Concepts of Natural Kinds
According to the standard model, concepts like water and heat are natural kind concepts subject to Twin Earth cases: they are not “in the head”, they have an “indexical component”, they function as mere “labels”. Alongside these concepts we have another range of “theoretical” concepts, such as H2O and molecular motion: these concepts pick out the same natural kinds as the former concepts, but they are concepts of a different type. They express the “real essence” of the natural kind, what it is intrinsically and necessarily. They are not mere labels, they are not indexical, and they are not subject to Twin Earth cases. Thus, if one of our counterparts on Twin Earth uses the term “H2O”, he refers to H2O (water), even if the water-like liquid on Twin Earth is not H2O but XYZ (so “water” on Twin Earth refers to XYZ while “H2O” refers to H2O). Twin Earth contains both hydrogen and oxygen and speakers refer to these elements with “hydrogen” and “oxygen” (we may suppose), so their phrase means the same as ours, which entails that their words “water” and “H2O” do not refer to the same natural kind. This suggests a general thesis: for any concept like water there is (or could be) a corresponding concept like H2O. For any concept C subject to Twin Earth cases there is a concept C* not subject to Twin Earth cases such that C* refers to the same natural kind as C. If this thesis is correct, then our basic concepts of reality are not natural kind concepts, as conceived by the standard model. Instead, they are analogous to the concepts H2O and molecular motion: descriptive, internal to the conceiving subject, and invariant between Earth and Twin Earth. Concepts of the real essence of natural kinds are not themselves natural kind concepts, as conceived by the standard model.
An objection may be raised to the thesis: Can’t we create Twin Earth cases for “oxygen” and “hydrogen”? What if Twin Earth does not contain these elements, though it does contain elements that are superficially similar to them: then won’t speakers on Twin Earth mean those elements not the ones we mean with the same terms? So the corresponding concepts will differ on Earth and Twin Earth, thus conforming to the standard model. If so, the concepts corresponding to the constituents of “H2O” will be natural kind concepts, as the standard model conceives them. But now there is a question about whether those concepts fit the thesis we are considering: won’t there be a corresponding concept for oxygen that fails to fit the standard model? Thus we can say that oxygen is the chemical element with atomic number 8, and now the question is whether that concept fits the standard model. It is hard to see how it could, since speakers on Twin Earth will have to mean the same thing we do by “chemical element with atomic number 8” (remember they are internally identical to us). And if that is so, then water will have a corresponding concept that is not itself of the type that water is—not a natural kind concept (though a concept of a natural kind).
Still, someone might object that even this highly theoretical concept contains constituents that can be made subject to the Twin Earth treatment—say, the concept atom or the concept electron. Couldn’t Twin Earth be bathed, not in electrons, but in some superficially similar kind of particle, indistinguishable from electrons by the inhabitants of Twin Earth? Then won’t they refer to that type of particle not electrons with “electron”? This is now becoming farfetched and difficult to articulate, not like the original intuitive story about “water”: how could there be particles that are superficially just like electrons but are not electrons? But even supposing such a thing to be possible, we can ask about other expressions for electrons—say, “negatively charged particle”. This concept is surely not subject to Twin Earth cases! In the end the chain of terms denoting natural kinds bottoms out in a description that contains no terms that are subject to Twin Earth cases. So reality is not ultimately conceivable onlyby means of concepts that fit the standard model. Natural kind concepts of the type exemplified by water and heatare not the ultimate way we conceive of reality—just as the initial cases suggested. When we conceive of a liquid as water we mentally represent it in a way that differs structurally from the way it is mentally represented by the concept H2O; and the latter type of concept is what is fundamental to representing reality. Natural kind concepts of the type expressed in the vernacular are really a dispensable substitute for more basic “theoretical” concepts. The possibility of Twin Earth cases is not universal to concepts for natural kinds. At the most basic level, our concepts of natural kinds are not natural kind concepts. For the basic concepts of natural kinds, meaning (content) is “in the head”, even if it is not for their vernacular counterparts.