If your covariate is subject-specific, you should:
1. define appropriate contrast on the individual level ,i.e. for each subject's model. According to your post these contrasts should be A+B and B-A, although I don't understand what do you mean by activation and deactivation.
2. One sample t-test --> enter all individual con images, enter your covariate (it should contain one value per subject). In your case you should do it twice, separately for each contrast.
3. Define contrasts: [0 1] for positive correlation, [0 -1] for negative correlation.
Hope this helps,