This experiment was designed to investigate the effect of experience with one pattern of events upon the way Ss learn later patterns. Suppose one group of Ss begins by learning that there is a consistent pattern of reinforcement on a two-choice gambling device such that one side always pays off. We now shift them to a schedule in which payoff on the two sides is randomly arranged, averaging out to 50:50 over a series of trials. After exposure to this irregular schedule, a consistent pattern is reintroduced and the gambling device is again paying off regularly on one side, not necessarily the same side as before. Contrast this group of Ss with another treated identically in all respects save that they have never experienced an initial period of learning a consistent pattern of payoff. How quickly will the two groups learn the final consistent pattern? Will the effect depend upon whether or not Ss in the first group begin and end with the same side always paying off? Can differences in speed of learning the