MVPA with behavioral measure

Hi all,

I would like to use a data-driven whole brain approach (i.e. MVPA) to find which regions of the brain correlate with a behavioral measure undergone outside the scanner.

I have got two groups of participants (patients and controls) that were all seen before and after treatment (pre and post).

As for the contrasts, is it correct to create 4 second-level covariates with the following contrasts?
- Patients: 0
- Controls: 0
- Behavioural measure patients: 1
- Behavioural measure controls: -1

...and then post>pre for the conditions?

Thank you!