Three Eras of Computational Logics of Discovery: Deductive Past, Inductive Present, and Abductive Future

The recent history of the field of “logics of discovery” spans half a century yet includes three major transformations in the general methodological framework. In order to discern the common patterns in this recent history of computational ideas, and to better guide future investigation in computer-related discovery in terms of reasoning by abduction, we first critically review the three phases of research from the computational point of view, focusing on the principles, achievements and obstacles in each stage. We show that the first, deductive phase, could not solve the problem of human-machine creativity. The second overemphasizes induction’s ability to discover novel hypotheses from data. The third, in turn, is an unfolding resolution to initiate the true discovery agenda via the abductive constructor we call Plausibly Anticipatorily Correct learning.