Hm. At first it would seem like StimLists might do the best job at what you need completed; however after contemplating the issue for a bit, it seems as though you may be better off simply making a trial for each pairing, and then using the WGR and BGR features to randomize the trials for you. If you make a trial for each pairing, you'll be able to have the WGR and BGR features properly randomize the trials for you - while making sure that each pairing still has the chance to be displayed.

For more information, look here:
RWG: http://www.empirisoft.com/directrt/h...amples_rwg.htm
RBG: http://www.empirisoft.com/directrt/h...amples_rbg.htm
RWG + RBG: http://www.empirisoft.com/directrt/h...es_rbg_rwg.htm