Three experiments examined the role of contingent associations in learning double-function, verbal-discrimination lists. Some 15-pair lists were constructed of category instances in such a way that the learning of three contingent associations based on category names would mediate correct performance for all 15 pairs. The first experiment gave no evidence that the three contingencies aided learning. The second experiment showed that subjects could be taught the three contingent associations in isolation and that they could then apply them successfully to the double-function list. To make the contingencies more apparent to the subjects, the category names were used during feedback in the third experiment. The learning was not facilitated by such feedback. It was concluded that the learning observed for the usual double-function list does not involve contingent associations. (Author)