Natural Kinds Without Essentialism

Many new dispositionalists (Ellis, Shoemaker, Bird and Molnar, among others) adopt essentialist positions, accepting the existence of causal powers, natural kinds and essences. There is even a tendency to draw parallels between ‘the new dis- positionalism' and ‘the new essentialism'.

Their main argument is that causal properties are essential, intrinsic, and that things are what they are, or of the kind they are, by virtue of these properties. Properties determine natural kinds, which are constituted and characterised by the essence of intrinsic properties. However, many authors recognise, as discussed above, that these essences are not determined intrinsically, but rather by their interactions with other properties.

The members of natural kinds are instances of properties and their essences. Kind membership is determined by essences: ‘nothing could acquire any set of kind-identifying properties without becoming a thing of this kind' (Ellis 2001, 237­238). Particulars are instances of real essences of intrinsic properties, which con­stitute natural kinds.

New essentialism is opposite to the categorical approach, which holds that natural properties are categorical rather than dispositional. The same property can exist in another possible world (Armstrong 1997). New essentialism does not invoke the Kripke-Putnam theory, in which direct reference is extended to natural kinds. However, as Mumford (2005, 428) points out, ‘Ellis's essentialism is of the Kripkean variety'. Mumford accepts N. Salmon's (1982) critique of this theory, namely, that it does not present a conclusive argument for essentialism about natural kinds.

The question to explore below is whether it is possible to accept the existence of dispositions, causal powers and natural kinds and reject essentialism. The central issue is whether we can accept the existence of natural kinds without accepting essentialism.

What theory on natural kinds allows us to avoid essentialism? Mumford (2005) offers an interesting argument. He develops an account of natural kinds that con­siders anti-essentialist. However, some aspects of his approach, one could argue, contain echoes of essentialism. This would be related to his approach on universals. Mumford holds that universals are real,

as are particulars: ‘Universals and particulars are distinct and both real” (Mumford 2005, 433). However, immediately after this statement he affirms that a universal exists‘only in its instances’, not in some Platonic universe:‘There is not some transcendent form that exists besides’ (Mumford 2005, 433).

Universals exist immanently in the particulars that carry them. This is so in the case of properties, which he defines in relation to natural kinds as ‘a universal but accidental characterising attribute [property]’ (Mumford 2005, 434). Natural kinds are ‘characterised by their attributes [properties]’ (Mumford 2005, 434). They are neither intrinsic nor essential, but they are universal. Therefore, the universality of natural kinds, which is affirmed by Mumford, is based on properties, which exist immanently in the totality of the particulars, (and each of them), that make up that kind. It is their shared property which makes the particulars are similar each other and constitute a universal natural kind. However, argues Mumford (2005, 435) These kinds can be characterised by attributes [properties] but a further assumption that any of these characterising attributes is essential is neither required nor inde­pendently motivated’.

The types are universal, while modes are particular. Modes belong to objects, and ‘The basic things that exist are objects-bearing-modes’ (Mumford 2005, 433). But, modes, what are modes of? They are modes of types, of course, and types are simply properties considered by themselves, rather than in their instances, their modes. Mumford sums this up nicely when he says that ‘Attributes (properties) are, therefore, types of modes-borne-by-objects’ (Mumford 2005 433). Properties (or attributes), redness for example, are simply ‘natural types of modes’, those that are borne by objects. But, what is that, borne by objects, making them constitute a natural kind? The answer here is redness, that which makes the property to be of that type.

So what makes a property to be of a certain type? One response could be ‘what the property is’, that which exists immanently in the objects and constitutes their modes, -of which the property is the type: ‘redness’, ‘solubility’, ‘fragility’. But what is redness? What is fragility? What is solubility? Applying Mumford’s argument, one could say that these are all ‘natural types’ existing beyond of, their modes (despite Mumford’s insistence that they are ‘types of modes’); or at least existing parallel to their modes. If this is the case, it raises the question of what extent it does not involve essentialist reminiscences. Mumford rejects this idea but one could contend that his arguments do seem to invoke it. We must not forget that types are universals real, existing, according to Mumford.^[153]

At this point we can ask: Is it possible to accept natural kinds and reject even the smallest hints of essentialism? The answer I propose below is in line with Mum­ford’s view for the most part, although without accepting the existence of real universals.

It could be said that properties are types, but this is merely a linguistic and conceptual resource that helps to classify and differentiate properties in order to define or characterise them. Therefore, I agree with Mumford (2005) that natural kinds are constituted’ by the omnitemporal totality of their member objects or particulars’ (Mumford 2005, 434) and that they can be defined or ‘characterised by their attributes’ (Mumford 2005, 434). But this is simply a perfectly valid linguistic and conceptual resource.

Thus, ‘natural types’ are not something of the world (real) but rather of language. What is of the world are the sets of objects with shared properties, dispositions and powers which constitute natural kinds; and to constitute a natural kind all that is needed is ‘resemblance’.^[154] Natural kinds are constituted by entities that share a property or properties (being soluble, fragile, irritable) and therefore certain dis­positions or powers.

I agree with Chakravartty (2008) that one can expect members of a kind that share properties to behave similarly in similar circumstances. This allows us to make generalisations, including many different types of laws, although as Chak- ravartty reminds us: ‘In many cases, however, causal generalisations are susceptible to exceptions and ceteris paribus qualifications’ (Chakravartty 2008, 160). The behaviour of the members of a kind depends on their causal powers, not on any essence. What matters is the possession of a causally efficacious property—a power or property that confers a power.^[155]

In my approach, natural kinds are constituted by sets of objects with certain shared properties and dispositions or powers; essences play no part here. This is a departure from Plato but not from Aristotle, because as Chakravartty (2008) argues, the concepts of powers and essences are not inextricably linked, contrary to what some Aristotelians believe. Therefore, causal generalisations are not determined by the essential properties of kinds of entities (objects). The behaviour of entities is determined by their causally efficacious dispositional properties.

7