Purpose: The purpose of this paper is to show that changes in social systems do not occur randomly. They are consistent with what has gone on before, with the history and identity of the system. As long as the organizing principles of a dominant culture remain unchallenged, behavior of all the social-units originating from this culture will remain unchanged. Design/methodology/approach: The concepts developed in this paper are the results of 30 years of real life experimentation with organizations and cultural transformations in different cultures. The paper demonstrates how an analytical culture, despite its well-known shortcomings, keeps reproducing the same set of non-solutions all over again. Findings: To change, systems need to go through an active process of unlearning. Unlearning is an iterative and collective process of the second-order learning. A participative and iterative design process with the aim of replacing the distorted shared images is the most effective learning tool to produce a second-order learning and a desired change in the behavioral pattern of a social system. Originality/value: The value is to appreciate the real source of resistance to change.