Sui Generis-ness, Parsimony and Innocence: The (Meta)2physics of Parthood

A metaphysical naturalist could find the following combination of claims attractive. First, part-whole and composition in physics are sui generis and lack some of the ‘core’ features we ascribe to these concepts and their worldly satisfiers in first-order metaphysics. Second, having agreed that some physical objects of interest satisfy sui generis concepts and/or relate by sui generis relations, none among these objects satisfies a classical concept or relate by a classical part-whole relation (e.g. the concept or relation of mereological part). The first claim I read as one of ‘appropriation’: the structural relations between physical objects of interest are sui generis and yet they pertain to the mereological kind. The second I read as one of ‘elimination’: metaphysically abstracted part-whole (or composition) has no instances in well- regarded physical domains. The dissertation argues for appropriation and against elimination. For appropriation, because current physics sanctions relata of part-whole relations (or at least satisfiers of part-whole concepts) that clash with intuitive, seemingly analytic principles for part-whole, e.g. the Antisymmetry postulate (x and y are mutual parts only if identical). Against elimination, because whether these objects of interest to physics also relate by ‘canonical’ part-whole (with the intuitive principles) is largely a question of parsimony. One removes instances of the canonical relations because these are not needed to account for the composition of objects that already relate by the non-canonical ones. But some of these relations at least (such as mereological part-whole) resist the pressure from parsimony, for they come at no cost once the objects already relate non-canonically (e.g. in opposition to the Antisymmetry postulate). The latter we can argue for in (at least) two ways: 1. canonical and non-canonical part-whole are members of a single kind, 2. canonical part-whole is of a kind with identity. Given either view and a preference for theories with minimal kinds, instances of the canonical relation do not increase a theory’s profligacy, because their kind is already instanced in a theory of objects that relate non-canonically. My preference is for the latter view.