Reading guide for Tues. 9/8: Okasha, ch. 2, sel., pp. 29-33; Harman, “The Inference to the Best Explanation” (on JSTOR), §I (88-91); Harman, “Enumerative Induction as Inference to the Best Explanation” (on JSTOR).
It is easy to see how empirical generalizations could be supported by an inductive inference that rests a generalization on observed instances of it. But the relation between theories and the evidence that supports them is different, and people have suggested that a different sort of inference is at work. One common term for that sort of inference is “abduction.” Another is “inference to the best explanation.” The latter is the term Okasha uses, and it was introduced in the first of the papers by Harman we will discuss.
• Okasha’s brief discussion of inference to the best explanation touches on quite a number of issues, and the last two pages (pp. 32-33) are worth reading and re-reading closely.
• One of the issues Okasha addresses is the question whether one of these two sorts of inference depends on the other. The papers by Harman argue that inference to the best explanation is the more fundamental. When referring to induction, Harman will speak of “enumerative induction.” It’s OK to assume initially that this is what Okasha simply calls “induction,” but Harman describes the idea more precisely near the bottom of p. 90 of the first paper. (Notice that I have assigned only section I of this paper; the rest turns to topics that are less relevant to this course.)
• In his second paper, Harman replies to a critique of the first one and, in the course of doing so, he fills out his idea of how inference to the best explanation works. He outlines his views in a series of numbered points on p. 530, but his discussion of a two examples on pp. 531-532 may be more helpful. (If you are curious about the paper he is replying to, it appears just before Harman’s paper, so you can reach it on JSTOR by following the link to “Previous Item.”)