The development of pupil behavior in the process of creative problem solving was investigated. A battery of pretests and post-tests (including written compositions, oral compositions, and open stories) was developed and administered as part of a program to develop cognitive skills. Treatment groups received instruction for 30, 60, or 90 minutes per week for 23 weeks. Results from 200 fifth-grade subjects in a total of 24 classes (six classes in each of the four treatment groups) significantly favored two experimental groups. These were (1) the group receiving the basic treatment in problem solving plus extra training in abstract thinking and (2) the group receiving the basic treatment plus extra practice in problem solving. These two groups also showed similar results when high, medium, and low IQ levels were analyzed. Tables, figures, and references are given. (Author/DE)