Sabbarton-Leary Nigel - The semantics and metaphysics of natural kinds

One source of philosophers interest in natural kinds is distinctively metaphysical: are there natural joints in nature, which our classificatory systemsin ordinary life or in sciencemight latch onto? For example, compare general terms such as cat, silver, carbon, electron, and planet, on the one hand, with expressions such as object bigger than a car, (call such objects nargs) and carbonated drink on the other. The members of the former list of terms, but not the members of the latter list, intuitively pick out a natural category of objects. But how are we to articulate what naturalness, in this context, amounts to?

One strategy for answering the question is to focus on what, if anything, is common to all the members of the class, and, relatedly, on the extent to which use of a given kind term delivers inductive or explanatory success. Nargs (which of course include buses, elephants and office blocks) have very little in common with one another; correspondingly, predictively speaking narg is an utterly useless kind, since nothing, aside from a lower bound on size, can be inferred from somethings being a member of that kind. W. V. Quine (1969) claimed that kindhood and similarity are variations or adaptations of a single notion (1969: 7), but that there are theoretical standards of similarity that lend themselves to better inductions. Thus [b]y primitive standards the marsupial mouse is more similar to the ordinary mouse than to the kangaroo; by theoretical standards the reverse is true (1969: 15). Similarly, colour is king in our innate quality space, but undistinguished in cosmic circles. Cosmically, colours would not qualify as kinds (1969: 14). So we might attempt to characterize the natural kinds as those kinds (unlike red thing and mouse, where the marsupial mouse counts as a mouse) that align with theoretical standards of similarity, thereby delivering superior inductive inferences. (This is not Quines way, however. Quine holds that it is a very special mark of the maturity of a branch of science that it no longer needs an irreducible notion of similarity and kind (1969: 52).)

Predictive (and explanatory) success has played a large role in the conception of natural kinds adopted by many philosophers of science (see e.g. Griffiths 1999 and Boyd, this volume), and in particular in Richard Boyds homeostatic property cluster account (1991). Many metaphysicians, however, hold that there must be something metaphysically or ontologically distinctive about natural as opposed to non-natural kinds; for example, that they are universals in something like David Armstrongs (1978) sense (see e.g. Ellis 2001: 678), or that they have essences in a metaphysically substantive sense (about which more later). For such metaphysicians, predictive successwhile it might be a common or perhaps even a universal feature of natural kind conceptscannot function as their defining characteristic, because such success comes in degrees. As Quine notes, [b]etween an innate concept of similarity or spacing of qualities and a scientifically sophisticated one, there are all gradations. Science, after all, differs from common sense only in degree of methodological sophistication (1969: 15). There is no cut-off, predictively speaking, anywhere in the spectrum between physics, chemistry, biology, the social sciences, and our ordinary, commonsense worldview; thus those who seek to endow natural kinds with a special metaphysical status must look elsewhere for the determining features of natural kinds.