Can the Epistemic Value of Natural Kinds Be Explained Independently of Their Metaphysics?

The account of natural kinds as stable property clusters is premised on the possibility of separating the epistemic value of natural kinds from their underlying metaphysics. On that account, the co-instantiation of any sub-cluster of the properties associated with a given natural kind raises the probability of the co-instantiation of the rest, and this clustering of property instantiation is invariant under all relevant counterfactual perturbations. We argue that it is not possible to evaluate the stability of a cluster of properties without taking stock of the metaphysical picture used to account for that stability. Thus, even on the stable property cluster account, the epistemic value of natural kinds remains partly grounded in their metaphysical status.