Abduction and Induction have existed mostly as separate subjects in Artificial Intelligence and other disciplines. They have each been studied as central inference mechanisms in order to tackle a set of particular but largely different problems. Abduction is mainly used to address problems of diagnosis and planning whereas induction is primarily used in the problem of machine learning. Within the wider context of the formation and the development of a knowledge theory, however, these two forms of reasoning interact and can be integrated together in a cooperative and enhancing way. In this paper we study this possibility of integration between abduction and induction and present a cycle of knowledge development based on this integration. This paper is a summary of part of the introductory chapter of a recent book (Flach and Kakas, 2000) which addresses more generally the issues of the relation and integration between abduction and induction.