In DirectRT, the random selection of items through stimlists occurs before the trials themselves are randomized through BGR and WGR.

Your trials occur in pairs--and each pair of trials is accessing the list 8 times. But there are 12 stimuli in the list--so look what happens... The first pair uses 8. The second pair uses the 4 that are left over and then draws 4 more from a fresh set of 12. This makes it very likely that the second set will have repeated stimuli. Because you are also randomizing the order of the three sets, you could see this repetition occur at any point--depending on where the second set ends up being presented.

Does that make sense?

An easy fix would be to place just 8 stimuli in your stimlist instead of 12. Or you could create a separate stimlist with 12 stim for each pair of trials. That way they can draw 8 (or any number) and it won't affect any other trial sets. The lists could even be identical so long as they have different names.

Hope that helps,
Blair