The present study was (1) to analyze the existing literature pertaining to thinking processes, (2) to develop a conceptual model for training thinking abilities, and (3) to design programs for promoting thinking abilities. In this study, thinking process was dichotomized to include centripetal and centrifugal thinking. The centripetal thinking was defined as the capacity of working memory and automatized thinking process which efficiently utilizes it. The centrifugal thinking includes ability for formulating problems, challenging assumptions, and generating alternatives. Two thinking ability development programs - one for centripetal and the other centrifugal - were developed to be implemented in 12 sessions each taking 40 minutes, and administered to the 6th grade students. The number of subjects of centripetal thinking program was 225(experimental group 109, control group 116), and that of centrifugal program was 224(experimental group 108, control group 116). The General Intelligence Test (Chung & Kim, 1990) was administered as pre- and post-test for the former, and the Creativity Test Short Form(Lee & Chung, 1989) for the latter. After two months, Centripetal Thinking Test and Centrifugal Thinking Test were administered as retention tests for each group. In the centripetal thinking program, statistically significant difference of increment was found only in numerical reasoning(p＜.05), while in word usage, verbal reasoning, arithmetic reasoning, figure reasoning, there were no differences. This result indicates that the program was not domain-specific. In retention test, statistically significant difference was also found only in numerical reasoning(p＜.01), while in sentence linkage, logical reasoning, verbal reasoning, argumentating and discriminating arguments and their bases, no differences were found. The reasons for the failure in obtaining expected result were the brevity of training sessions for developing automatizing thinking and the possiblity that students' partly successful strategy had hindered the learning of new strategies. In the centrifugal thinking program, statistically significant difference of increment was shown in total creativity score(p＜.01) and f1uency(p＜.001), while in flexibility, originality, openness, there were no differences. What was intended in this program was not the automatization of thinking but change in attitudes or viewpoints, the latter being relatively easier to come about. That is, change in attitudes or viewpoint needs less time than automatization. In retention test, statistically significant difference was found in generating alternatives(p＜.01) and challenging assumptions(p＜.05). But in formulating problems, the control group was found to perform better than experimental group (p＜.05). The fact that the fluency in the formulating problems of control group was higher than that of experimental group, while the fluency in generating alternatives of experimental group was higher than that of control group, was taken as evidence that as a result of the training the experimental group took the test too deliberately utilizing elaboration strategies. These results indicate that working memory and automatization processes are not domain-specific but content-free, supporting the automatization theory, working memory and parallel processing models. It was concluded that an extra-curricular thinking skills development program could be helpful for developing thinking abilities but as one study shows(e.g. Lee et al., 1992), other factors such as family environment, everyday experiences, institutional infrastructure(e.g. classroom structure, teacher, etc.) are no less important than the program itself in promoting students' thinking abilities.