Abstract

According to best systems accounts, laws of nature are generalizations in the best systematization of particular matters of fact. Metrics such as simplicity and strength determine which systematization is best, but these are notoriously language relative. For this reason, David Lewis proposed a constraint on languages of inquiry: all predicates must be natural. This constraint is sometimes interpreted as requiring us to know which natural properties are instantiated in our world prior to scientific theorizing. I argue that this interpretation is incorrect. I provide a better interpretation and show how it undercuts an influential epistemological objection to Lewis's best systems account of laws due to Bas van Fraassen.