Abstract

In this paper I lay out a non-formal kernel for a heuristic logic—a set of rational procedures for scientific discovery and ampliative reasoning—specifically, the rules that govern how we generate hypotheses to solve problems. To this end, first I outline the reasons for a heuristic logic (Sect. 1) and then I discuss the theoretical framework needed to back it (Sect. 2). I examine the methodological machinery of a heuristic logic (Sect. 3), and the meaning of notions like ‘logic’, ‘rule’, and ‘method’. Then I offer a characterization of a heuristic logic (Sect. 4) by arguing that heuristics are ways of building problem-spaces (Sect. 4.1). I examine (Sect. 4.2) the role of background knowledge for the solution to problems, and how a heuristic logic builds upon a unity of problem-solving and problem-finding (Sect. 4.3). I offer a first classification of heuristic rules (Sect. 5): primitive and derived. Primitive heuristic procedures are basically analogy and induction of various kinds (Sect. 5.1). Examples of derived heuristic procedures (Sect. 6) are inversion heuristics (Sect. 6.1) and heuristics of switching (Sect. 6.2), as well as other kinds of derived heuristics (Sect. 6.3). I then show how derived heuristics can be reduced to primitive ones (Sect. 7). I examine another classification of heuristics, the generative and selective (Sect. 8), and I discuss the (lack of) ampliativity and the derivative nature of selective heuristics (Sect. 9). Lastly I show the power of combining heuristics for solving problems (Sect. 10).