Can there be an Inductive Logic?

Abstract

Abstract The results of the preceding chapter will now be brought to bear on an old question in the philosophy of science: can there be an inductive logic? Traditionally it was always assumed that logic had two branches--deductive and inductive. However, as we shall see, the development of modern logic from its initiation with Frege in 1879 up to the early 1970s tended to bring about a divergence between deductive and inductive logic. Carnap and Popper, despite their differences, agreed in reducing inductive logic to the theory of confirmation (or corroboration). Now confirmation theory certainly appeared something very different from deductive logic, as that existed in the 1950s and 1960s. So it seemed to many unreasonable to use the term ‘logic’ to cover both areas. In retrospect this was perhaps the point of maximum divergence between deductive and inductive logic, for the trend towards divergence has been reversed by the new results in artificial intelligence. Machine learning has introduced inductive rules of inference, while PROLOG has introduced control into deductive logic.