have trouble understanding what you're trying to achieve. If I have to guess, I would suggest that you obtain session-level effects for each subject by either (1) creating 5 separate regressors for each session, or (2) analyzing each session separately.

Gang