Two experiments confirmed sequential mediation of social interaction by investigating the effects of generalized expectancies on specific expectancies and the effects of specific expectancies on performance. Both experiments used a simulated tutoring task in which the subject took the role of tutor while a confederate took the role of student. In Experiment 1 subjects combined generalized expectancies about the effectiveness of certain tutoring responses with specific situational information to produce specific expectancies about the results of the tutoring responses under the experimental circumstances. Experiment 2 replicated this finding and showed that specific expectancies of the relative effectiveness of different responses influenced which response was performed more. Results were discussed in terms of cognitive motivation theory and social learning theory.