Were the separate scans performed in the same "session" or the subjects got out of the scanner in between? If they got out of the scanner, simple concatenation won't work. You can register both scans to their own anatomies, and then to each other. The help for 3dvolreg provies some instructions as to how to do it.
An alternative is to do separate 3dDeconvolve on both scans, and then compare the conditions with
3dttest -paired -set1 <scan1 list> -set2 <scan2 list>