Induction, Normality and Reasoning with Arbitrary Objects

This paper concerns the apparent fact — discussed by Sinan Dogramaci (2010) and Brian Weatherson (2012) — that inductive reasoning often interacts in disastrous ways with patterns of reasoning that seem perfectly fine in the deductive case. In contrast to Dogramaci's and Weatherson's own suggestions, I argue that these cases show that we cannot reason inductively about arbitrary objects. Moreover, as I argue, this prohibition is neatly explained by a certain hypothesis about the rational basis of inductive reasoning — namely, the hypothesis that inductive reasoning is fundamentally reasoning about what normally happens (in a non‐statistical sense).