Natural Kinds: (Thick) Essentialism or Promiscuous Realism? Theoretical identity statements of the form water is H 2 O are allegedly necessary truths knowable a posteriori, and assert that nothing could be water and not be H 2 O. The necessary a posteriori nature of these identity claims have been taken to justify a move from talk of reference (language) to talk of essence (metaphysics). In this paper I will contest this conclusion, and argue that i) the only way to derive essentialism from semantics (specifically direct reference) is to assume it as a premise (Salmon 1982), ii) that contemporary essentialism is a metaphysical assumption not a thesis, and iii) assuming the accuracy of the analogy between proper names and natural kind terms present an alternative version of natural kind classification that is interest relative. I will also, for the sake of this paper, be using the term essentialism to refer to the substantive metaphysical thesis that H 2 O is the very nature of water (as a matter of primordial metaphysical fact), as opposed to the more innocuous thesis that H 2 O is the nature of water (given our world, and our conceptual schema) 1. The (allegedly) necessarily true identity statements, knowable a posteriori, that I am interested in are those concerning natural kinds. Hilary Putnam and Keith Donnellan are key exponents of the necessary a posteriori for natural kinds, and although both have been shown to be question begging (see Salmon 1982) Putnam even changed his view on this issue sometime during the early 80s (Putnam 1982) I will present their view as I take it to be the inspiration driving contemporary essentialism (Ellis 2002/2004, Lowe 2006, Simons 2006, Bird forthcoming). The example that has become indicative of the view I am attacking is the Twin Earth style thought experiment. The general structure of the thought experiment, briefly, is: we discover on Earth that the microstructure of water is H 2 O, and compare it to a substance on a distant planet called Twin Earth which has a macroscopically, functionally identical substance, also called water, but with the chemical composition XYZ. Although native speakers of both worlds call their chemical substance water, and, if suddenly switched, would not know the difference between the substances, given that what we call water is H 2 O, we will not have discovered a possible world in which water is XYZ but merely a possible world in which there are lakes of XYZ, [and] people drink XYZ (and not water) (Putnam 1975, p. 232/3). The upshot of this (allegedly) intuitive thought experiment is that while it is conceptually possible that water isn t H 2 O, it is conceivable but it isn t possible! (ibid) 1 The distinction is between thick and thin essentialism, where thin essentialism concerns unproblematic statements about the essence of substances here, plus various factors about practical interest and the state of scientific investigation, and thick essentialism to be the deep metaphysical claims about the very nature of substances made by Lowe (2006), Simons (2006) and Ellis (2002/2004) that construes essence as a deep and (marvellous) metaphysical discovery about the nature of an object. 1
The Twin Earth thought experiment is intended to demonstrate the a posteriori necessity of natural kind identity statements such as water is H 2 O via three premises: first, the ostensive definition supplied by the theory of direct reference, second, an empirical claim about the hidden nature of the paradigm (Salmon 1982, p. 163), and third, what being an instance of the same kind consists in. So, to elaborate: in the case of something like an element, given the ostensive definition of gold (premise 1) as that stuff, to generate the necessary a posteriori truth that gold is the element with atomic number 79, we need to know (or at least an expert needs to know, given Putnam s division of linguistic labour) that the paradigm has the atomic number 79 (premise 2), and that to be consubstantial consists in two (or more) samples sharing the same atomic structure (premise 3). The view Kripke advocates, which is slightly different, is that we take the objects we do have in this world, and which we can identify, and then ask whether certain things might have been true of these objects (Kripke 1980, p. 53), in counterfactual situations. Exegetically Kripke may only be committed to premises 1 and 2, but his notion of transworld identification, which always concerns this object, applied to kinds, carries an implicit consists in relation. According to Kripke substance samples are members of the kind in virtue of certain identifying marks (Kripke 1980, p. 118). These marks can, in some cases, form the basis of kind identity, but often we discovered that certain properties were true of [X] in addition to the initial identifying marks by which we identified it (Kripke 1980, p. 119), and it is these properties that Kripke claims differentiate one kind for another. So, in the case of gold, present scientific theory is such that it is part of the nature of gold as we have it to be an element with atomic number 79 (Kripke 1980, p. 125, my italics). Thus it turns out that necessarily gold is the element with atomic number 79, because in this world that is what gold turned out to be 2. Kripke s claim about gold is certainly underpinned by something like premise 3, where, mutatis mutandis, membership of a kind consists in sharing an important characteristic with a paradigmatic instance, namely its atomic structure, and this is the metaphysical necessity that, traditionally, was only associated with analyticity and a prioriticity. The major problem with this thesis is that premise 3 assumes that for two elements to be the same they have to be consubstantial, and that consubstantiality consists in having identical chemical structure (Salmon 1982, p. 176). So, consubstantiality is a necessary condition of identity, at least for chemical kinds, and thus the essence of a chemical element is its chemical structure. While, as an intuition, this may or may not be something people subscribe to, the point is simply 2 It seems to me that here, again, Kripke is vague, and that the metaphysical status of identity statements should be qualified. Option 1 (the sensible option in my opinion) is that gold is the element with atomic number 79 in this world because of an accidental property of this world, but given that it is a property of this world when we refer to gold we refer to and only to the substance with the atomic number 79. The other option is far more metaphysically significant and is that the discovery of gold s atomic number is a deep, primordial discovery about the nature of reality. 2
that there is an essentialist premise within the argument for necessary a posteriori truths, and as such the necessary a posteriori cannot ground essentialism as this would be circular. Putnam s thesis rests on the following claim: If there is a hidden structure, then generally it determines what it is to be a member of the natural kind, not only in the actual world, but in all possible worlds. Put another way, it determines what we can and cannot counterfactually suppose about the natural kind. (Putnam 1975, p. 241) Apparently, there is a hidden structure that determines whether or not an instance of an object is a member of a given natural kind, because membership of the kind is dependent on having the aforementioned hidden structure. But this is an unsubstantiated essentialist assumption, and one that, although natural, should be resisted. First, we have no reason to believe that our intuitions are accurate reflections of the metaphysical structure of reality, they may be mistaken, naïve and based on a lack of evidence. Second, take Putnam s Twin Earth example: to echo a thought from Joseph LaPorte (2004), it is so fantastical that we cannot have practical, well founded intuitions because of the very nature of the example. LaPorte s real world analogous case of jade provides a paradigmatic natural kind counterexample, where two hidden-structurally-distinct elements (jadeite and nephrite) both count as members of the kind: jadeite has the chemical composition NaAlsi 2 O 6, whereas nephrite has the chemical composition Ca 2 (Mg, Fe) 5 Si 8 O 22 (OH) 2, but both qualify as jade, strictly speaking. This is an important (albeit contentious) example for at least two reasons, it demonstrates first that hidden structures do not necessarily determine membership of a given kind, and second the kind of reaction we might have were we ever to encounter a real Twin Earth example e.g. we could subsume both hidden-structurally-distinct entities under a common (macro) kind 3. If this is right, then our reactions to real Twin Earth style occurrences are motivated by practical considerations, which may or may not coincide with atomic structure. Take another example, water molecules are made up of Hydrogen and Oxygen atoms, both of which have naturally occurring isotopes. Hydrogen has three, Protium ( 1 H) and Deuterium ( 2 H or D) and Tritium ( 3 H or T), whereas Oxygen has 17, only three of which are stable 16 O, 17 O and 18 O. These atomic isotopes (naturally) bond to form water molecules as found in lakes and rivers, some more abundant than others. Strictly speaking the natural kind term water has a number of disjunctive members depending on what level of reduction we apply: H 2 O and D 2 O; H 16 2 O, H 17 2 O and 3 This discussion seems to be one of reductionism, and specifically what level of reduction is appropriate for kind identification. However, there is, I think, a deeper issue about whether all property differences are indicative of different kinds, even at the microscopic level: for example, if H 2O had been bonded in a different fashion (instead of a polar covalent bond) then it would have had markedly different properties, thus there could be H 2O which was so radically different as to be unrecognisable, and the question is how does that effect our kind classification? 3
H 18 2 O, plus the same for Deuterium and Tritium. This being the case Putnam s consubstantiality relation having an identical chemical structure is not 100% accurate, even in this world. Imagine a further case, where we make it to Twin Earth, and, via special spectacles, are able to perceive (visually) differences in chemical composition. Thus we actually see that what we call water is H 2 O, and that what Twin Earthers call water is XYZ; it appears in this case that the is in the water is H 2 O and water is XYZ has become an is of predication, not an is of identity. But why should the property being predicated of the kind, in the water is H 2 O case, be more significant than, say, the identical macro properties of the two substance? Granted there may be some property dependence relations, but the question is what intuitions should we have if we perceived chemical compositions in the same fashion as, say, transparency? To my mind there is no clear answer. A stronger case yet might be the following adaptation of the Twin Earth thought experiment. Imagine that we, as in the previous case, make it to Twin Earth, and find that while everything seems to be macroscopically identical, microscopically everything that exists on Twin Earth is different (perhaps the people on Twin Earth are not even carbon based life forms). Let us also assume, for the sake of argument, that the linguistic practices are identical, and they call XYZ water, just as we call H 2 O water, and so on mutandis mutatis for all other occurent substances on Twin Earth. Thus, for every substance on Earth there is a counterpart substance on Twin Earth, carrying the same name and ostensive properties, but a different microstructure. Are they samples of the same kind or not? Again, there is no clear answer. Prima facie nothing precludes disjunctive kinds indeed in this example we accept a race with wildly different neurochemistry, but have little problem attributing beliefs and language to them and if for every case of water is H 2 O there is a corresponding case where water is XYZ, disjunctive kinds seem intuitive. Kripke, however, argues in the case of gold, and by inference water, that: One should not say that it would still be gold in this possible world, though gold would then lack the atomic number 79. It would be some other stuff, some other substance. (Once again, whether people counterfactually called it gold is irrelevant. We do not describe it as gold.) (Kripke 1980, p. 124) In the above quote Kripke is subscribing to, or at least endorsing, metaphysical essentialism: the view that substance identity consists in sharing an identical atomic structure; indeed he states that it is part of the very nature (Kripke 1980, p. 124) of gold that it has atomic number 79, and this 4
can be read as a substantial metaphysical claim about the structure of reality 4. If this is correct then Kripke, like Putnam, is question begging with his essentialism. My alternative picture is a version of promiscuous realism (Dupré 1995), that acknowledges the difficulty of assessing what would happen if we were to come across critical examples where new chemical elements, exhibiting identical manifest behaviours and properties to an element on our periodic table, but possessing a unique atomic composition, were to appear. The type of claim I contest is exemplified by necessarily water is H 2 O, where the necessity is metaphysical (e.g. across all possible words) and I contest it on the grounds that although in this world there is a substance we call water, and samples of water here only qualify as water if they have the chemical composition H 2 O 5, any metaphysical assertion to the effect that i) atomic structure is the nature, or essence, of an object, and therefore a necessary feature of its identity, and ii) substances can, and should, be differentiated in virtue of the atoms that constitute their atomic structure, is clearly an essentialist assumption. Essentialist intuitions may be fairly standard, but they are a by-product of a metaphysical assumption that underpins our current scientific investigation, one which a) may or may not turn out to be true, b) should be open to (potential) revision under the relevant circumstances (perhaps a Kuhnian revolution of some sort), and c) is, at best, a statement of fact about how this world happens to be as opposed to some statement of primordial metaphysical fact, that holds across all possible worlds. As far as I am concerned this world, call it world 1, might have turned out to be that world, call it world 2 or Twin Earth in which the chemical composition of water was radically different 6. Current scientific endeavour is underpinned by the metaphysical paradigm (in the Kuhnian sense) atomism, where elements, minerals, and chemicals can be differentiated via their atomic structures and chemical compositions, and this paradigm holds because there are few counterexamples, and it is an incredibly useful system of classification. But this system of classification would not be so useful in the face of an extreme version of the Twin Earth thought experiment explicated earlier. Faced with overwhelming empirical evidence to the contrary it would be illogical to hold that natural kinds are differentiated by their chemical composition if every element had an analogous atomic-compositionally divergent counterpart. If essentialism derived 4 There are other issues with Kripke, specifically his analogy between the Hesperus is Phosphorus case and the water is H 2O case, which seems to imply that he considers both to be instances of names, and the identity relation that exists between them is a semantic one. But in the water is H 2O case, H 2O is playing quite a different function, or at least is of a very different form to the term water. It looks, at first blush, very much like a definite description, which would be decidedly Fregean, and rather anti-direct reference, or it is a proper name, and should be presented as something like Aitch-two-owe (although I have has this thought independently I acknowledge E.J. Lowe s [unpublished] paper Essentialism and Chemical Kinds, where he made a similar point, albeit for different purposes), and should not therefore be the centre of all this controversy surrounding essentialism. 5 Although even this has been shown not to be, strictly speaking, true. 6 Again, this thought is one I have had independently, but, as I understand, is a classic two dimensionalist intuition, but I have yet to explore this avenue of literature fully. However, leading two dimensionalists include Frank Jackson and David Chalmers (2002), with Scott Soames (2004) as an obvious critic of the 2D project. 5
from semantics is question begging then Twin Earth style thought experiments should have the opposite effect to that which Putnam intended rather than vindicating essentialism they demonstrate the need to judge on a case by case basis that is dependent on our practical interests and requirements. Our method of classifying the elements that occur (and even some which don t) in this world should not be construed as necessarily true a posteriori, where the necessity is metaphysical necessity. Rather they should be interpreted as true relative to a metaphysical paradigm, true relative to the way the world turned out to be, and true relative to the interests of the observer. Essentialism, if sound, should be far more conservative: natural kind classifications depend, in part, on our interests, and those interests (as far as chemical kinds go) are dictated by the scientific community, the limits of our research capabilities, and the types of elements that happen to exist in our world. We need not necessarily convert to anti-essentialism (particularly as I don t think my examples constitute arguments), but what we should consider is that any essentialism we do subscribe to is interest relative, and although that interest may be species wide it is still only relative to this world. 6
Bibliography Donnellan, Keith. (1977). Speaking of Nothing in Naming, Necessity and Natural Kinds edited by S. Schwartz. Cornell University Press. Dupré, John. (1995). The Disorder of Things. Harvard University Press Kripke, Saul. (1980). Naming and Necessity. Harvard University Press. Kuhn, Thomas. (1996). The Structure of Scientific Revolutions. University of Chicago Press, (3 rd ed.) LaPorte, Joseph. (2004). Natural Kinds and Conceptual Change. Cambridge University Press Mellor, D. H. (1977). Natural Kinds in British Journal for the Philosophy of Science 28, pp 299 312 Putnam, Hilary. (1973). Meaning and Reference in The Journal of Philosophy, Vol. 70, No.19, 699-711. Nov. 8. Putnam, Hilary. (1975). The Meaning of Meaning in Keith Gunderson, editor. Minnesota Studies in the Philosophy of Science, vol. VII. Minneapolis: University of Minnesota Press Putnam, Hilary. (1982). Why there isn t a ready-made world in Synthese, Vol. 51, No. 2, May Quine, W. V. (1969). Natural Kinds in Ontological Reality and Other Essays. Columbia University Press Salmon, Nathan. (1982). Reference and Essence. Basil Blackwell. Soames, S. (2002). Beyond Rigidity: the unfinished semantic agenda of Naming and Necessity. Oxford University Press, 7