Defining intrinsicality via naturalness: The incoherence of Lewis’s persistent perfectionism

Langton and Lewis hold that an intrinsic property cannot differ between duplicates. They characterize duplicates as entities that share the same basic intrinsic properties, which they in turn define in terms of naturalness. Lewis had earlier maintained that duplicates share the same perfectly natural properties. Langton and Lewis argue that their definition, though more complex, improves on Lewis’s earlier proposal because it does not presuppose that every perfectly natural property is intrinsic (PNI). They note, nonetheless, that Lewis still endorses (PNI). I argue that, given Lewis’s modal realism, he cannot coherently endorse the Langton–Lewis definition while maintaining (PNI). I then turn to Marshall, who similarly defines intrinsicality in terms of perfectly natural properties, while dismissing grounding-based definitions as illusory. In response, I revise Rosen’s influential grounding-based definition to show that such approaches remain viable. Finally, I argue that Marshall’s own definition of intrinsicality fails both extensionally and conceptually.