JOSEPH LAPORTE
RIGIDITY AND KIND (Received 13 March 1998)
Saul Kripke has argued famously that identity statements lik...
9 downloads
420 Views
83KB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
JOSEPH LAPORTE
RIGIDITY AND KIND (Received 13 March 1998)
Saul Kripke has argued famously that identity statements like ‘Hesperus = Phosphorus’ are necessary, if true at all, as a result of containing two rigid designators. In the same way, ‘Water = H2 O’ is supposed to be necessary if true, as a result of containing two rigid designators. But the applicability of arguments for the former claim, which involves concrete-object designators, to the latter claim, which involves kind designators, is hardly straightforward. A considerable number of questions, problems, criticisms, and complications concerning the extension have been raised in the literature. At the center of these difficulties rests rigidity. The crucial question is whether an appropriate understanding can be provided for the rigidity of kind designators. An affirmative answer will be defended here.
I
One of the first puzzles that was ever addressed concerning the rigidity of kind terms pertains to what they could possibly rigidly designate. Rigidity would seem to be more straightforwardly applied to object designators. ‘Hesperus’ rigidly designates the planet Venus because it designates Venus in each possible world (that contains it).1 ‘The brightest celestial object, apart from the moon, in the evening sky’, by contrast, designates Venus nonrigidly, since it designates that planet in fact, but there are possible worlds in which it designates, say, Neptune. Neptune might have shone more brightly than Venus, and in possible worlds in which it does, ‘the brightest celestial object, apart from the moon, in the evening sky’ designates Neptune, and not Venus. So far so good. But if vernacular terms like ‘water’ and ‘whale’ and theoretical terms like ‘H2 O’ (for more examples, see Kripke Philosophical Studies 97(3): 293–316, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.
294
JOSEPH LAPORTE
1980, p. 134) are supposed to be rigid as well as names like ‘Hesperus’, what are these kind terms supposed to rigidly designate? At first blush, it would seem that they don’t rigidly designate anything at all; rather, it would seem, ‘water’ and ‘whale’ designate different entities in different possible worlds, and so are non-rigid. A term like ‘whale’ designates in this world, it is natural to suppose, its extension: the thing or set of things to which it applies, which is, in this case, the set of actual whales out there in the ocean, including Gigi and Flipper. But there are worlds in which Gigi and Flipper never come into existence, but other whales, who haven’t in fact come into existence, do. ‘Whale’ has a different extension in such worlds. Similarly, in the case of water, there are possible worlds in which the water that actually exists doesn’t exist: ‘water’ has an extension in (some of) those worlds, however. Its extension in those worlds consists of water particles that as a matter of fact don’t exist, but might have. So apparently if kind designators designate their extensions, they aren’t rigid. Fortunately, there is another proposal concerning what these terms designate. Steven Boër (1985, pp. 129–135), Keith Donnellan (1983, pp. 90f.), and Fabrizio Mondadori (1978) have all urged that kind designators be taken to designate kinds, or other abstract entities.2 This is to think of the terms, as Donnellan puts it, following Mill, as “abstract nouns” (1983, p. 90). It is easy to see how the “abstract designator” account allows for kind designators’ rigidity. The problem of what could be designated in all possible worlds is settled thus: a kind designator designates the same abstract kind (or other entity) in every possible world, even if the concrete individuals instantiating the respective kind, and comprising the term’s extension, vary from world to world. So here is an account that allows it to be seen how rigidity might be applicable to kind designators.
II
Of course, that the account at issue can show how rigidity might be applicable to kind designators does not assure its acceptability. If it is to be satisfactory, the account must allow for terms relevantly like ‘water’ and ‘H2 O’ to come out rigid. And, presumably, it must make
RIGIDITY AND KIND
295
other expressions designating kinds come out non-rigid. A number of writers have expressed misgivings over whether the account offered can apply to the appropriate designators in this way. Some have cast more general doubts over whether any account of rigidity applying to kind designators can apply to the right designators. Criticisms fall into two camps: first, it is common to charge that the above account (or any account) of rigidity for kind terms makes too many expressions rigid. And second, it is commonly charged that on the account in question, too few expressions come out rigid. It is my aim to show that these charges do not tell against the account put forward above: the account withstands criticisms from both directions. Criticisms of the former variety will be examined in the present section, and those of the latter type will be examined in the following section. One reason for supposing that any account of rigidity like the one characterized here applies to too many expressions is rooted in the conviction that artificial-kind terms should not be counted rigid. For on the proposal in question, a natural-kind term such as ‘whale’ is rigid on account of designating the whale kind across all possible worlds. But observe that non-natural kind designators designate abstract entities (artificial kinds, in their case) across all possible worlds too, so rigidity must be extended to them, too, if the account advanced here is right. For example, ‘pencil’, taken as an abstract noun, refers in every possible world to the pencil kind, or to pencilhood. Since Putnam is only too ready to extend his semantics for natural-kind terms to artifact terms such as ‘pencil’ (see Putnam 1975, pp. 242ff.; cf. Wiggins 1994, p. 204), the similarity might at first be thought only to confirm this connection. In fact, it does not. Putnam himself refuses to extend his semantics for natural-kind terms to so-called “one-criterion” words, which he takes to be analytically defined (see Putnam 1983, p. 89). ‘Hunter’ is an example. Putnam holds this word to possess a traditional, analytic definition: ‘hunter’ is defined one who hunts. Yet, taken as an abstract noun, ‘hunter’ refers to hunterhood in all possible worlds, while individuals instantiating hunterhood vary from world to world. ‘Bachelor’, another strong candidate for one-criterion, analytically defined status, similarly refers to the bachelor kind, or to bachelorhood, in all possible worlds. Hence, ‘bachelor’ and
296
JOSEPH LAPORTE
‘hunter’ are rigid if ‘water’ and ‘whale’ are. This is commonly perceived to be unacceptable, and hence, as Schwartz calls it, a “glaring problem” for the account discussed above (Schwartz 1980, p. 190; see also his 1977, pp. 37ff.).3 But what is the glaring problem? What reasons are there for supposing that rigidity should have to be restricted to natural, as opposed to artificial-kind terms? One reason for supposing that artificial-kind terms cannot be counted rigid is that to count them rigid would trivialize rigidity by counting every kind designator rigid. Surely some designators for kinds are intended to be non-rigid, or else there would apparently be no reason to take special note of the rigidity of certain kind designators, as if to distinguish them. Since, on the preferred account of rigidity above, every kind designator comes out rigid, something would appear to be wrong. Something like this worry seems to be what bothers a number of authors about proposals like the above. Schwartz asserts that “Just as with singular terms there is a contrast between proper names and definite descriptions, in that the former are rigid whereas the latter are non-rigid, with general nouns there is a contrast between those that are natural kind terms and those that are not” (1980, p. 196). He thus objects to proposals that extend rigidity to non-natural-kind terms. Danielle Macbeth seems to have something like this concern in mind also, when she criticizes accounts like the above on the grounds that they render rigidity trivial: “in this trivial sense, every predicate, whether or not it is a natural kind term, is rigid” (1995, p. 263; see also pp. 266, 268–271, 276–277).4 How does this general criticism fare? Not well. My preferred account does not trivialize rigidity. It is simply not the case that every kind designator rigidly designates its kind on my preferred account. ‘Honeybee’ (or, alternatively, ‘the honeybee’),5 which names a species of bee, seems to pick out the honeybee kind rigidly. But other kind designators pick out this kind by way of the kind’s fitting a particular accidental description. ‘The insect species that is typically farmed for honey’ is an example: this expression designates the honeybee, since the honeybee is typically farmed for honey and no other insect is. But it is not a designator that picks out its kind in every other possible world. Rather, it picks out that kind by means
RIGIDITY AND KIND
297
of the honeybee’s happening to satisfy a particular description, one that fails to apply to the honeybee in various other possible worlds. It might, for example, have been typical to farm bumblebees for honey, rather than any other species of bee. Had that been so, ‘the insect species that is typically farmed for honey’ would have designated the bumblebee, not the honeybee. ‘The insect species that is typically farmed for honey’ happens to designate the honeybee, but it does not rigidly designate the honeybee. The rigidity of ‘the honeybee’ would seem, then, to find a contrast in certain descriptions, just as the rigidity of names such as ‘Hesperus’ and ‘Ben Franklin’ do. It would not seem to be a “trivial” matter at all that expressions like ‘the honeybee’ are rigid; on the contrary, it is significant that such expressions are rigid, and that others, such as ‘the insect species that is typically farmed for honey’, are not. It is significant because it is necessary to distinguish these types of expressions for the purpose of determining whether certain statements are necessarily true. The differences between these two types of expressions for kinds are significant for the very reasons that the differences between expressions like ‘Ben Franklin’ and ‘the inventor of bifocals’ are significant. Let me elaborate. Like Hesperus, the honeybee has more than one name. It has the vernacular ‘the honeybee’ and the scientific ‘Apis mellifera’. If Kripke’s arguments for the necessity (in case of truth) of identity statements involving two rigid (concrete-)object designators, such as ‘Hesperus = Phosphorus’, are successful, then those arguments ought to work also for identity statements involving two rigid kind designators, such as ‘The honeybee = Apis mellifera’. The line of reasoning is the same: if ‘the honeybee = Apis mellifera’ is true, then ‘the honeybee’ and ‘Apis mellifera’ must designate the same particular (abstract) object. Moreover, because the designators are rigid, each designates just the object it actually designates in all possible worlds in which that object exists. But in that case there is no world for which (using our language) ‘The honeybee = Apis mellifera’ can possibly be false: in every world, ‘the honeybee’ and ‘Apis mellifera’ designate the same thing, that thing each designates in the actual world: the honeybee. Now, clearly no parallel argument can support the necessity of a statement like ‘The honeybee = the insect species that is typically
298
JOSEPH LAPORTE
farmed for honey’, any more than a parallel argument could support the necessity of ‘Hesperus = the brightest celestial object, apart from the moon, in the evening sky’, or, to use Kripke’s favorite example of a contingent identity statement, ‘Ben Franklin = the inventor of bifocals’. ‘Hesperus = the brightest celestial object, apart from the moon, in the evening sky’ is true, because Hesperus is indeed brighter than any other object in the evening sky, excluding the moon. But this did not have to be the case. Hesperus might have been overshadowed by some other still brighter object. In worlds in which that is the case, the sentence ‘Hesperus = the brightest celestial object, apart from the moon, in the evening sky’ is false. The sentence is not true in all possible worlds. It is not necessarily true.6 Just as such familiar statements involving concrete objects are only contingently true, the above statement ‘The honeybee = the insect species that is typically farmed for honey’, which involves kinds, is, for similar reasons, true but not necessarily true. With respect to those worlds in which the bumblebee is the usual creature to farm for honey, the sentence ‘The honeybee = the insect species that is typically farmed for honey’ comes out false: for in those worlds, the species that is typically farmed is the bumblebee, and obviously the honeybee is not identical to that. The reason that no argument parallel to that supporting the necessity of ‘The honeybee = Apis mellifera’ can support the necessity of a statement like ‘The honeybee = the insect species that is typically farmed for honey’ is that, as we have seen, the second designator in the latter statement doesn’t pick out the honeybee kind in every possible world, but rather picks it out in some worlds and not in others, so it isn’t rigid. Since one of the designators in the identity statement does not rigidly designate its kind, there is no reason to take the statement to be necessary, even if it is true. Having drawn a parallel between ‘the honeybee’ and ‘Hesperus’, on the one hand, and ‘the insect species that is typically farmed for honey’ and ‘the brightest celestial object, apart from the moon, in the evening sky’ on the other, we are in a better position to see whether artificial-kind terms allow such a contrast. It will be noticed that the rigid/non-rigid distinction found above in natural kind designators seems readily applicable to designators of arti-
RIGIDITY AND KIND
299
ficial kinds. The bachelor kind is apparently rigidly picked out by ‘bachelor’ but non-rigidly picked out by the expression ‘the kind most commonly broached in discussions about analyticity’. The bachelor kind, or bachelorhood, couldn’t have failed to be the bachelor kind, or bachelorhood. But the bachelor kind could have failed to be the kind most commonly broached in discussions about analyticity. Philosophers might have preferred to appeal to ‘spinster’ and spinsterhood in illustrating (or trying to illustrate) analyticity. If philosophers had done so, ‘the kind most commonly broached in discussions about analyticity’ would not have designated the bachelor kind, though it happens to do so as things are. There are worlds, then, with respect to which ‘the kind most commonly broached in discussions about analyticity’ designates the bachelor kind, and others, such as the spinsterhood-preferring world, in which the expression designates some other kind. Thus, that expression does not rigidly designate the bachelor kind. ‘Bachelor’, by contrast, does. Similarly, ‘soda pop’, ‘soda’, and ‘pop’ all rigidly designate the soda pop kind; but ‘the beverage my uncle requests at Super Bowl parties’ only accidentally designates the kind. The connection to necessity is straightforward. ‘Soda = the beverage my uncle requests at Super Bowl parties’ is true but not necessarily true, since the second designator is not rigid; it refers to milk, or juice in some worlds, and in those worlds, the sentence is false. ‘Soda = pop’ is, on the other hand, necessarily true, since it is true and both designators rigidly designate soda pop. I cannot agree with various suggestions, then, that non-natural, or nominal kind terms stand in contrast to natural-kind terms over rigidity, just as descriptions like ‘the inventor of bifocals’ contrast as non-rigid designators with names, such as ‘Ben Franklin’.7 The proper contrast over rigidity is that between non-rigid descriptions for kinds (either natural or artificial), on the one hand, and rigid names/descriptions for them on the other. Before continuing, let me pause to discuss a natural line of objections to the proposal above. It might appear that my proposal about the rigidity and non-rigidity of certain terms for kinds is committed, in a way that corresponding proposals about terms for concrete objects are not, to controversial metaphysical claims concerning the existence of various kinds.
300
JOSEPH LAPORTE
Suppose, for example, that in addition to a soda kind, which my uncle happens to request, there is a beverage my uncle requests at Super Bowl parties kind (call it the ‘BMURASP-kind’). The latter kind is instantiated, in any possible world, by the beverage my uncle requests at Super Bowl parties in that world. So in this world it is instantiated by the soda kind, soda being what my uncle requests, but it is not identical to that kind; in other possible worlds, where my uncle prefers other beverages, it is instantiated by milk, or apple juice. The soda kind, on the other hand, is of course not instantiated by milk or apple juice in any possible world. Now, it might be objected that ‘the beverage my uncle requests at Super Bowl parties’ actually designates the BMURASP kind rather than the soda kind. If this is so, then the expression ‘the beverage my uncle requests at Super Bowl parties’ would, it seems, rigidly designate its kind. For in every possible world, something is a member of the BMURASP kind just in case it is the beverage my uncle requests at Super Bowl parties. Hence, it might appear that ‘the beverage my uncle requests at Super Bowl parties’ is a rigid designator after all, and it does not non-rigidly designate soda pop.8 If all of this is so, then the sentence ‘Soda = the beverage my uncle requests at Super Bowl parties’ is, contrary to initial appearances, false. For in order to be true, soda would have to be identical to BMURASP, and I have just observed that this is not the case: soda and BMURASP are distinct kinds.9 What to say? Well, the first thing to observe is that, contrary to the above suggestion, the metaphysical issue about kinds behind the objection pertains as well to concrete objects. So the challenge that this metaphysical issue presents to my account of kind designators’ rigidity faces accounts of concrete-object designators’ rigidity as well. Consider ‘the president of the USA’. This description apparently designates Clinton non-rigidly. But suppose there are, in addition to human persons, “office persons” (to use a term introduced by Alan Sidelle 1992; see also his 1995). An office person is an individual who is constituted at any given time by the man or woman in a given office, much as a river is constituted by the water particles in it. So “Prez” is an individual currently composed by Clinton, though in 1790 Prez was composed by George Washington. When someone
RIGIDITY AND KIND
301
leaves office, Prez does not die, but rather comes to be constituted by the person who has just sworn in to be the new U.S. president. Prez is not identical to any man or woman in office any more than a river is identical to the water particles in it, or a statue is identical to its clay. Now, if there is such an object as Prez then, as Sidelle points out (1992, pp. 417, 419), ‘the president of the USA’ might rigidly designate Prez, rather than non-rigidly designate Clinton. For in every possible world featuring Prez, Prez is the president of the USA. Of course, one need not recognize “unusual” objects like office persons in one’s ontology. But neither must one recognize “unusual” kinds, like the BMURASP kind. The point that the above discussion of office persons is supposed to make is that if claims about the reference of kind designators become tangled in metaphysical issues of the variety in question, so do claims about reference to concrete objects. My application of rigidity and non-rigidity to kind designators does not introduce any special problems not faced by the widely accepted application of rigidity and non-rigidity to concrete-object designators. So I have argued. But there may still be a question as to whether some fundamental flaw in ascriptions of rigidity and non-rigidity to kind terms and concrete-object terms both has been uncovered. I do not believe that such a flaw has been uncovered. It seems evident that we do use some expressions to refer to kinds nonrigidly. The species of insect that I name five sentences down is the honeybee (and not some artificial kind instantiated by different species in different worlds – I never name such a thing), the kind of bread this bag is fashioned to hold is French bread, and so on. If there are unusual kinds, we might want to conclude that an expression like ‘the kind of bread this bag is fashioned to hold’ or ‘the beverage my uncle requests at Super Bowl parties’ has more than one use, but not that there is no non-rigid use of such an expression.10 If ‘the beverage my uncle requests at Super Bowl parties’ has different uses, then it can be used to refer to the unusual kind BMURASP or to soda, depending on the context. Such multiple use is not troublesome. It attends, for example, expressions like ‘the honeybee’. ‘The honeybee’ may, as is observed above (see
302
JOSEPH LAPORTE
note 5), be used to designate the honeybee kind, as in ‘The honeybee is a kind essentially marked by its . . . ’, but it can also be used to refer to a particular insect, as when a speaker relates an encounter with one: ‘The honeybee stung Martha when she swatted it’. The bearing of these observations upon the modal status of sentences should be clear. I have argued that the sentence ‘Soda = the beverage my uncle requests at Super Bowl parties’ is contingently true. Obviously any discussion of the modal status of a sentence must begin with an interpretation of the sentence, and here I have assumed a particular use for ‘the beverage my uncle requests at Super Bowl parties’. I have assumed that this expression is used to non-rigidly designate whatever actual beverage my uncle requests, soda, as it happens to be. Given such a use, the sentence is contingently true.11 I conclude, then, that my division of kind designators into a rigid and a non-rigid camp is not threatened by metaphysical troubles concerning unusual kinds. More generally, I conclude that the charge that on my preferred account of rigidity, all kind designators end up rigid, seems unfounded. If the charge were true, then my admission of artificial kind designators into the rigid camp might face trouble. As things are, no good reason for barring artificial-kind designators from rigid status has yet been found.12 There is, however, yet another common suggestion for why artificial kind designators cannot be rigid. According to that suggestion, rigidity is what does the work of putting meanings “out of the head,” to use Putnam’s phrase, and as such rigidity secures the reference of a kind term to a particular real, as opposed to a nominal, kind: hence, artificial-kind terms like ‘bachelor’ cannot be rigid, as they designate nominal kinds.13 This account of rigidity’s work seems to be fairly widespread. Thus, Schwartz discusses the new theory according to which “natural kind terms like ‘gold’, ‘tiger’, and ‘water’, are indexical or rigid. This means that contrary to traditional theories, the meaning of such a term is not given by a conjunction or cluster of properties semantically associated with the term” (1978, p. 566).14 Ronald de Sousa suggests agreement, calling attention to the recent restoration of Aristotelianism and rejection of the empiricist doctrine that natural kinds have “conventionally fixed intensions,” or “nominal definitions”, remarking that “the crucial
RIGIDITY AND KIND
303
prop for this restoration is the notion of rigid designators applied to natural kind names. Rigid designators allow us to believe in essences without having to know what they are” (de Sousa 1984, pp. 566– 567). If rigidity is indeed distinctive of kind terms that designate real kinds, setting them apart from those that refer by way of nominal definitions, then ‘pencil’ and ‘bachelor’ cannot be rigid, at least assuming they name nominal kinds: so the above account of rigidity, according to which artificial-kind terms are rigid, is unsuccessful. (See, in addition to the above, Putnam 1975, p. 234, discussed below; Schwartz 1980, pp. 196f.; 1977, pp. 37ff.; Macbeth 1995, esp. pp. 265–266.) This objection can be effective only if a proponent of the view that kind terms rigidly designate abstract objects is unable to distinguish the way artifact terms secure their referents from the way natural-kind terms secure their referents. That is, on the proposal in question, rigidity applies to all kind terms, those naming real kinds as well as those naming nominal kinds, but the objection is that rigidity is precisely the device that distinguishes some of those terms, those naming real and hence natural kinds, from those naming nominal and hence artificial kinds: the former are distinguished by their rigidity. A proponent of the challenged view of rigidity is thus under obligation to show that she still has the resources to account for the reference of ‘water’, on the one hand, to a real kind and ‘bachelor’, on the other hand, to a nominal kind, despite the rigidity of both terms. A proponent must answer the question: If rigidity is not what secures a natural-kind term’s reference to a kind with an unknown underlying real essence, what does? Fortunately, this challenging question admits a ready answer. What secures a term’s reference to a kind with an unknown underlying essence is not rigidity, but rather the mechanism marked by the causal theory of reference.15 According to the causal theory of reference (CTR), sample items are used to “anchor” a term. A term is used to refer to whatever underlying essence a paradigmatic sample has.16 It may help to imagine a kind’s baptism taking place under artificially ideal circumstances. In Kripke’s words, “If we imagine a hypothetical . . . baptism of the substance, we must imagine it picked out as by some such ‘definition’ as, ‘[X] is the substance instantiated by the items over there, or at any rate, by
304
JOSEPH LAPORTE
almost all of them’ ” (1980, p. 135). The “almost all” qualification is there to ensure that the foreign matter (dust, other chemicals perhaps resembling the paradigms, etc.) that will typically be found in a sample representing a kind is discounted. So on the causal theory, a term like ‘whale’ is linked to its referent by causal contact with individual whales. ‘Whale’ was baptized by a (probably informal) dubbing act: “the term is to refer to that kind of thing” (the dubber points to some whales). It is thus the causal theory that allows ‘whale’ to designate a hidden essence. On the causal theory what makes a thing belong to the extension of ‘whale’ will not be properties like having a fish-like appearance or exhaling a large, visible spout of vapor that a speaker associates with whales, but rather underlying properties and relations that guarantee sameness of kind (to paradigm samples). The essence of whalehood will not be nominal. As one writer explains, “CTR . . . allow[s] the element on which reference depends to be unknown. . . . We do not stabilize a term’s reference by associating certain properties with continued use of the term” (Leplin 1988, p. 497). This stands in contrast to a term like ‘bachelor’, which is not causally grounded in sample bachelors. The reference of ‘bachelor’ is determined, rather, by conditions established a priori to mark bachelor status: unmarried malehood, and so on. These conditions amount to something like a nominal essence for bachelorhood.17 Natural kind terms can get by without such analytic definitions because “actual things . . . have played a certain causal role in our acquisition and use of the terms,” thereby grounding our reference (Putnam 1983, p. 73). It is the causal theory that explains how a term like ‘whale’ is different from a term like ‘bachelor’. There is no reason both terms cannot be rigid, referring to the same kind across all possible worlds, though only the former term has its reference determined by causal contact with paradigm samples of the relevant kind. No appeal to a discrepancy over rigidity is needed to distinguish the two cases. The differences are certainly respected on my preferred account of rigidity above. It seems to be a running together of different elements of the Kripke-Putnam theory of reference, in particular the running together of rigidity with the causal theory, that has led to
RIGIDITY AND KIND
305
the belief that rigidity is what causes terms to refer to real, and not nominal, kinds. Putnam seems to have compounded the confusion noted here by supposing that causally grounded terms are indexical, because, he says, their reference depends upon what sample items around here turn out to be like. This attribution of indexicality seems to have been a mistake, for reasons Burge (1982, pp. 102–107) specifies at some length, but it is not necessary to discuss the mistake here. I mention talk of indexicality only to point out that the above confusion of the causal theory with rigidity can also take the form of a confusion of “indexicality” with rigidity. This was evidenced above in Schwartz’s claim that natural-kind terms are “indexical or rigid” (see note 14 and accompanying text). Putnam falls into the same confusion. “It should be clear . . . that Kripke’s doctrine that natural-kind words are rigid designators and our doctrine that they are indexical are but two ways of making the same point” (1975, p. 234). Actually these do not at all seem to be two ways of making the same point. Rigidity has little to do with how a term gets hooked up to its referent, whether this is by way of samples, as with “indexical” terms, or by description, etc.; rather, rigidity has to do with whether the term refers to the same entity in all possible worlds. There is apparently no reason that kind terms Putnam takes to be non-indexical, such as ‘hunter’ and ‘bachelor’, cannot refer to the same thing in all possible worlds just as well as terms he takes for indexical, like ‘water’ and ‘tiger’.18 No good reason has been identified for withholding rigid status from terms for artificial kinds. That the account of rigidity under consideration extends rigidity to artificial-kind terms is no strike against it. On the contrary, it seems apt to hold rigidity applicable to artificial-kind terms like ‘bachelor’ as well as to natural-kind terms. It does not appear that the account in question makes for too much rigidity.
III
A hurdle has been passed. On the proposed account, rigidity does not seem to extend to inappropriate kind designators. The fear that too many designators might count as rigid is allayed. However, a
306
JOSEPH LAPORTE
threat from the other direction must also be addressed: the account must not be such that too few expressions are counted as rigid designators. Kripke’s argument for the necessity (in case of truth) of theoretical identities such as ‘Water = H2 O’ rests, of course, on the terms ‘water’ and ‘H2 O’ being rigid. Yet a number of writers have, in recent years, independently cast doubts on whether natural-kind terms and/or theoretical expressions typically come out rigid on any account of rigidity that extends to kinds. It is worth investigating these doubts. One prominent skeptic, Helen Steward (1990, pp. 390ff.), describes a situation in which H2 O is drastically affected in its superficial properties: it becomes opaque, pink, and solid. Now, she says, this opaque, pink, solid stuff would not be the same substance that H2 O composes in the actual world. So ‘H2 O’ is not rigid: it doesn’t refer to the same substance in all possible worlds, because it refers to an opaque, pink, solid substance in some worlds, while referring to the quite different substance H2 O actually composes in other worlds such as our own. Steward does not offer much in the way of criteria for distinguishing substances. Not having much of an idea how to delimit them myself, I am not in a position to dispute her claim that ‘H2 O’ would name, in the world described above, stuff of a different substance than the substance composed of H2 O in the actual world. So ‘H2 O’ does not, for all I know, name the same substance in all possible worlds. But should that make it non-rigid? That a term fails to rigidly designate a substance does not indicate that it fails to rigidly designate some other sort of entity. Some names (‘water’, for example, if Steward is right) might pick out the same substance in every possible world. These would rigidly refer to a substance (taken as an abstract entity). But other names might rigidly refer to a chemical composition (say H2 O), even if they do not rigidly refer to a substance (whatever that might be), as Steward supposes. Others might rigidly refer to a species, an occupation, a marital status, and so on. In each case, the objects that instantiate the kind vary from world to world. But the kind itself is the same, from world to world. Unless some reason can be adduced for saying that substances can be referred to rigidly, but that other kinds can’t, it seems unprincipled
RIGIDITY AND KIND
307
to discriminate on that basis. And Steward offers no such reason. Hence, her claim that ‘H2 O’ is not rigid seems to be an unpromising one. Another animadversion that merits attention is launched by C. J. Bolton (1996, p. 148). Bolton argues that theoretical expressions like ‘H2 O’ look more like descriptions than like proper names. And descriptions aren’t always rigid. Some descriptions happen to be rigid, such as ‘the positive square root of 81’. But others, such as ‘the celebrated Irish wit of fin de siècle London’ are not. Names, on the other hand, are always rigid. Kripke claims that rigid descriptions are rigid de facto, while names are rigid by stipulation, or de jure (1980, p. 21n.). Since some descriptions are not rigid, Bolton continues, some argument is needed to show that ‘H2 O’ is rigid, if its rigidity is to be established. But no argument is given in Kripke’s work. So the claim is moot. Bolton wants an argument that ‘H2 O’ is rigid. Now, Kripke does offer an “intuitive test” for discriminating between rigid and nonrigid designators (Kripke 1980, pp. 48–49; see also Sidelle 1989, p. 65). Kripke, of course, focuses primarily on object designators, leaving it up to the reader to apply his treatment to kind designators. The rigidity of ‘Bill Clinton’ is evinced by the negative answer to the question “Could Bill Clinton have failed to be Bill Clinton?” The non-rigidity of ‘the president of the USA’ is, on the other hand, revealed by the positive answer to (the true reading of) “Could the president of the USA have failed to be the president of the USA?” Consider, now, two kind designators, one rigid and one nonrigid: ‘the honeybee’ and ‘the insect species that is typically farmed for honey’. “Could the insect species that is typically farmed for honey have failed to be the insect species that is typically farmed for honey?” receives an affirmative answer: of course it could have failed to be the insect typically farmed for honey, since bumblebees could have been farmed for honey instead.19 But “Could the honeybee have failed to be the honeybee?” receives a negative answer. Surely the honeybee kind could not have been some other kind, such as the bumblebee kind, any more than the property redness could have been the property blueness. If, in any possible world, there exists a kind that isn’t the honeybee kind (in that world), then
308
JOSEPH LAPORTE
we would say it is not what we discuss as “the honeybee kind” at all. To apply this test to the theoretical kind term ‘H2 O’ we must answer: could H2 O (the kind) have failed to be H2 O (the kind)? It is hard to see how anyone could imagine an affirmative answer to be true.20 So ‘H2 O’ seems rigid. We might run a similar inquiry for ‘water’. Could water (the kind) have failed to be water (the kind)? It seems not. So ‘water’ would seem to be rigid. A complication arises: is ‘H2 O’ rigid de facto or de jure? Claims to de facto rigidity are supported by essentialist intuitions, which may be what Bolton and others want an argument for. Moreover, even if ‘H2 O’ is rigid de jure,21 there are other expressions that Kripke takes for rigid that certainly seem rigid de facto if they are to be counted rigid at all: ‘the species with X DNA’ is an example. Suppose tigers have X DNA structure. Whether the above description is rigid may seem to depend on whether tigers possess their DNA structure essentially. If species’ DNA structure is contingently possessed, just as striped species’ stripedness is contingently possessed, then the above description would seem no more rigid than ‘the large, striped feline species’: it will designate different species in different possible worlds, rather than designating the tiger species in all possible worlds. Is the truth of Kripke’s claim that a theoretical identity statement like ‘The tiger = the species with X DNA’ contains two rigid designators thus dependent on DNA structure being an essential trait of the tiger species? No. There is a particular object, a kind, that the statement’s description is intended to rigidly designate: namely, the kind instantiated in any given possible world by just those organisms possessing X DNA structure. If the description fails to grasp this single referent in all possible worlds, some other expression will have to replace it. Kripke discusses a similar complication involving the description ‘C-fibers’, dismissing it as a non-problem: “the point is unimportant; if ‘C-fibers’ is not a rigid designator, simply replace it by one which is, or suppose it used as a rigid designator in the present context” (1980, p. 149). We can invent an expression to rigidly designate, de jure, the intended kind: ‘X DNA kind’ would suffice, or just the description above, used in a special sense. When a description is made rigid by stipulation like this, its rigid status
RIGIDITY AND KIND
309
will admit of no interesting dispute. Of course, whether ‘The tiger = X DNA kind’ would, even if it contains two rigid designators, be true is another question. But the sentence would be necessary if true, which is what rigidity is supposed to guarantee. Neither Steward nor Bolton has given us reason to doubt the rigidity of the terms they discuss. A final argument to be considered is raised by Ghiselin. According to Ghiselin, We scientists do not attach a name to a class, then discover the defining properties which are its essence, but rather redefine our terms as knowledge advances. Therefore the view of Kripke (1980) and his followers (see Schwartz, 1977) that natural kind terms are, like proper names, “rigid designators,” should be dismissed as nugatory, and with it the accompanying essentialism
(Ghiselin 1987, p. 135). Ghiselin is apparently right to place the above proponents of the rigidity view in the camp of those who hold the discovery view described. Kripke strongly suggests the discovery view when he urges that “scientific discoveries of species essence do not constitute a ‘change of meaning’ ” (1980, p. 138). Schwartz, whom Ghiselin also mentions, suggests sympathy for that view as well: “In the philosophy of science, adherents of the new theory of reference oppose the . . . view that new discoveries become part of new definitions of terms” (1977, p. 35). But, though proponents of the view that natural-kind terms are rigid happen to be proponents of the discovery thesis, the two doctrines must be distinguished. Let’s assume that the discovery thesis is false, and that our natural-kind terms are constantly being redefined, so that the precise kind a term refers to continues to change as the term’s meaning continually changes. In that case, we should not expect a kind term to rigidly designate the same kind before the rise of modern science as it designates after the rise of science. But at any given time it could still rigidly designate a single kind: that is, at any given time in the history of the English language, a kind term might designate, in each possible world, the same kind, though the term may not at other times designate that kind since its meaning varies from time to time. Indeed, Thomas Kuhn develops just this position in his (1990). So the mere evolution of meaning does not destroy the rigidity thesis. A term need not keep its meaning over time in order to be rigid at a time.
310
JOSEPH LAPORTE
Ghiselin’s objection does not succeed. But it is of some service, since it belies a popular doctrine often associated with the rigidity of kind terms: it suggests that our conclusions about what essences various kind terms rigidly designate are not really discovered to be true, as is often thought. Ghiselin’s observations about the advance of scientific terminology in biology22 are supported, for example (contrary to what Boyd suggests: 1992, p. 142), by the history of the attempt to determine the essence of species. Countless pages have been devoted since Darwin to the search for the nature of biological species, but it appears that we are hardly much closer to a resolution now than we were in Darwin’s time. One anthology (Ereshefsky 1992) offers over a half dozen different definitions from the recent literature. And the reason for the competition seems to be that there simply is no answer as to what the defining properties of the prescientific category “species” are. Darwin saw this as clearly as any modern author: “It is really laughable to see what different ideas are prominent in various naturalists’ minds, when they speak of ‘species’,” Darwin wrote in an 1856 letter to his friend J. D. Hooker. “It all comes, I believe, from trying to define the indefinable” (Darwin 1888, vol. 1, p. 446). It is not unlikely that scientists will settle upon one or another definition for ‘species’, but that isn’t to say that they will discover how we have been using the term all along. It’s largely a question of how we want to decide to use the term. “Realistically, the use of the term ‘species’ will be determined as much by historical and sociological factors as by logic and biological considerations,” as two well-known systematists confess (de Queiroz and Donoghue 1988, p. 335). Similar words apply to higher taxa, examples of which include oak, cat, and mammal. There are competing taxonomic schools, each giving a different answer about what the essence of such groups is, and there doesn’t seem to be a truly correct way to arbitrate the dispute. “Having whatever information and methods one wants,” the scientist L. Van Valen asks, “what criteria should be used to incorporate the results into a classification? This question lacks an objective answer. . . . That is why there has been war in systematics” (1989, p. 100). However the wars turn out, the resulting claims that this or that feature, or “essence” is what distinguishes the oaks or the cats cannot be said to have been discovered: rather, conclu-
RIGIDITY AND KIND
311
sions reflect decisions to adjust the meanings of our natural-kind terms. We refine them continuously, bringing them up to date, as we progress in knowledge. So while Ghiselin fails to show natural-kind terms to be nonrigid, he does put his finger on an important and fallacious doctrine often associated with rigidity. Helpful as this and other such criticisms may be, no such argument seems to indicate that rigidity is not applicable to natural kind or theoretical-kind designators, to the tune of important consequences concerning the necessity of true theoretical identity statements.23
NOTES 1
A couple of clarifications are in order here: first, a rigid designator designates the same object in all possible worlds as it is used in the actual world, not as it is used in other possible worlds in which the object gets picked out: for although we identify objects in other worlds by our own names, natives of some of these worlds use other names. Second, ‘Hesperus’ is rigid because it picks out Hesperus (Venus) in all worlds that contain Hesperus. In worlds not containing Hesperus, it suffices here to say that the designator fails to name anything other than Hesperus. Salmon (1980, pp. 32ff.) distinguishes between more than one understanding of a rigid designator that conforms to that requirement, but it is not necessary to further specify here. 2 Just what sort of abstract entity might be designated by a natural kind term is a matter of dispute, one that for the purposes at hand does not require a resolution: besides the view that ‘whale’ names the whale kind (see, e.g., Mondadori 1978) there is the position (espoused by, e.g., Boër 1985) that it is the property or attribute whales have in virtue of being whales that is designated. I will speak, for simplicity, of the designated entity as a kind. 3 See also Monte Cook (1980), who balks at the prospect of granting rigid status to terms like ‘bachelor’, and who accordingly attempts to offer an alternative understanding of rigidity on which terms like ‘cat’, but not terms like ‘bachelor’, come out rigid. Cook’s account seems mistaken to me for a number of reasons, though it would take me too far afield to elaborate here (for some discussion see, e.g., Boër 1985, pp. 131ff.; LaPorte 1997). I hope to show here that artificialkind terms should not be denied rigidity, so that the alternative account of rigidity to which Cook appeals in order to avoid attributing rigidity to such terms is unneeded. 4 Both Schwartz and Macbeth discuss natural kind terms as predicates here, rather than as singular terms. The intent of both is apparently to discuss natural kind terms in general, however, so it appears that there is no intention to exclude from their observations kind terms used as singular terms. In any case, the issue
312
JOSEPH LAPORTE
here is natural kind designators, so I will discuss the criticism that my proposal for kind term rigidity makes all kind designators rigid, which destroys any contrast between rigid and non-rigid kind terms. Replies related to the one I will offer apply to the relevant claim about predicates in particular. 5 ‘The honeybee’ can be used as a singular term for a kind, as in ‘The honeybee is a social animal.’ Of course, ‘the honeybee’ has other uses: cf. ‘The honeybee over there stung my friend.’ 6 Even if ‘Hesperus = the brightest celestial object, apart from the moon, in the evening sky’ is not necessary, it will nevertheless be necessarily true of the object that is the brightest in the evening sky, Venus, that it is identical to Hesperus. Thus, Kripke’s limited claim: “If ‘a’ and ‘b’ are rigid designators, it follows that ‘a = b’, if true, is a necessary truth. If ‘a’ and ‘b’ are not rigid designators, no such conclusion follows about the statement ‘a = b’ ” (1980, p. 3, Kripke’s emphasis). It should be noted that on some accounts, ‘Hesperus = the brightest celestial object, apart from the moon, in the evening sky’ would be ambiguous as a result of ‘the brightest celestial object, apart from the moon, in the evening sky’ being ambiguous: on these accounts such descriptions sometimes rigidly designate and sometimes non-rigidly designate. An ambiguity of this sort in the contained description would give the sentence in question two readings, one necessarily true and one contingently true. Kripke rejects this alleged ambiguity in (1980, pp. 59–60n.), but even if the ambiguity exists, identity statements involving just rigid designators are distinguished from those involving designators that are at least sometimes used non-rigidly in that the former do not have a contingently true reading, while the latter do. 7 Putnam suggests this contrast (1975, p. 265), as does, again, Schwartz (1980, p. 196). There may be some parallel between artificial kind terms and descriptions, on the one hand, and proper names and natural kind terms on the other, inasmuch as the causal theory of reference might apply to both members of the latter pair, and to neither member of the former pair (see below). But these parallels do not reflect rigidity. 8 I do not suggest that anyone I have cited would hold that kind designators would all be rigid on my account for this reason. On the contrary, it seems that these opponents have simply overlooked descriptions like ‘the beverage my uncle requests at Super Bowl parties’ when making the claim that accounts like my own give rise to universal rigidity for kind designators. 9 I am grateful to an anonymous referee for suggesting the above criticisms. 10 See also note 19. Along the above lines Sidelle argues that ‘the president of the USA’ can be used rigidly or non-rigidly for particular concrete objects even if Prez exists (Sidelle 1992, pp. 422, 423, 428). 11 In the same way, Kripke insists that names with multiple bearers present no problem to his claims about the truth conditions of sentences containing, say, ‘Aristotle’. To speak of ‘the truth conditions’ of a sentence . . . it must be taken to express a single proposition – otherwise its truth conditions even with respect to the
RIGIDITY AND KIND
313
actual world are indeterminate. Thus ambiguous words or homonyms . . . must be read in a determinate way . . . , indexicals must be assigned determinate reference, syntactic ambiguities must be resolved, [etc.] (1980, p. 9). “In practice,” says Kripke, “it is usual to suppose that what is meant in a particular use of a sentence is understood from the context” (p. 9). 12 It has been suggested to me that my thesis appears to be that simple, or simply stated designators, such as ‘the honeybee’, and ‘humankind’ are rigid, while complex ones like ‘the insect species that is typically farmed for honey’ are not. This is not my thesis. The non-rigid kind designators I present do have a form something like ‘the kind that . . . ’, where some accidental property properly ascribed to the kind is used to pick it out. But there is no simple split over rigidity between complex expressions on the one hand and simple ones on the other. Just as ‘humankind’ rigidly designates the human kind, ‘bald-human kind’ rigidly designates the bald-human kind, ‘bald-happy-human kind’ rigidly designates the kind instantiated in any possible world by bald, happy humans, and so on. One candidate for an expression that rigidly designates bachelorhood is ‘unmarried-eligible-male-hood’. Whether or not the expression rigidly designates bachelorhood, it does rigidly designate unmarried-eligible-male-hood. 13 The contrasting of real and nominal essences harks back, of course, to Locke (1975). Roughly, a nominal essence consists of properties speakers associate with a kind, and a real essence consists of an underlying constitution responsible for the observable properties. A real kind has its membership determined, naturally enough, by a real essence, and a nominal kind by a nominal essence. 14 Bringing indexicality into the discussion obscures the issues somewhat, as I will explain below, but in any case Schwartz’s point about rigidity is rendered clear by his explanation: “This means that. . . . ” 15 Kripke 1980 and Putnam 1977, 1975 are, of course, the leading champions of this theory. Devitt and Sterelny 1987, pp. 70–75 offer a clear exposition and helpful discussion of the theory as it is applied to natural kinds; see also Devitt 1996, pp. 167ff. It is because the causal theory is what does the work in moving essences out of the head that it has come under fire on that score from various quarters: see, e.g., Douven and van Brakel (forthcoming), Nersessian (1984), and Shapere (1982). 16 There is also a second component of the causal theory: the second component explains not how a term gets hooked up to its referent, but how the name gets passed on, and used by those with no contact with the referent. The first component alone guarantees that reference is not determined by properties associated by speakers with the kind, so discussion of the second component is not required here. 17 This has occasionally been disputed in the case of ‘bachelor’, but for the sake of sparing the popular foil we can assume that tradition is right about this. 18 Perhaps it should be noted, to quell any doubts about the matter, that Putnam intends to apply ‘rigid’, following Kripke, to any term that “refers to the same individual in every possible world” (1975, p. 231); he is not introducing a novel
314
JOSEPH LAPORTE
use for the term. Simple confusion about the relation between rigidity and the causal theory seems to explain why he believes that terms whose meaning is “in the head,” like ‘hunter’, are not rigid, though causally grounded terms like ‘water’ are (see, besides the references immediately above, 1975, p. 265). 19 This intuitive test reinforces the claim that there is indeed at least (if not at most) a non-rigid reading of ‘the insect species that is typically farmed for honey’. That expression cannot, then, be nothing more than a rigid designator for an “unusual” kind of the sort discussed last section. 20 It is important to keep in mind when applying the “test” to kind designators that what is designated by a tested expression is a kind, so the question will ask whether a certain kind could have failed to be a certain kind: this reveals whether a term designates its kind in all possible worlds. It is irrelevant whether the individuals instantiating a particular kind could have failed to instantiate it. 21 It is plausible to suppose that this is the case. For the question ‘Is it the case that H2 O might not have been H2 O?’ does not seem to be asking of something that is just picked out by a description whether it could have failed to possess a property given by description; rather, it is apparently asking whether a thing (the H2 O-kind, or H2 O-hood), could have failed to be itself. In this way, ‘H2 O’ contrasts with ‘the liquid that has chemical composition H2 O’, which seems de facto rigid, if it is rigid. 22 Van Brakel (1992), Aune (1994), LaPorte (1996), and others have made similar observations concerning kind terms from chemistry, as well. 23 Earlier drafts of this paper were substantially improved by comments from Bruce Aune and an anonymous referee. For discussion of various ideas in this paper I thank Margaret Speas and Maribel Romero.
REFERENCES Aune, Bruce (1994): ‘Determinate Meaning and Analytic Truth’, in G. Debrock and M. Hulswit (eds.), Living Doubt, Netherlands: Kluwer Academic Publishers, pp. 55–65. Boër, Steven (1985): ‘Substance and Kind: Reflections on the New Theory of Reference’, in B. K. Matilal and J. L. Shaw (eds.), Analytical Philosophy in Comparative Perspective, Dordrecht: D. Reidel, pp. 103–150. Bolton, C. J. (1996): ‘Proper Names, Taxonomic Names and Necessity’, The Philosophical Quarterly 46, 145–157. Boyd, Richard (1992): ‘Constructivism, Realism, and Philosophical Method’, in John Earman (ed.), Inference, Explanation, and Other Frustrations, Berkeley: University of California Press, pp. 131–198. Burge, Tyler (1982): ‘Other Bodies’, in A Woodfield (ed.), Thought and Object, Oxford: Oxford University Press, pp. 97–120. Cook, Monte (1980): ‘If “Cat” is a Rigid Designator, What Does it Designate?’, Philosophical Studies 37, 61–64.
RIGIDITY AND KIND
315
Darwin, Charles (ed.) (1888): The Life and Letters of Charles Darwin, ed. Francis Darwin, 2 vols. New York: D. Appleton and Co. De Sousa, Ronald (1984): ‘The Natural Shiftiness of Natural Kinds’, Canadian Journal of Philosophy 14, 561–580. De Queiroz, Kevin and Michael J. Donoghue (1988): ‘Phylogenetic Systematics and the Species Problem’, Cladistics 4, 317–338. Devitt, Michael (1996): Coming to Our Senses: A Naturalistic Program for Semantic Localism, Cambridge: Cambridge University Press. Devitt, Michael and Sterelny Kim (1987): Language and Reality: An Introduction to the Philosophy of Language, Oxford: Basil Blackwell. Donnellan, Keith (1983): ‘Kripke and Putnam on Natural Kind Terms’, in Carl Ginet and Sydney Shoemaker (eds.), Knowledge and Mind, New York: Oxford University Press, pp. 84–104. Douven, Igor and J. van Brakel (forthcoming): ‘Can the World Help Us in Fixing the Unique Reference of Natural Kind Terms?’, Journal for General Philosophy of Science. Ereshefsky, Marc (ed.) (1992): The Units of Evolution: Essays on the Nature of Species, Cambridge, MA: MIT Press. Ghiselin, Michael (1987): ‘Species Concepts, Individuality, and Objectivity’, Biology and Philosophy 2, 127–143. Kripke, Saul (1980): Naming and Necessity, Cambridge, MA: Harvard University Press. Kuhn, Thomas (1990): ‘Dubbing and Redubbing: The Vulnerability of Rigid Designation’, in C. Wade Savage (ed.), Minnesota Studies in the Philosophy of Science 14: Scientific Theories, Minneapolis: University of Minnesota Press, pp. 208–318. LaPorte, J. (1996): ‘Chemical Kind Term Reference and the Discovery of Essence’, Noûs 30, 112–132. LaPorte, J. (1997): ‘Essential Membership’, Philosophy of Science 64, 96–112. Leplin, Jarrett (1988): ‘Is Essentialism Unscientific?’, Philosophy of Science 55, 493–510. Locke, John (1975): An Essay Concerning Human Understanding, ed. Peter H. Nidditch. Oxford: Clarendon Press. Macbeth, Danielle (1995): ‘Names, Natural Kind Terms, and Rigid Designation’, Philosophical Studies 79, 259–281. Mondadori, Fabrizio (1978): ‘Interpreting Modal Semantics’, in F. Guenther and C. Rohrer (eds.), Studies in Formal Semantics, Amsterdam: North-Holland Publishing Co., pp. 13–40. Nersessian, Nancy (1984): Faraday to Einstein: Constructing Meaning in Scientific Theories, Dordrecht: Martinus Nijhoff. Putnam, Hilary (1975): ‘The Meaning of “Meaning” ’, in Mind, Language and Reality, Cambridge: Cambridge University Press, pp. 215–271. Putnam, Hilary (1977): ‘Is Semantics Possible?’, in Steven P. Schwartz (ed.), Naming, Necessity, and Natural Kinds, Ithaca: Cornell University Press, pp. 102–118.
316
JOSEPH LAPORTE
Putnam, Hilary (1983): Realism and Reason, Cambridge: Cambridge University Press. Salmon, Nathan (1980): Reference and Essence, Princeton: Princeton University Press. Schwartz, Steven P. (1977): ‘Introduction’, in his Naming, Necessity, and Natural Kinds, Ithaca: Cornell University Press, pp. 13–41. Schwartz, Steven P. (1978): ‘Putnam on Artifacts’, The Philosophical Review 87, 566–574. Schwartz, Steven P. (1980): ‘Formal Semantics and Natural Kind Terms’, Philosophical Studies 38, 189–198. Shapere, Dudley (1982): ‘Reason, Reference, and the Quest for Knowledge’, Philosophy of Science 49, 1–24. Sidelle, Alan (1989): Necessity, Essence, and Individuation: A Defense of Conventionalism, Ithaca: Cornell University Press. Sidelle, Alan (1992): ‘Rigidity, Ontology, and Semantic Structure’, The Journal of Philosophy 89, 410–430. Sidelle, Alan (1995): ‘A Semantic Account of Rigidity’, Philosophical Studies 80, 69–105. Steward, Helen (1990): ‘Identity Statements and the Necessary A Posteriori’, The Journal of Philosophy 87, 385–398. Van Brakel, J. (1992): ‘Natural Kinds and Manifest Forms of Life’, Dialectica 46, 243–261. Van Valen, L. (1989): ‘Metascience’, Evolutionary Theory 9, 99–103. Wiggins, David (1994): ‘Putnam’s Doctrine of Natural Kind Words and Frege’s Doctrines of Sense, Reference, and Extension: Can They Cohere?’, in Peter Clark and Bob Hale (eds.), Reading Putnam, Oxford: Basil Blackwell, pp. 13– 41.
Department of Philosophy Bartlett Hall University of Massachusetts at Amherst Amherst MA 01003-0525 USA