ARTIFICIAL AND NATURAL KIND TERMS
One of the primary purposes of language is to pick out things in the world. Most of the time language users accomplish this task by naming those things. For example, when a community wants to identify a particular substance that just happens to be important for irrigating crops and drinking, they name that substance water. When the named thing has a natural essence, or an essence constituted by a common structure, these names become natural kind terms. Natural kind terms contrast with artificial kind terms, because people view the defined common structure of an artificial kind term as arbitrary.
Despite its reliance on examples that seem completely benign and natural, the concept of natural kind terms is controversial. Various theories explain and detail the rules for defining natural kind terms. This essay will focus on the most widely accepted theory of natural kind terms, as argued by Kripke and Putnam. Despite its wide acceptance, the Kripke-Putnam Theory still attracts many critics. For example, Mellor advocates a rejection of the Kripke-Putnam Theory based on the argument that natural kind terms are not rigid designators (Mellor defends a Fregean approach; see Mellor, 1977). In addition, Ben-Yami (2001) rejects the Kripke-Putnam Theory because one cannot found the essence of natural kinds on structural properties.
Mellor and Ben-Yami argue that we should reject the Kripke-Putnam Theory, but neither advocates an outright rejection of the concept of natural kind terms. I will argue that there are far fewer natural kind terms than Kripke and Putnam would enumerate. I will conclude that most known kind terms are artificial and that only elementary particles are natural kinds. In order to accomplish this, I will review Kripke’s and Putnam’s theory of natural kind terms and explain why we should dispense with the prevailing distinction between natural and artificial kind terms. In its place, characterizing most kind terms as artificial kind terms will function without succumbing to the objections against the traditional distinction between natural and artificial kind terms.
Kripke’s philosophy of language extends far beyond the scope of natural kind terms. His metaphysics and the concept of the rigid designator are important aspects that lie at the foundation of natural kind terms and essentialism. I will briefly review Kripke’s metaphysics and theory of rigidity. This will reveal the framework essential for his theory of natural kind terms.
There are several aspects of Kripke’s metaphysics that are important to the subject of natural kind terms. Kripke advocates the existence of structural properties, notably aspects that things will have as a “kind of thing” (Kripke, 1980, p 122). These properties will be important when Kripke moves from naming one particular thing (e.g., ‘this is water’) to using that particular as a “paradigmatic instance” (Kripke, 1980, p. 122) of a natural kind. For Kripke, a given natural kind constitutes a set of structural properties that a particular must possess in order for the natural kind term to fit that particular. For example, assume that tigers are a natural kind and one of the essential properties of tigers is stripedness. An animal that does not possess the property of ‘striped’ is not a tiger.
Kripke rejects the existence of bare particulars (a notion that particulars have no properties; 1980, p. 52), indicating that nothing stands as a substratum behind a particular’s properties. This rejection of bare particulars entails that names will not be able to refer to an object without also referring to at least some of its properties. Kripke rejects the notion that a particular thing is only a bundle of properties (a notion that a particular is constituted solely by the aggregate of its properties; 1980, p. 52). The rejection of both the bare particular view and the bundle-of-properties view means that referring to a particular thing will require a name to refer to both an object and some of its properties.
Kripke separates epistemological and metaphysical claims. The identification of natural kind terms will pick out paradigmatic instances, and some of the properties of the paradigmatic instances will be essential to the natural kind term. This essentialism of properties is metaphysical, not epistemological, which means that an agent may fail to know the set of properties that are essential to a natural kind. A person may use the term ‘water’ to identify a pool of water, and lack a complete or even correct awareness of water’s properties. Fortunately, it is possible that a language knows these metaphysical truths, but knowing a natural kind’s properties and knowing what natural kinds are will be an a posteriori endeavor. (Kripke, 1980)
In order to introduce Kripke’s concept of rigidity properly, I must first explain Kripke’s view of possible worlds. One competing view is that possible worlds are actual self-contained universes that have whole histories and exist independently of our world. This view is championed by David Lewis. In contrast, Kripke rejects Lewis’ theory, stating that possible worlds are not a distant place that we can observe through a telescope or some other set of technology. Kripke argues that we should not interpret possible worlds as foreign countries that one could view through an apparatus, but as stories that philosophers stipulate. Thus, for Kripke, possible worlds are derivative of the world that we live in. (Kripke, 1980)
Kripke’s view of possible worlds is important to the concept of rigidity. In the case of counterfactuals, we can stipulate some things to be different, and the things that remain unstipulated will remain identical to the actual world. Kripke argues that proper names always refer to the same object, regardless of whether we use them in the normal sense or in counterfactual conditionals. When we consider what would have happened to Nixon if he never became President, the name ‘Nixon’ refers to the same Nixon, only with a different stipulated property. It is possible that Nixon could have not been President, but it is impossible that the proper name ‘Nixon’ fails to refer to Nixon. According to Kripke, natural kind terms are rigid: they refer to the same kind of thing in all possible worlds. (Kripke, 1980). More specifically, a natural kind refers to the same paradigmatic instance in all possible worlds.
Natural kind terms rely on rigid paradigmatic instances. These paradigmatic instances will provide a number of structural properties that will determine whether other objects are of the same kind. Thus, if a particular possesses all the essential structural properties of a natural kind, then that particular belongs to that natural kind. A paradigmatic instance of a tiger will provide a number of properties that another thing must have in order to be a tiger. If an animal fails to have the essential properties of a tiger, then the animal is not a tiger. However, these internal properties only define whether an object is a specific kind—it is not necessarily the case that a language user will use these properties to identify whether a particular object is a specific natural kind. For Kripke, it is more important to understand the nature of what makes a natural kind than to ensure that language users are right when they differentiate things that are of a certain natural kind and things that are not a certain natural kind. (Kripke, 1980)
Putnam also argues for the existence of natural kind terms. In Meaning and Reference, Putnam argues that water is a natural kind term determined by an initial indexical reference. In addition, Putnam argues that cooperation among individuals allows society to divide the linguistic labor between sociological groups.
Putnam’s main argument is, “The traditional concept of meaning is a concept which rests on a false theory” (Putnam, 2001, p. 289). We can trace Putnam’s discomfort to one of the assumptions that Putnam attaches to the traditional concept of intension/extension:
(2) That the meaning of a term determines its extension (in the sense that sameness of intension entails sameness of extension). (Putnam, 2001, p. 289)
Instead of viewing extension as an effect of intension, Putnam argues that a word’s extension determines its intention. Putnam uses his famous Twin Earth experiment to show how this is the case.
Consider Putnam’s Twin Earth example. Twin Earthians are exact duplicates of Earthians (except perhaps that the “water” in their bodies has a different molecular structure); water on Earth is H2O, and a different sort of water on Twin Earth is XYZ (a shortened formula of a complex chemical). Putnam stipulates that in 1750 the Doppelgangers on Twin Earth have the same psychological states as their respective twins on Earth. This stipulation forces the mental criteria of water to be the same on Earth and Twin Earth. However, due to the differences in molecular chemistry, water on Earth is very different from the water on Twin Earth, except for the fact that the two versions of water have the same superficial characteristics. H2O boils at the same temperature as XYZ—they also have the same freezing point. In fact, without equipment for molecular testing, Earthians and Twin Earthians have no way to discriminate between H2O and XYZ.
Putnam describes through an example of aluminum and molybdenum that human society frequently commissions ‘experts’ to cooperatively define the meaning of a term and to prescribe that meaning to those individuals who are not ‘experts.’ This notion of linguistic outsourcing coheres with Putnam’s formal hypothesis:
Hypothesis of the universality of the division of linguistic labor: Every linguistic community exemplifies the sort of division of linguistic labor just described; that is, it possesses at least some terms whose associated “criteria” are known only to a subset of the speakers who acquire the terms, and whose use by the other speakers depends upon a structured cooperation between them and the speakers in the relevant subsets. (Putnam, 2001, p. 292)
There are several important facets of this hypothesis. The hypothesis defines two subsets of a linguistic community: a subset of speakers “who acquire the terms,” and a subset of speakers who depend “upon a structured cooperation between them and the speakers in the relevant subsets.” This second subset refers to people who depend on the analysis of experts in order to create criteria for the meaning of terms.
Putnam denies that a pattern of social cooperation initially determines the meaning of a natural kind term. Since extension determines meaning, the Twin Earth water example features no division of linguistic labor. There were no experts in the Twin Earth example in 1750. Putnam’s analysis involves a theory of reference, rather than a theory of the universality of the division of linguistic labor. Putnam presents two hypotheses, rejects the first, and affirms the second. The first hypothetical theory of reference is this:
(1) One might hold that ‘water’ was world-relative but constant in meaning (i.e., the word has a constant relative meaning). On this theory, ‘water’ means the same in W1 and W2; it’s just that water is H2O in W1, and water is XYZ in W2. (Putnam, 2001, p. 292)
This hypothetical theory denies Putnam’s assumption that “two terms cannot differ in extension and have the same intension” (2001, p. 288). Putnam wants terms to have a complete meaning and have a one-to-one relation with the world, so he prefers the second hypothesis:
(2) One might hold that water is H2O in all worlds (the stuff called “water” in W2 isn’t water), but ‘water’ doesn’t have the same meaning in W1 and W2. (Putnam, 2001, p. 292)
If we agree with the conclusions that Putnam makes from his examples and his claim that “two terms cannot differ in extension and have the same intension” (2001, p. 288), then water a natural kind term and a rigid designator. Natural kind terms will then consist of a set of essential structural properties, such as H2O, which will hold in all possible worlds.
CRITIQUE OF NATURAL KINDS
In response to Kripke and Putnam’s analysis of natural kind terms, Mellor advances several arguments against the idea of natural kind terms. Mellor analyzes Putnam’s thought experiment of Twin Earth and concludes that Earthians and Twin Earthians both refer to water.
Mellor points out that Putnam assumes that the microstructure of water on Earth and Twin Earth are a basis to separate them into two different natural kinds. Mellor challenges this assumption, stating:
There was water on both planets alike, and there still is. We simply discovered that not all water has the same microstructure; why should it? Because its microstructure is an essential property of water? Well, that is what’s in question. (Mellor, 1977, p. 303)
Mellor justifies this challenge by specifying scientific examples. Mellor describes various heavy waters and chlorine isotopes that have varying microstructures, yet chemists in our world still refer to all chlorine isotopes as chlorine and all densities of water as water. Putnam’s only available answer is that “chlorine and water have been found not to be natural kinds after all, but rather mixtures of natural kinds” (Mellor, 1977, p. 303). According to Zymach (as cited by Mellor, 1977, p. 303), this response may entail that there are no natural kind terms at all. As we delve further into the microstructure of natural kinds, scientists will discover fine distinctions among the particular molecules that we were previously unable to detect. If those differences are significant, then our scientific discoveries will reveal that the supposed natural kind terms were merely rough-and-ready properties. If we evaluate those differences as insignificant, then we effectively trivialize these new distinctions and appeal to the molecules’ previously visible “superficial characteristics” (Putnam, 1975, p. 241, as cited by Ben-Yami, 1999, p. 169). Either way, Putnam’s Twin Earth example begins to come apart. Consider what would happen if we utilized new scientific advancements to discover that half of Earth’s water had a subtly different microstructure from the other half. If Earthians concluded that this discovery meant that half of what we considered water was not water, this creates a disturbing precedent against the existence of natural kinds. If, instead, Earthians concluded that this discovery was trivial and did not warrant a new identity of liquid, then this possibility argues against Putnam’s premise that the liquid on Twin Earth was not water. Mellor’s claim is that water on Twin Earth is water, and the specified difference in microstructure leads us to affirm that water exists both on Earth and Twin Earth.
Mellor admits that Putnam can always fall back on his theory of the division of linguistic labor. Putnam can stipulate that XYZ is not water, since his theory of the division of linguistic labor allows experts to delineate these distinctions. Putnam addresses what would happen when Earthians and Twin Earthians use space ships to experience the differences between water1 and water2. These explorers will eventually discover the differences between the two Earth’s waters, and experts on each of the two planets may eventually embrace these distinctions. However, when experts on Earth say that XYZ is not water or experts on Twin Earth say that H2O is not water, these decisions to make a difference of microstructure relevant will be arbitrary. It is equally plausible that Earthian scientists will trivialize the difference between XYZ and H2O and categorize XYZ as water. Putnam can tell a story where two civilizations think that two chemicals requires different common names, but we receive no explanation why the difference between XYZ and H2O (or any two superficially similar things) should be significant.
Putnam assumes that XYZ and H2O are a basis to divide water1 and water2 into separate categories, but he does not explain how such a division is natural. Putnam can divide XYZ and H2O into two categories by relying on a society’s linguistic labor, and this leaves us with unsatisfactory methods of discriminating natural from artificial kind terms. We understand the difference between a natural and an artificial kind term in the following way: we see natural kind terms as a way of acknowledging natural differences between the kinds specified, and we view artificial kind terms as ad hoc mechanisms that differentiate two kinds. In order to reify the distinction of water1 and water2, Putnam relies on a society’s linguistic labor to generate an arbitrary standard of differentiation, which effectively destroys the distinction between natural and artificial kind terms.
Kripke recognizes that new discoveries can disrupt our understanding of natural kind terms. For example, it is possible that a scientific discovery shows that there are two different types of gold. In this case, scientists can split gold into two types, or drop the name ‘gold’ altogether (Kripke, 1980, p. 134). This sequence of events is bad news for natural kind terms, since it seems that new scientific discoveries will deprecate rough-and-ready natural kinds, and we will eventually abandon these newer natural kind terms once future discoveries demonstrate that these new natural kind terms are inadequate. In this epistemic sequence, the essential properties that constitute natural kinds are beyond the limits of the experts’ skill.
CRITIQUE OF STRUCTURAL PROPERTIES
In order to explain why we use natural kinds to refer to aspects of nature that exist beyond our knowledge, Ben-Yami advances a different explanation of natural kind terms. Ben-Yami does not agree that natural kinds refer to essential structural properties. As the most salient examples, Ben-Yami states that “a paradigm of natural kinds are elementary particles; if anything is a natural kind, electrons, quarks and photons are” (Ben-Yami, 2001, p. 166). Electrons, quarks, photons, and other indivisible subatomic particles, according to Ben-Yami, have no apparent structural properties, since we presumably know nothing and assume nothing about the internal structure of these elemental particles to posit any structural properties. Elemental particles, as Ben-Yami says, “are elementary particles” (2001, p. 166).
This description of natural kinds is inconsistent with the Kripke-Putnam theory of natural kinds, since Kripke and Putnam require that natural kinds share a common set of structural properties, such as having a certain genetic pattern or a molecular structure or some other salient internal constitution. Ben-Yami explains that for quarks, photons, and electrons, “no unknown structure is assumed to exist, and the definition of the kind is not by reference to paradigms” (2001, p. 166). Quarks, photons, and electrons only seem to share basic properties.
Kripke and Putnam do not account for the existence of natural kinds that only share basic properties (Ben-Yami, 2001, p. 166). In fact, Kripke and Putnam rule out the possibility. In the case of elementary particles, quarks, photons, and electrons only share basic properties. Ben-Yami suggests that if elementary particles are natural kinds in virtue of basic properties, then we should disband the Kripke-Putnam rule that mandates that internal structural properties constitute natural kind terms. Ben-Yami argues that elementary particles, not structural properties, capture the essence of natural kind terms (1999). This move from structural properties to elemental particles shifts the notion of natural kind terms from a metaphysical to an epistemic notion, since we traditionally expect science to reveal that basic properties are disguised complex properties.
Although structural properties do not constitute natural kinds, they are essential for artificial kinds. Consider a structural property: water has a structure of H2O (on Earth). Scientists are aware of this type of structural property, and scientists recognize certain underlying explanations for this structural property. For example, scientists understand that a particular subatomic force, the polar covalent bond, causes H2O to maintain its stable structure. It is conceptually possible for an H2O molecule to exist without the constitution of a polar covalent bond, although scientists have not produced any cases. Therefore, while scientists admit that all observed water molecules are bound by the polar covalent bond, scientists do not consider such a bond as an essential structural property of water. Instead, scientists rely on superficial physical characteristics and ways that water reacts chemically with either different kinds of molecules or various environmental phenomena. If an experiment produced a stable H2O molecule did not exhibit a polar covalent bond but otherwise exhibits the same superficial characteristics as a water molecule would, we would not say that the outlier was not water. Scientists might say, based on future experiments, that a polar covalent bond is not a part of the essence of water. However, at this time, such a judgment would be arbitrary, since we see no natural reason why a non-polar-covalent-bound H2O is or is not water. Using structural properties will only function for characterizing artificial kinds.
If we reject the notion that natural kinds depend on structural properties and view elemental particles as constitutive of natural kinds, not only is it clearer how society utilizes natural kind terms, it is easier to see how the essence of natural kinds depends on how much a society knows. Since elementary particles are natural kinds, and elementary particles only share basic epistemic properties, structural properties differentiate artificial kinds. For the elementary particles, we have no way to differentiate one quark from another. As our scientific expertise increases, we may discover structural properties where we previously saw none, and this revision of basic properties into structural properties will eventually disqualify quarks, photons, and electrons as natural kinds. This may be why Ben-Yami says that “natural kinds have an inexhaustible number of distinctive properties, which are not logically derivable from each other” (2001, p. 181). This slow evolution of natural kinds into artificial kinds suggests that metaphysically there are no natural kinds at all. People use natural kind terms in an epistemological sense, not in a metaphysical sense. Therefore, if we want to use a kind term in a metaphysical sense, we are constrained from using natural kind terms.
In opposition to the Kipke-Putnam Theory, Ben-Yami advocates an alternative to using natural kind terms. Philosophers, according to Ben-Yami, analyze natural languages “by means of the predicate calculus” (2001, p. 176). For example, the sentence ‘All humans are mortal’ translates into predicate logic as (∀x)(Hx ⊃ Mx). This method characterizes the common noun ‘human’ as a predicate and characterizes the adjective ‘mortal’ as another predicate (Ben-Yami, 2001, pp. 176-177). Ben-Yami thinks that characterizing common nouns as predicates is a mistake.
When it comes to using common nouns, Ben-Yami suggests that we should use ‘kind terms’ instead of natural kind terms. Kind terms, Ben-Yami says, should be used only as referring expressions (2001, pp. 176-182), not as predicates. These referring expressions “refer to, ‘stand for’, or denote particulars” (Ben-Yami, 2001, p. 177), but these referring expressions do not ascribe properties to the named things. For example, “ ‘Tiger’ in ‘Tigers are dangerous’ is not a predicate, but a referring expression” (Ben-Yami, 2001, p. 177). ‘Tiger’ merely picks out the whole population of tigers. In contrast, adjectives refer to properties and are predicates. For example, “ ‘Are pretty’ and ‘runs fast’ do not refer to things, but say what properties things have” (Ben-Yami, 2001, p. 177). In summation:
The only semantic function of kind terms in their basic occurrences in natural language is to refer to particulars. The properties that we think members of a kind have are not part of the meaning of the kind term. We rely on some of these properties when we identify members of the kind to which the term applies, but the semantic function of the term is to refer to members of the kind and not to ascribe any property to them. (Ben-Yami, 2001, p. 177)
In sentences such as ‘This animal is a tiger’, Ben-Yami claims that ‘tiger’ is not an adjective. ‘Tiger’ remains a referring expression, which merely classifies the animal as one of the things referred to by the term ‘tiger’ (2001, p. 178). The sentence ‘This man is Paul’ is similar. ‘Paul’ ascribes no properties to the named man, but classifies ‘Paul’ as the thing referred to by ‘Paul’.
Characterizing common nouns as referring expressions is clever, but it is metaphysically problematic. A close reading of Kripke suggests that Ben-Yami’s alternative risks referring to bare particulars. If we use ‘tiger’ as a referring expression, we would only be referring to the total group of actual tigers. This structure of referencing picks out each individual particular tiger, without referring to any corresponding properties. If we use ‘Nixon’ merely as a referring expression, we would only refer to Nixon. This reference picks out Nixon without referring to any of Nixon’s properties. Kripke objected to this strategy, saying that it would be erroneous to refer to “propertyless substrata underlying the qualities” (1980, p. 52) of particulars. Referring expressions that pick out particulars without simultaneously picking out at least some properties will always violate this principle and is an unacceptable alternative to natural kind terms.
It may make more sense to reify what we have learned about natural kind terms, artificial kind terms, and understand the function of each. We use natural kinds only in an epistemic way in order to identify elementary particles. In this fashion, we use natural kinds to posit basic properties, not structural properties. When a society discovers that a natural kind possesses structural properties, such a kind will cease to be a natural kind. In contrast, society uses an artificial kind term to identify structural properties. Of course, one may object that the idea of natural kinds is becoming obsolete. When scientists discover that so-called natural kinds possess structural properties, they will probably also posit new basic properties. For example, scientists previously believed that protons and neutrons were basic particles, and before that, that atoms were basic particles. Some day, scientists may understand how to split photons or why electrons have unique spins. When scientists make these advancements, they will recognize that photons and electrons are not basic particles, and will eventually discover new elementary particles that are more fundamental than the natural kinds they previously named. In this process, scientists will rename the old natural kinds into artificial kinds and commission new natural kinds. These scientists will derive the new natural kinds from the limits of a society’s knowledge, not from metaphysics. Natural kinds, therefore, will signal what a society does not know about what it names.
Ben-Yami, H. (1999). The semantics of kind terms. Philosophical Studies, 102, 155-184
Kripke, S. A. (1980). Naming and necessity. Cambridge: Harvard University Press.
Mellor, D. H. (1977). Natural kinds. The British Journal for the Philosophy of Science, 28(4), 299-312.
Putnam, H. (2001). Meaning and reference. In A. P. Martinich (Ed.), The philosophy of language (Fourth ed.) (pp. 288-195). New York: Oxford University Press.