Source
Alexander Bird and Emma Tobin, “Natural Kinds”, Stanford Encyclopedia of Philosophy (2008; substantive revision 28 January 2022), LIT-244, read in full in NOTE-769: §1.1.1, §1.2.1 and §3.
What was actually shown
- Not sufficient for kindhood (§1.1.1). Support for inductive inference is “a necessary but not sufficient condition”, and “does not go beyond” the condition that members share a natural property, which white things, negatively charged things and things of mass 1 kg meet without forming kinds. Participation in laws fails the same way: laws “may concern natural qualities and quantities that do not define kinds (e.g. Newton’s law of gravitation)”.
- Induction is over properties and relations (§1.2.1). “The real answer to ‘what tends to confirm an induction?’ is not ‘natural kinds and their instances’ but is ‘natural properties and their instances’”; and inductions “often concern relations … But we do not suppose that objects that enter into a natural relation form a natural kind.”
- Not sufficient against convention (§3). Lemeire (2018), as reported: epistemic success “is constrained by deference to the world not only in cases of natural kind classification, but also in cases of conventional and pragmatic classification”, so a purely epistemic approach “fails to give us a principled way of distinguishing between the natural kinds and their conventional counterparts”. Kendig and Grey (2021), reported likewise: a cluster’s stability cannot be assessed without the metaphysics that accounts for it.
The counterexamples could have failed: a class of white things or of charged objects could have turned out to support the projections a kind supports. That they do not is the survey’s case.
What this does not say
- Not that usefulness is irrelevant. Every account in the survey keeps inductive support as necessary, and the epistemology-first accounts (Slater 2015; Ereshefsky and Reydon 2015) make it central.
- Not what does suffice. The survey leaves the “kindhood” condition (criterion 4) unclear, and offers candidates without endorsing one: Boyd’s homeostatic mechanisms, Slater’s stability, Hacking’s empirical discovery of a category’s instances, essences.
- Not a refutation of the epistemic approach. Lemeire’s argument is reported, not reproduced, and Slater’s stability clause may escape it. This is why the account is Proposed.
- Not about compression or information. The entry never mentions either. The parallel with THEORY-036 is the record’s, not the survey’s.
Connections
- QUESTION-028. The question asks when generalisable, intervention-relevant relational structure warrants a real kind. On this account those qualifiers are necessary and, by themselves, insufficient; an answer needs a further clause, stated before the data. This does not answer the question, which asks for that clause.
- CLAIM-150, the record’s guard that predictive success does not establish constitution, is the same point made for communicative categories; Lemeire’s argument is a published version of it.
- THEORY-036 says Dennett’s compressibility over-admits real patterns and that projectibility clauses do the excluding. This account says the same of usefulness for kinds. The two are independent and agree.
- THEORY-057 reaches a negative verdict on emotion as a natural kind, which the survey reports from Griffiths (1997).