Four groups of rats received one food (F) trial and one water (W) trial daily in a runway apparatus. Two of the groups received the incentives in the sequence FW, while the remaining two groups received the sequence WF. One group receiving each sequence was hungry in Phase 1 and thirsty in Phase 2; the remaining two groups were thirsty in Phase 1 and hungry in Phase 2. All four groups demonstrated negative transfer in Phase 2 relative to naive controls. This finding indicated that the animals failed to form expectancies involving specific rewards in Phase 1. It was concluded that (1) Trapold and Overmier's (1972) two-process theory of instrumental performance must be elaborated so as to specify the conditions under which animals do and do not form specific reward expectancies, and (2) the present procedure affords a technique whereby the nature of such conditions may be empirically determined.