At first, I thought "absolutely nope." Especially with MediaLab. I could imagine ways but given how much randomization you are talking about, these "ways" would be very cumbersome. Especially the part where you are randomly pairing images with the labels (not even sure how you are doing that! --in MediaLab??).

But THEN I saw you were using DirectRT also. Here, there *might* be a way. If you create all you randomization for a given session using a DirectRT input file, DirectRT saves a COPY of that input file with all of the randomization outcomes in it--in the same folder as your original input file. You could take THAT input file (i.e., the copy) and administer that to your second subject. It will be identical in all respects because it will be the same input file that the first subject received. That's because if randomization functions are used in a DirectRT input file, it's actually the COPY that is used in the session--not the original.

But again, if it's mostly MediaLab content, then it gets trickier to do a design like this. Does this help at all?