I have a few questions to add:

In this scenario, is the AllSubjects variable necessary?  If AllSubjects is intended to be included with your study groups, then the model for your proposed analysis below would need to include that [0 0 0 -1 1 0 0 0].  But, I would clarify if AllSubjects is needed.

I have set-up gender in the past as "1" and "2" values.  The key is not setting up a dichotomous covariate as "1" and "0."

Also, I have noticed that CONN will not recognize "0" value covariates, which can be an issue for any covariate that has been mean centered (de-meaned).  In these cases, I've been recoding "0" value covariates to 0.000001.  This way the covariate is not ignored (or at least I think it's being ignored based upon the N pop-up message one get when selecting a covariate).

Hope this helps,
Jeff Browndyke