In the previous posts, I hinted that the unification challenge for knowledge-first epistemology outlined there could be best answered via the claim that knowledge itself was a psychological natural kind. But what is a natural kind, such that this is a plausible suggestion? In this post, I’ll outline a recent theory of natural kinds, due to P.D.Magnus, which can do this job - or so I will later argue. First, however, I’ll start by outlining the standard view which is associated with the idea that knowledge is a natural kind - the naturalistic epistemology defended by Hilary Kornblith (2002, 2007). While the view that I’ll end up with has things in common with Kornblith’s theory of knowledge, it has further resources which ultimately make it more compelling for an account of knowledge and - by extension - as an answer to the unification challenge.
Kornblith’s naturalistic epistemology takes as its focus Richard Boyd’s influential homeostatic property cluster theory of natural kinds (Boyd 1989, 1991, 1999). ‘HPC’ theories of natural kinds, in opposition to previous essentialist accounts of kinds in the philosophy of science, aim to account for the so-called projectability of kinds (viz. the idea that bona fide natural kinds, in opposition to mere conventional and pragmatic groupings, are those which feature in and explain the epistemic reliability of a large class of enumerative inductive inferences) via the claim that natural kinds [at least in the special sciences] are structured clusters or repetitions of (macro-structural, extrinsic or functional) properties whose reliable co-occurence in nature is explained and maintained by the operation of a causal mechanism. It’s central claims are as follows:
Property Cluster: natural kinds [in the special sciences] are identified with a clusters of properties that regularly occur together.
2. Homeostatic Mechanism: a causal mechanism underlies and explains why the clustered properties occur together.
Causal Import: natural kinds, [(1-2) explain] feature in important causal inductive generalisations.
This feeds directly in Kornblith’s version of naturalistic of epsitemology, which is comprised of the following:
Knowledge (and not mere true belief) plays a causal and explanatory role within one scientific domain (viz. cognitive ethology).
Within this domain of cognitive ethology, knowledge picks out a specific homeostatic property cluster: viz. reliably formed true belief.
‘Knowledge’ picks out the same homeostatic cluster of properties (RTB) in all domains.
Knowledge is RTB.
As discussed in yesterday’s post, Kornblith’s naturalistic epistemology is also typically associated with 5:
(5) the sole target of epistemological theorising is knowledge itself, and not the concept of knowledge.
Emerging in direct response to these challenges, recent literature on natural kinds has sought to approach the topic of natural kinds in a different way. Taking as their basis the central explanandum of a theory of natural kinds viz. their projectability - this third and broadly ‘pragmatic’ approach to natural kinds aims to provide a philosophical theory of natural kinds grounded solely in the actual (and a posteriori) epistemic role that individual natural kinds in various domains perform in contemporary scientific theorising. One of the most developed accounts is that of P.D.Magnus in Scientific Enquiry and Natural Kinds: From Planets to Mallards (2012) and (2014, 2018). This takes as its starting point a distinction between two questions that can be asked about natural kinds have been systematically conflated by proponents of traditional metaphysical approaches to natural kinds. In this metaphysical framework, attempts to answer the naturalness question (that of what, the naturalness of paradigmatically natural kinds consists in) proceed by providing an account of the metaphysical nature or structure of such kinds - for example as intrinsic essences, or causally maintained property clusters - in order to explain the role that natural kinds play in our inductive and explanatory practices. They thus treat he ontological constitution and epistemic or explanatory utility of natural kinds as aspects of a single question. Magnus, along with proponents of other epistemic accounts rejects this thesis, in favour of the claim that epistemic and ontological questions with respect to natural kinds can and ought to be distinguished:
Epistemic Question: what distinguishes a category which is natural from an arbitrary class? Ontology Question: what feature in the world allows natural categories to perform this function?
Epistemic approaches thus take the provision of an answer to the first, and more general, taxonomic question as the primary of objective of philosophical theories of natural kinds. Inspired by Boyd, Magnus provides an answer to the taxonomic [from here on the ‘epistemic question’] question which he calls pragmatic naturalism:
Pragmatic naturalism: A category k is a natural kind for domain d if the following conditions are met: (1)[The success clause]: k is a part of a taxonomy that allows scientific enquiry into d to achieve inductive and explanatory success and (2) [The restriction clause]: any taxonomy that excluded k would not do so. (PM)
This theory is taken to be motivated by the adoption of three independently plausible theses or assumptions, which Magnus refers to as ’induction’ ‘science’ and ‘domain relativity. According to these claims, natural kinds are essentially those categories which (i) support and feature in successful exploratory and inductive inferences, (ii) form components of a successful scientific taxonomy [that is are necessarily scientific categories] and (iii) are, crucially, always relative to a specific domain of enquiry. This latter claim is the most controversial. Found also in Boyd (1999) and Quine (1959) the claim contained in (iii) constitutes a denial of the so-called ‘simpliciter assumption’ viz. the claim that there is a fact of the matter whether or not (for any natural kind candidate X) X is a natural kind (Goodman 2014, Bird and Tobin 2015,). Instead, the PM view is that natural kinds are always domain relative: natural kindness is not a one place predicate’ (‘electron is a natural kind’) but is rather a relational notion which relates natural kinds to specific domains of scientific enquiry (‘electron is a natural kind for the domain of particle physics’):
‘‘even though the kind predator does no work in the high-energy or small-scale domains where top quarks matter, top quark does no work in the biological or ecological domains in which predator is a natural kind. Neither has a claim to absolute importance, but they are important in relation to different contexts and phenomena. The very same kind top quark is a natural kind in the domain of particle physics but not in the domain of ecology; and vice versa for the kind predator’’ (Magnus; in draft;3, 2012).
Magnus motivates and defends a pluralistic view with respect to the ontological question.Whilst there are, according to the pragmatic naturalist, explanatory features which are unique to - and shared by each and every - natural kind, Magnus argues that this needn’t be the same for the ontological account of natural kind the naturalist provides. We can simultaneously hold that natural kinds form a unified phenomenon, whilst also endorsing the view that the ontological nature and constitution of natural kinds in each case can be very different.
In short, pragmatic naturalism offers us a new theory of natural kinds with which to approach and characterise the project of naturalistic epistemology. In the interests of brevity, I’ll save the details of the new naturalistic epistemology which this gives rise to - which - spoiler - I believe, is both knowledge-first and contextualist - and its relation to Kornblith’s version of naturalistic epistemology, for my next post. This involves the claim that knowledge is a natural kind relative to certain domains, a claim which is compatible with distinct ontological claims about knowledge’s underlying nature. In my view, this captures all the ‘fundamentality’ and 'theoretical priority' a knowledge-firster could - and should, if they are naturalists -want. Crucially, I will argue, this theory can explain why the metaphysical claims (M2-5) should be endorsed together .