Printer Friendly

Natural kinds: direct reference, realism, and the impossibility of necessary a posteriori truth.

Scientists have discovered that water is [H.sub.2]O. "Water is [H.sub.2]O" is true. But is it a necessary truth? In other words, is it true in all possible worlds? Some people think it is. For example Hilary Putnam, in his well-known Twin Earth argument, concludes that "water is [H.sub.2]O" is necessarily true; thus a liquid which phenomenally resembles [H.sub.2]O and fits the description of water in almost all aspects, but has the chemical formula XYZ, cannot be water.(1) Saul Kripke has made a similar claim about the necessary identity between water and [H.sub.2]O. Because this type of truth is based on empirical discoveries, Kripke calls truths of this sort "necessary a posteriori."(2) The thesis shared by Putnam and Kripke has two premises: a realist view that natural kinds exist independently of human cognition, and a theory of direct reference of natural kind terms. Opposing the view that natural kind terms pick out objects through descriptions, Putnam and Kripke hold that natural kind terms pick out natural kinds in the world in a direct way. Based on these two premises, they argue that, if two natural kind terms, A and B, designate the same thing in the world, "A = B" expresses a necessary identity.

Objections to the Putnam-Kripke thesis have been numerous.(3) My argument against this thesis takes a different route from that taken by most of their critics. I will grant Kripke and Putnam as many assumptions of their theory as possible, but show a different logical conclusion. In particular, I grant that Kripke and Putnam are right in their theory of how natural kinds are named; I also grant that they are right in their realist view of natural kinds. I will show that, contrary to their belief, even with these assumptions, their conclusion about the necessary a posteriori truth of natural-kind identity is flawed. Specifically, I argue that, if Kripke and Putnam are right, then in naming a natural kind, there is always a certain vagueness as to at what taxonomic level the term is posited, hence a vagueness as to the exact scope of that natural kind, and consequently, an inevitable indeterminacy as to what is named. This feature of natural-kind naming determines that true identity statements of natural kinds can never express necessary truth.


Indeterminacy of Natural Kind Term. I will begin with a difference between a kind and an individual and the implications of this difference for naming. Unlike an individual object a kind could, in principle, have an inexhaustible number of instances. This results in an important difference between naming a natural kind and naming an individual. In naming an individual, we can have the individual right in front of us and name it by ostensive definition. In naming a natural kind, we cannot have the kind right in front of us. Even if we could collect all instances of a kind which currently exist, we nevertheless would not have the kind at hand, because a kind is supposed to include instances which have existed in the past and instances that will come into existence in the future. That is to say, in naming we do not have direct access to a kind as we do to an individual. This characteristic of naming natural kinds, as I shall show, has direct consequences for determining what is named.

According to Kripke and Putnam, a natural kind can be named by ostension through paradigmatic instances of a kind.[4] For example, we point to a glass of water and say, "Let's call this (liquid) ~water'." In doing so, we have directly named a kind of thing, namely, water. We might be wrong about the properties of the kind being named; water might turn out not to be colorless and tasteless even though we think it is. Should that happen, we would not think we had named something that does not exist, but rather that we had been wrong about the kind of thing we had named. This is so because, according to Kripke and Putnam, in naming we fix the reference directly to what is named, and what we have named is not dependent on the properties we have thought it to possess. In other words, in naming a kind, we directly name "that kind of thing."(5)

But what kind of thing? A natural response to this question would be, "The kind of which the instances that have been used in naming are paradigmatic." By answering in this way, however, one has assumed that a group of instances can be paradigmatic of only one definite kind. This is by no means a simple matter. On the issue of the relation between an object and its kind(s) there are at least three views. First, one may think that an object can be in only one kind. For example, one may think that since an object has only one essence, it has to belong to only one species which is determined by its essence, and that only biological-species terms are natural-kind terms. Holding this view, however, will drastically reduce the scope of the direct reference theory and leave most natural-kind terms (for instance, genus terms and subspecies terms) unaccounted for. The theory of direct reference has been proposed as a theory for naming in general. It should have explanatory power for all direct reference terms of the same mechanism. If a genus term is introduced into the language through the same mechanism as a species term, it should have the same direct-reference character as a species term does. Thus this path would lead to very odd consequences for the direct-reference theory. As a matter of fact, neither Kripke nor Putnam seems to have taken this path. Putnam has listed "acid" as a natural kind term, which stands for several compound substances that have different chemical formulas.(6) Kripke thinks "common names" is an expression "quite appropriate" for natural kind terms, and uses "cow," which stands for the mature female of the genus Bos, as an example.(7) Unless they were to change their views in a drastic way and consequently greatly reduce the power of their theory Kripke and Putnam, I assume, would not accept the view that objects can be in only one kind. They would like their account to apply to all natural kind terms, including species and genus terms.(8)

The second view is represented by writers such as Bruce Goldberg, W. V. O. Quine, John Dupre, and Philip Kitcher.(9) They have argued that one item may have membership in overlapping natural kinds or classes at the same taxonomic level. For instance, Quine writes, "kinds can . . . overlap; the red things can comprise one kind, the round another."(10) Whether this is true of natural kinds has been a matter of dispute. In biology, for example, Willi Hennig in his celebrated book Phylogenetic Systematics maintains that only one taxonomic system can be regarded as privileged, and accordingly that one organism cannot properly belong equally well to two overlapping kinds at the same taxonomic level.(11) Putnam and Kripke would agree with Hennig in rejecting the second view. There is a third option. It is the view that there is one correct taxonomic theory and one object or instance belongs to only one species and one genus, but not to two or more overlapping kinds at the same taxonomic level. Holders of this view usually rely on the idea of scientific taxonomy, the idea that it is up to the scientists to find out the taxonomic structure of natural kinds. This view is favorable to the Putnam-Kripke theory of direct reference. On the one hand it allows, as Putnam and Kripke intend, a broad application of their direct reference theory to both genus and species terms; on the other hand, it avoids some difficulties that are raised by the Quinean view. For the sake of argument, I will assume the third view is the only correct one.(12)

The third view, however, does not imply that an object can be in only one natural kind, because both species and genera are natural kinds.(13) Acid is a natural kind (per Putnam), but I have every reason to say that [H.sub.2][SO.sub.4] (sulphuric acid) is also a natural kind. Likewise, cow (the mature female of the genus Bos) is a natural kind (per Kripke), but I have every reason to say that a species of the genus Bos is also a natural kind. Similarly, we can say that apple as well as Red Delicious are natural kinds. This view poses a difficulty for Putnam and Kripke. A Red Delicious can be an instance of the kind Red Delicious and it can also be an instance of the kind apple. In other words, natural kinds are nested in a taxonomic framework, and within this framework instead of a one-one relation there is a one-many relation between an individual and its kinds. When we point to some Red Deliciouses and name "that kind of thing" the question now is, What kind of thing has been named?

To Illustrate this point, let us imagine a possible world W (or, for those who do not like possible world semantics, an isolated island people with their own language) in which there are no fruits and consequently no such natural-kind terms as "apple" in the W people's language. Let us now throw a number of Red Deliciouses into that world. Suppose the W people point to the Red Deliciouses and name that kind of thing "ABC." Let's further suppose the W people endorse the Putnam-Kripke theory of direct reference and believe that ABCs do exist even though they might have been mistaken about the properties of these objects, such as roundness, sweetness, and so forth. The problem, however, is that there is a taxonomic hierarchy (for instance, Red Delicious/Delicious/Apple/ . . . ) in which kind terms such as "ABC" have to be nested. What "ABC" means depends on at what taxonomic level the term is posited, and its level cannot be clearly determined through paradigmatic instances alone. In the case of ABC, what kind of thing has been named? Red Delicious? Delicious? Apple? Or something even broader in the taxonomic system? At this stage there is no way to tell. If we asked the W people what "ABC" meant, they would probably say a natural kind represented by those Red Deliciouses. But if we pressed further and asked whether by "ABC" they meant Red Delicious, Delicious, apple, or something broader than apple in the taxonomic system, they would look at us, puzzled, and probably ask us, What is the difference? Of course, there is a difference. We know there is a big difference. If ABC is Red Delicious, a Macintosh is not an ABC. If ABC is something broader than apple, not only a Macintosh but also a pear might be an ABC. The point is that the W people have not recognized the difference.

One may contend that whether the W people realize anything or not is irrelevant, and what ABCs are is something they need to discover in the world, not something to be decided by them. In other words, the issue is ontological, not epistemological. This objection confuses the issue. I do not suggest that the way we name objects can change objects in the world. Even if we assume the realist claim about the independent existence of the world is correct, the above problem for the theory of direct reference remains. While the existence of species is objectively independent of our mind, the existence of genera and subspecies is also objectively independent of our mind. An object is an instance both of a species and of a genus, which are both natural kinds. The issue here is how names are connected to natural kinds in the world, and in particular, how a name is connected to a species instead of a genus or vice versa. It is a matter of naming. We cannot expect a natural kind term to have reference even before the term has been adopted into language. Nor can we discover what ABCs are before we understand what it takes for there to be ABCs. Assuming the Putnam-Kripke realist view of natural kinds is correct, ostension through paradigmatic instances alone does not determine what natural kind has been named.

Taking place in a certain context, an ostensive definition does exclude objects from the kind. For example, perhaps no one in W, unless extremely mistaken, would wonder whether a table is an ABC. In our case, however, nothing in the W people's minds could definitively determine whether Macintoshes are also ABCs. In naming, the namers usually do not specify features of the kind in great detail as the criteria of the kind they name. Also, as Kripke points out, they may be wrong about these features. Because the contrast between Red Deliciouses and Macintoshes does not occur in the W people's minds at the time of naming, the issue of whether ABC includes Macintoshes simply does not arise until after the name has been put into use. After the name had been put into use, it does not bear enough import to determine whether a Macintosh is also an ABC. The problem with the W people is this: On the one hand, Macintoshes share sufficient similarities with Red Deliciouses to be one kind; on the other hand, they also bear sufficient dissimilarities with Red Deliciouses to be a different kind. Therefore, the W people have to make a decision as to whether a Macintosh is an ABC, and this decision cannot be dictated by what they had in their minds at the time of naming.

Putnam and Kripke would like to say that it is up to the scientists to find this out. Scientists in W, however, can do no better than decide whether a Macintosh is also an ABC. Shifting the burden to scientists does not change the nature of the issue. The naming of a natural kind cannot be accomplished once and for all by using paradigmatic instances. The only way to find out more about what kind of thing has been named "ABC" is to show the W people some fruits other than Red Deliciouses and ask them whether these fruits are also ABCs. If we show them some Macintoshes and they call them "ABCs," then probably "ABC" does not mean "Red Delicious." If in addition to Red Deliciouses the W people call pears "ABCs," then "ABC" cannot merely mean apple. It is not the case here, however, that the W people make a judgment according to a fixed standard, because there is no such established standard. In further focusing on what is named, they need to decide whether a Macintosh is also an ABC.

This kind of decision is also a decision about what nature or sameness relation is involved and what kind of thing ABCs are. Putnam maintains that the kind is determined by a certain sameness relation among its instances.(14) But what sameness relation? A Red Delicious bears a certain sameness relation to other Red Deliciouses insofar as they are Red Deliciouses. It also bears a certain sameness relation to Macintoshes insofar as both Red Deliciouses and Macintoshes are apples. The W people need to make the further decision because the initial ostensive definition does not bear enough import regarding whether, for instance, a Macintosh, which bears a sameness relation with Red Deliciouses at the taxonomic level of apple, is an ABC.

We should note also that the decision is not merely a matter of whether we call some objects "ABCs." It has been decided to call both a financial bank and a river bank "bank," yet this does not make them the same kind. To decide to call both Red Deliciouses and Macintoshes "apples," however, is to recognize a certain sameness relation between Red Deliciouses and Macintoshes, and this decision does indicate that they are in a sense the same kind of fruit, namely, apple. That is, to decide whether Macintoshes and Red Deliciouses both are ABCs is to decide what kind of sameness relation is required for something to be an instance of ABC. The decision has to do with what ABC is or what nature ABC has.

One may think that since the W people use Red Deliciouses as paradigmatic instances, an ABC must be a Red Delicious. In making this assumption one has hastily ruled out the possibility that, if we later showed the W people a Golden Delicious or a Macintosh, they might accept it as an instance of ABC - a possibility that I do not think entirely unlikely. The history of biology shows a history of discoveries of new species and subspecies that belong to some kinds that have been named. To see why it is not entirely unlikely for the W people to accept Macintoshes as ABCs, let us look at another natural kind term, "elephant." The term "elephant" applies to animals of two different species, the Asian elephant and the African elephant. The Chinese counterpart of "elephant" is xiang. The term must have been first used to designate the Asian elephant, because ft was the only elephant the Chinese could have known when they started using the term. Today xiang designates both the Asian elephant and the African elephant. It is reasonable to think the Chinese first used the Asian elephant in fixing the reference and later included the African elephant under the kind name xiang. In the same way, it is possible that the W people should also take Macintoshes as ABCs.

In deciding what to call an "ABC" there is no necessity compelling the W people to call or not to call a Macintosh an "ABC." Were there such a necessity, there would be an exact cross-cultural correspondence between natural-kind terms in two cultures that have the same natural kinds. Obviously, however, we do not have such a uniformity. For example, in Chinese, Lao-shu stands both for mouse and for rat, while Yan and Er stand for wild goose (one that flies) and domestic goose (one that does not fly) respectively. It is possible that, with the same paradigmatic instances, the Chinese have made decisions different from those of English-speakers in introducing these natural-kind terms. The Chinese think mice and rats bear a certain sameness relation in both being Lao-shu, while the English think Yan and Er bear a certain sameness relation in both being geese. Science may prove that they are both right: while one way focuses on the species, the other focuses on the genus. The difference indicates the possibility of going either to the species or to the genus from the same paradigmatic instances.

One should not confuse the case outlined above with that of reference changes of natural kind terms. Sometimes terms do change their references. Our case, however, is different. When the Chinese decided that xiang also referred to the African elephant, it was not the case that the reference of the kind term changed from a narrower taxonomic kind to a broader one. This would have been the case only if the status of the kind term had been already determined. Before the decision was made, however, the taxonomic status of the term was far from having been determined. An ostensive definition alone does not complete the process of naming a natural kind if by naming we mean fixing the reference rigidly to a specified scope which draws a clear line between what is and what is not of the kind. Decisions following an ostensive definition to further specify the kind are also part of the process of naming. The W people would not need to make a decision until they encountered things different from Red Deliciouses but similar enough to them (Macintoshes, for instance) for the W people to wonder seriously whether these new things were also ABCs, and they might not encounter such objects in a million years or ever at all. Even after they encountered some objects of this sort, further decisions would still be needed. Generally speaking, when we come across a new object O that makes us seriously wonder whether it belongs to a kind (which we have already named in the past without further focusing to determine whether objects like O are or are not instances of the kind), we need to decide whether it is of the kind.

Therefore, the ultimate reference (or the scope) of a natural kind term is pending and hence indeterminate as long as the entire process of naming is not finalized. The process can never be finalized. Of five hundred million species of animal and plant life estimated to have existed on Earth, 99.75 percent are extinct. There are at least five hundred species of fleas and one hundred species of mosquitoes.(15) There are always new species emerging. Furthermore, we can always imagine an instance (Putnam's XYZ is a good example) that makes us seriously wonder whether it belongs to a kind which has been named, but for which no effort has been made to determine whether that instance is one of the named kind. The ongoing dispute over the identity of water itself shows that the taxonomic status of the natural kind term "water" needs further focusing. The very fact that we are debating over the issue whether XYZ is water proves that the taxonomic status of the term "water" has not been finally determined.


Term Indeterminacy and the Impossibility of Necessary Truth. Together with the thesis of the necessary identity of natural kinds is found the following doctrine: If there is a natural kind K which has been correctly identified as a kind I (with the characteristic i), and if we imagine a kind 0, which resembles I in almost all aspects, but is not I, then, since "K" rigidly designates the kind K, necessarily O cannot be K. For example, water has been correctly identified as [H.sub.2]O; a liquid X resembles but is not [H.sub.2]O; necessarily X is not water. Both Kripke and Putnam embrace this doctrine. I believe their conclusion is unwarranted.

There are two ways for a natural kind K to be identified as I, by definition or by empirical discoveries. K is identified as I by definition when we name a natural kind K that is characterized as I. For example, we point to an object and announce "let's call this kind of thing with the feature i, ~K.'" By doing so, we identify K as a kind with the feature i. This does not mean, however, that K could not have been otherwise. As Kripke has noted, we might be wrong in our identification. We might later find that there are some instances of K without i. Therefore, even though we have defined (or, to use Kripke's term, fixed the reference of) K as possessing i, something that lacks i might also be K. It is not necessary, therefore, that all Ks possess i. The other way in which a natural kind K is identified as I is by empirical discoveries. This is the way, according to Kripke, to reach necessary a posteriori truth. As has been noted, a kind may have an inexhaustible number of instances. No empirical examinations can exhaust all instances of a kind. Thus empirical discovery can by no means guarantee necessary identity between K and I. Even if all instances of K we have discovered are Is, the taxonomic status of K may still need to be determined. Given the nature of natural-kind naming, there is always the possibility of a genus-species (or kind-subkind) relation between K and I, and thus K and I are not necessarily identical. In other words, for K and I to be necessarily identical, "K" and "I" must both rigidly designate the same kind. Because of the indeterminacy of the ultimate reference of these terms, such rigidity does not obtain; therefore, the necessity of their identity is not warranted.

Putnam's water argument may be put as this: on Twin Earth there is a liquid the chemical formula of which is XYZ. XYZ is indistinguishable from the [H.sub.2]O of our Earth at normal temperatures and pressures. On Twin Earth XYZ fills up oceans and lakes, people drink XYZ and use it to relieve drought for their crops, and so forth. is XYZ water? Putnam argues that, since water is [H.sub.2]O and XYZ does not bear the sameness relation to [H.sub.2]O, it cannot be water.(16)

I contend that, because of the indeterminacy of the ultimate reference of the term "water," there is no necessity that XYZ not be water. As we have said earlier in discussing the ABC case, the W people do not know more about what sameness relation has been included in being ABC until it becomes an issue whether fruits other than Red Deliciouses are ABCs, and they need to decide whether, for example, Macintoshes are ABCs. Whether Red Deliciouses and Macintoshes bear the sameness relation to each other at the level of being ABC depends on whether ABC is defined as apple, Red Delicious, or something else; this cannot be decided until an object such as a Macintosh (or at least the description of it) is presented to the namer. In the same way, whether XYZ is water depends on whether it has a sameness relation to [H.sub.2]O sufficient for being water, and whether XYZ bears a sameness relation to [H.sub.2]O sufficient for being water depends on the level at which "water" is posited. We cannot say whether the sameness relation would allow XYZ to be water until the decision is made, and this decision is not made until whether XYZ is water becomes a real issue. Given that Red Deliciouses and Macintoshes bear a certain sameness relation in being apples, and that [H.sub.2][SO.sub.4] (sulphuric acid) and HCI (hydrochloric acid) bear a certain sameness relation in being acid, there are also reasons to think that XYZ and [H.sub.2]O might bear a certain sameness relation sufficient for both being water.

It may be a fact that we originally used instances of [H.sub.2]O as Paradigmatic instances in fixing the reference of "water" and that so far all instances of water that have been found are [H.sub.2]O. Does that mean that necessarily XYZ cannot be water? It does not. The Chinese used Asian elephants for the initial ostensive definition of xiang; yet the scope of xiang was not monopolized by the Asian elephant. It is fallacious to say that, because Asian elephants were used as paradigmatic instances of the kind xiang, necessarily the African elephant cannot be a xiang. Similarly as xiang is a generic (or family) term posited between "mammal" and "Asian/African elephant," "water" might be a genus term posited between "[H.sub.2]O/XYZ" and a taxonomically higher kind.

One may object that for [H.sub.2]O and XYZ to be two species of the same genus they must have the essential sameness relation; Putnam's point is that, even if XYZ has the same phenomenal properties as [H.sub.2]O, because it does not have the essential sameness relation XYZ cannot be water. What sameness relation is essential, however, depends on whether it is essential to the species or genus. For example, presumably even though [H.sub.2][SO.sub.4] does not have the essential sameness relation for being HCI and vice versa, they both have the essential sameness relation of being acid. Even though different species of cows do not have the essential sameness relation for being the same species, they nevertheless have the essential sameness relation for belonging to the same genus. Therefore, even though XYZ does not have the essential sameness relation for being [H.sub.2]O and [H.sub.2]O does not have the essential sameness relation for being XYZ either, they nevertheless may have the essential sameness relation for being the same genus. Therefore, it is still possible for them both to be water. Unless Putnam and Kripke choose to hold that only species are natural kinds and have essential properties - and hence to drastically reduce the explanatory power of their theory - they have to face the possibility of the two alternatives illustrated here and hence the possibility of XYZ being water.

Like Putnam, Kripke does not argue on a priori grounds that nothing other than [H.sub.2]O can be water. En fact, he would probably allow the epistemic possibility that water might turn out to be XYZ, since it is an empirical discovery that water is [H.sub.2]O and we might have been mistaken about this. Kripke argues that natural-kind tenus are rigid designators,(17) and since we use a natural-kind term T to designate the kind, nothing which is not the kind can be T. For example, he writes, "we use the term ~tiger' to designate a species, and . . . anything not of this species, even though it looks like a tiger, is not in fact a tiger.(18) I do not think this argument really adds anything in Kripke's favor. First, speaking of a kind, we all agree that, if T designates a kind, anything not of this kind cannot be an instance of T. But the question is precisely whether something is of this kind. Second, even though the term "tiger" is currently used as a species term, it is not necessary that tiger be one species. Assuming that we are right in believing that all tigers in the world are in fact one species (that is, Felis tigris), we nevertheless cannot rule out the possibility that there might be another species of tiger. Suppose we now find another species which has genes different from Felis tigris, but resembles Felis tigris enough (as the African elephant resembles the Asian elephant). Is it tiger? To say we must call it "tiger" may be an overstatement, but given what has happened to xiang ("elephant") it seems at least that we cannot say that necessarily the species is not one of tiger. As a matter of fact, even Kripke himself does not rule out the possibility that there might be "two kinds of tigers."(19) If there were another kind of tiger, a tiger of that kind would not have to be a Felis tigris. Therefore, "tiger = Felis tigris" is not necessarily true.

In the case of water, the disagreement is precisely about whether XYZ is water (the kind). We can all agree that if XYZ does not belong to the natural kind water, then it is not water (a tautology). But the dispute is really about whether water is a taxonomically broader kind which contains [H.sub.2]O and XYZ as subkinds. Just as there might turn out to be two kinds of tigers, there is no reason to assert that since [H.sub.2]O is water, XYZ cannot be water: there might be two kinds of water and both [H.sub.2]O and XYZ might thus both be water. Just as there are two kinds of elephants but the term "elephant" is still a natural kind term designating the family Elephantidae, if there are two kinds of water, the term "water" can still be a natural kind term designating the genus water. Therefore, "water = [H.sub.2]O" is not a necessary truth.

Perhaps at this point one may want to say that, if there might be two species of tigers or XYZ might be water, this only shows that identity statements like "tiger = Felis tigris" and "water = [H.sub.2]O" are untrue; and such an eventuality would do no harm to Kripke's thesis because he only says that if such a statement is true it is necessarily true. What, however, is the truth condition for such a statement? We have shown that the ultimate reference of a natural kind term is indeterminate. If by saying that a statement like "tiger = Felis tigris" or "water = [H.sub.2]O" is true we mean the two involved terms have the same ultimate reference, we can never be in a position to say that such a statement is true. We simply do not know whether the two terms have the same ultimate reference. In that case, Kripke's thesis that "if such a statement is true it is necessarily true" would mean little, if not nothing, because the antecedent is never warranted. It seems unlikely that Kripke would be content with this agnostic view. If he refuses this view and takes the position (as most people would) that we can know whether such statements are true, he has to accept that a statement such as "tiger = Felis tigris" is true if and only if every instance of tiger is also an instance of Felis tigris and vice versa. This has to be so in this world or Kripke's thesis would be a cheap tautology. His thesis would not amount to much if it were to say that if every instance of tiger is an instance of Felis tigris and vice versa in all possible worlds then "tiger = Felis tigris" is a necessary truth; because that is the way we define necessary truth. If the proposition, "tiger = Felis tigris," is meant to be meaningful, and if there is any reasonable way for us to discover this proposition to be true or false, it has to allow us to discover this in this world. It has to allow us to rely on empirical scientific evidence we can gather in this world. Science has provided sufficient evidence - if anything counts as sufficient evidence - that every instance of tiger is an instance of Felis tigris and every instance of Felis tigris is an instance of tiger in this world. Therefore, "tiger = Felis tigris" is true in this world. But it is by no means a necessary truth.

Let me emphasize that my argument is not that in a possible world the word "tiger" or "water" might be used in a different way; for instance, that people might call a chair "tiger." That is something about which everyone would agree. My argument is that in this world the ultimate reference of such terms is indeterminate, and therefore, statements like "tiger = Felis tigris" do not express necessary truths. If "tiger" turns out to be a term designating a broader scope than "Felis tigris," is it the same term as the term "tiger" we have been using so far? The answer has to be affirmative. Just as xiang has remained the same term while now being used to designate the African elephant in addition to the Asian elephant (it remains the same entry in the dictionary), "tiger," if its use were expanded beyond Felis tigris, would still be the same term in the same sense as xiang is. As Kripke would agree, in that case we would say not that "tiger" had changed its meaning but that we had discovered a new species of tiger. The same holds true for "water." Because of the indeterminacy of the ultimate reference of natural-kind terms and the need for further focusing of these terms, the discovery of a new species is only a step forward in the process of the term's being further focused.


A Brief Summary. I have argued that, because we have no direct access to a natural kind as we do to an individual object, when we name a natural kind exactly what the kind is cannot be determined without further focusing. The process of further focusing can never come to an end. What happens when we later find some instances which are so similar to the paradigmatic instances that we have used in naming that we ask whether these newly discovered instances belong to the same kind? This situation calls for us to decide whether the new instances are the same kind as the paradigmatic instances and, which is the same thing, what the kind is. Therefore, when we have two natural kind terms, A and B, and "A = B" is true, there is the possibility that they are related as genus and species (or better, as kind and subkind), even though the genus may have only one species (but it could have more). My argument has direct consequences for the Putnam-Kripke thesis of the necessary a posteriori truth of identity statements of natural kinds. I have shown that either Putnam and Kripke have to say that it can never be found out whether such statements are true, or they have to accept a truth condition which does not guarantee necessary truth. Either way, the Putnam-Kripke thesis of necessary identity is flawed.(20)

(1) Hilary Putnam, "Meaning and Reference," Journal of Philosophy 70, no. 19 (November 8, 1973): 699- 711. (2) Saul Kripke, Naming and Necessity (Cambridge: Harvard University Press, 1980). Kripke and Putnam each independently developed a view of direct reference of natural kind terms. While their views have differences, they are the same in spirit. In this paper I discuss them mostly together, and separately only when needed. (3) To name a few, the list includes Eddy Zemach, John Dupre, Mark Wilson, Hector-Neri Castaneda, and Avrum Stroll. See Eddy Zemach, "Putnam's Theory on the Reference of Substance Terms," Journal of Philosophy 73, no. 5 (March 11, 1976): 116-127; John Dupre, "Natural Kinds and Biological Taxa," Philosophical Review 90, no. 1 (January 1981): 66-90; Mark Wilson, "Predicate Meets Property," Philosophical Review 91, no. 4 (October 1982): 549-89; Hector-Neri Castaneda, "Semantic Holism Without Semantic Socialism: Twin Earths, Thinking, language, Bodies, and the World," Midwest Studies in Philosophy 14 (1989): 101-26; and Avnun Stroll, "What Water Is or Back to Thales," Midwest Studies in Philosophy 14 (1989): 258-74. (4) Kripke, Naming and Necessity, 122, 136; Putnam, "Meaning and Reference," 702. (5) Kripke, Naming and Necessity, 122. (6) Hilary Putnam, "Is Semantics Possible?" in Naming, Necessity, and Natural Kinds, ed. Stephen A. Schwartz (Ithaca: Cornell University Press, 1977), 102-32. (7) Kripke, Naming and Necessity, 127. (8) In a recent article, T. E. Wilkerson proposes the view that neither species nor genera are natural kinds; see T. E. Wilkerson, "Species, Essence and the Names of Natural Kinds," Philosophical Quarterly 43, no. 170 (January 1993): 1-19. Wilkerson holds that there are many more natural kinds than species and that many natural kinds perhaps only have one member. He even goes so far as to "leave open the very definite possibility that there will be as many kinds as individuals" (p. 16). I leave this view aside for at least two reasons. First, it is a view that neither Kripke nor Putnam is likely to endorse. Second, by reducing kinds to individuals one may be able to avoid many criticisms of the realist view of natural kinds, but at the same time one also rejects the entire concept of a natural kind. (9) Bruce Goldberg, "The Linguistic Expression of Feeling," American Philosophical Quarterly 8, no. 1 (January 1971): 86-92; W. V. O. Quine, "Natural Kinds," in Naming, Necessity, and Natural Kinds, ed. Stephen P. Schwartz (Ithaca Cornell University Press, 1977), 155- 75; John Dupre, "Natural Kinds and Biological Taxa"; Philip Kitcher, "Species," Philosophy of Science 51, no. 2 (June 1984): 308-33. (10) Quine, "Natural Kinds," 159. (11) Willi Hennig, Phylogenetic Systematics, trans. D. Dwight Davis and Rainer Zangerl (Urbana University of Illinois Press, 1979), 9. (12) This distinguishes my argument from that of many of the above-mentioned philosophers. Mark Wilson's argument, for example, relies heavily on examples of artificial kinds and contends that to distinguish sharply between natural kinds and artificial kinds and to limit the Putnam-Kripke account to natural kinds is "incoherent." Since there are many philosophers who believe that there is, or at least should be, a distinction between natural kinds and artificial kinds, an argument that does not rely on the rejection of this distinction is needed. My argument is also different from the argument presented by Hartry Field, who is primarily concerned with scientific terms such as "mass" and draws his motivating examples entirely from Kuhnian "scientific revolutions." See Mark Wilson, "Predicate Meets Property," Philosophical Review 91, no. 4 (October 1982): 549-89; and Hartry Field, "Theory Change and the Indeterminacy of Reference," Journal of Philosophy 70, no. 14 (August 16, 1973): 462-81. (13) For the sake of simplicity, I use "taxonomy" in a very broad sense. What I say about species and genera applies to biological species and genera as well as to element and chemical compound kinds. Also "species" and "genus" are used in the way that they are in logic; that is, "genus" means a relatively larger kind and "species" a relatively smaller kind or subkind of the genus. Thus, when "species" is used along with "genus," they are intended to stand for, in addition to biological concepts, concepts of two adjacent levels within the taxonomical hierarchy in general. (14) Putnam, "Meaning and Reference," 702. (15) Ronald C. Pine, Science and the Human Prospect (Belmont: Wadsworth, worth 1989), 18, 74. (16) Helen Steward argues that the "is" of this type is "simply predicative," and not the "is" of identity. For the sake of argument, I assume Putnam and Kripke's interpretation is correct. See Helen Steward, "Identity Statements and the Necessary A Posteriori," Journal of Philosophy 87, no. 8 (August 1990): 385-98. (17) For some problems with Kripke's concept of rigid designation see Chenyang Li, "Kripke's Two Definitions of Rigid Designation," Iyyun: The Jerusalem Philosophical Quarterly 41, no. 1 (January 1992): 63-71. (18) Kripke, Naming and Necessity, 121. (19) Kripke, Naming and Necessity, 121. (20) This paper was developed from a part of my Ph.D. dissertation, "Toward A Contextual Approach To The Question Of Being," written at the University of Connecticut. I would like to take this opportunity to thank Joel J. Kupperman, my major adviser, and Samuel C. Wheeler III and Garry M. Brodsky, my associate advisers, for their guidance and assistance. Under the title, "Identity, Necessity, and Natural Kind Terms," an early version of this paper was presented at the Ninetieth Annual Meeting of the Central Division of the American Philosophical Association in Louisville, Kentucky, on April 23-26, 1992. I would like to Tomis Kapitan for commenting on the paper. My gratitude also goes to Eddy M. Zemach, Ruth B. Marcus, Ruth G. Donald L. Baxter, Crawford L Elder, and John G. Troyer for their valuable comments and suggestions. The revision of this paper was partly supported by office facilities provided by the Department of Philosophy of the University of Connecticut in the summer of 1993.
COPYRIGHT 1993 Philosophy Education Society, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Li, Chenyang
Publication:The Review of Metaphysics
Date:Dec 1, 1993
Previous Article:Logical implication and the ambiguity of extensional logic.
Next Article:Postmodernity or late modernity? Ambiguities in Richard Rorty's thought.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters