The main purpose of the research was to make a comparative study of: (1) trial-and-error learning, in which a subject performs and is provided with knowledge of results; and (2) observational learning, in which a subject either observes the performance of another subject or is otherwise provided with equivalent information as to the correctness and incorrectness of responses. In general, the earlier results showing superiority of observation learning were not substantiated. Other results indicated that male college students form new associations more rapidly under performance and less rapidly under observation, and that both sexes tend to repeat errors more often over trials under performance and less often under observation. Also, college students in other experiments learned more effectively on the second of two functions, whether it was performance or observation, indicating the role of a warm-up or learning-to-learn factor. (Author)