More Than You Want to Know About
Kripke's
Theory of Names


Some Background on Theories of Names 

Compare sentence (1) and (2)

(1) It was a scientific discovery that Hesperus = Phosphorus

(2) It was a scientific discovery that Hesperus = Hesperus


Intuitively, (1) is true and (2) is false. Yet we might wonder how this can be. If names are just ‘tags’ or ‘labels’ for things out in the world, and if there is no real content to a name except for the thing or object that it corresponds to in the world, then ‘Hesperus’ and ‘Phosphorus’ are simply two tags for the same object. And if this is right, then (1) and (2) should have the same content, since there is only one object—one planet—that is under discussion; there is only one object that each of sentence (1) and (2) are talking about. So how can it be that it was a scientific discovery that this one object is identical to itself? Shouldn’t (1) and (2) both be false?

In order to solve this mystery, some (e.g., Russell) have introduced the descriptive theory of names. This says that a name is not an empty ‘tag’ for objects in the world, whose only semantic content is the object that it is pointing to or indicating. Rather, names are disguised definite descriptions. So, for example, instead of ‘Hesperus’ and ‘Phosphorus’ having the same meaning—viz., the (one) planet that they pick out—they instead stand for different descriptions such as: Hesperus = the brightest star in the evening, Phosphorus = the brightest star in the morning. In this way, the ‘real’ content of (1) and (2) above is expressed by (1*) and (2*) respectively:

(1*) It was a scientific discovery that the brightest star in the evening = the brightest star in the morning.
(2*) It was a scientific discovery that the brightest star in the evening = the brightest star in the evening. 

This honors our original intuitions that (1) is true while (2) is false, thus seemingly providing us with a nice solution to the initial puzzle.

However, Kripke shows how the descriptive theory of names is ultimately unsatisfactory for two main reasons: one epistemological, the other metaphysical (or, more specifically, modal).



Kripke’s Epistemological Argument Against the Descriptive Theory of Names

Kripke has two kinds of epistemological arguments against the Descriptive theory of names: the argument from error and the argument from ignorance. <> 

Argument from Error: If names have no other content than their description—if a name such as ‘Hesperus’, say, means nothing more than ‘the brightest star in the evening’, then it looks like we will never be able to in error about an object’s identifying descriptive properties. For example, if we were to find out that we had been mistaken all along, and that actually there was this other planet, very much like Hesperus, that was the brightest star in the evening, but that our telescopes had been too weak to distinguish the two planets, we would want to say something like (3): 

(3)   Hesperus is not the brightest star in the evening.

But if the name ‘Hesperus’ just means ‘the brightest star in the evening’, we can’t very well say something like (4): 

(4)   The brightest star in the evening is not the brightest star in the evening.

Given the descriptive theory of names, (4) will be false by definition. So it looks like the descriptive theory of names has prevented us from being wrong about our descriptions; we can’t, according to this view, be in error about an object’s identifying descriptive properties, since that description is it’s name—it’s what the name means. 

A suggested fix given by the descriptive theorist is the cluster theory. This theory claims that names aren’t identical to just one description, but to a whole family or cluster of descriptions. In this way, one or a few of the descriptions can be false, so long as the majority of the descriptions hold.

But the argument from error just repeats itself at the level of the whole cluster. If we ever found out that all of the descriptions of a certain cluster wee false, we wouldn’t be able to coherent express this. But intuitively we can. [An example here if I have time.]


Argument from Ignorance: It seems that people can competently use names even if they are ignorant of most or all of an object’s identifying descriptions. For example, some person may know hardly anything about Hesperus, except that it was the planet that his astronomy teacher talked about for some length of time during lecture, during which the student scarcely paid any attention. This would not distinguish Hesperus from any other planet that the teacher might have talked about that day or any other, yet it still seems as if the student can (somewhat) competently use the name and talk about the planet in question. (Notice that even if one has a cluster theory of names, this won’t help, since one can be ignorant of the relevant cluster or family of descriptions and still competently use the name.)

A suggested fix given by the descriptive theorist for this particular problem of ignorance is the borrowed credentials theory. This theory runs as follows: take the individual who knows little to nothing about Hesperus except for the fact that his astronomy teacher talked about it in class. While the student knew very little about Hesperus, the teacher did. The student can borrow the teacher’s knowledge of the appropriate cluster of descriptions for ‘Hesperus’ and in this way, the student can competently use the name, too.

However, despite the suggested fix, the theory can be shown to be inadequate. For it requires that the borrower—in this case, the student—remember and recall where, and on who’s authority, he learned of the name. But, the student could have completely forgotten where he heard of ‘Hesperus’. He might recall that he heard it from somewhere, and that it is a planet of some kind, but this might be the extent of his knowledge. That he can’t remember that his astronomy teacher talked about it prohibits him from ‘borrowing credentials’ from the teacher, yet we would still be inclined to say that he can (somewhat) competently use the name ‘Hesperus’, and talk about the planet.

 

Kripke’s Modal Argument Against the Descriptive Theory of Names 

Consider the following modal claims (5), (6), and (7):

(5)   Hesperus might not have been Hesperus.
(6)   Hesperus might not have been Phosphorus.
(7)   Hesperus might not have been the brightest star in the evening.

Kripke thinks that it is clear that (5) and (6) are false, given that Hesperus is indeed identical to Phosphorus, but that (7) is easily true. Just imagine the case explained above: that we found out that we had been mistaken all along, and that actually there was this other planet, very much like Hesperus, that was the brightest star in the evening, but that our telescopes had been too weak to distinguish the two planets apart. This is enough to make (7) come out true. But nothing, Kripke argues, could make (5) or (6) come out false, given that Hesperus = Phosphorus. And this shows, he continues, that the descriptive theory of names is false. It shows, in other words, that it can’t be true by definition that either Hesperus is the brightest star seen in the evening, or that Phosphorus is the brightest star seen in the morning.

The fact that names can track individuals across counterfactual situations—even counterfactual situations in which the individuals’ (actual) identifying descriptions are false of them—shows, Kripke thinks, an interesting feature of names that the descriptive theory of names leaves out. Namely, that proper names are rigid, or, as Kripke describes them, rigid designators. I will explain more on this below. 

A suggested fix on behalf of the descriptive theorist against Kripke’s modal argument is that the descriptions used as names have an implicit ‘actuality’ operator in them. [More on this later, if you wish...]

But notice that this move is made only in response to Kripke’s arguments, and might thereby be seen as a rather ad hoc move on behalf of the descriptive theorist. Moreover, what the ‘actuality’ operator is doing is simply making a non-rigid description rigid. So insofar as descriptions with the actuality operator are doing the same sort of work as Kripke’s names—i.e., they are functioning as rigid designators—then this move is not really one that shows that descriptive theory is any better than Kripke’s. If anything, it is admission that Kripke is right: there are terms in our language (whether descriptions or not) that function as Kripke claims rigid designators do. Yeah, Kripke.     

 

Kripke’s Positive View: Baptism, Transmission,  and Rigid Designation

Baptism and Transmission: In addition to his arguments against a Descriptive Theory of Names, Kripke also proposes an alternative view about how names get introduced. He claims that while names might start out as descriptions, once the name ‘fixes’ onto an object, the description is dispensable. There is a kind of ‘baptism’, Kripke thinks, when names are introduced. Someone looks at an object, and says something like: “I will hereafter call you ‘Bob’.” Once the name is introduced, and sticks, it can then be passed from user to user with no descriptive content whatsoever. So, contra the (cluster) descriptive theory, even with the added condition about borrowed credentials, Kripke’s theory does not require that the (competent) user of a name need to know where or from whom he heard of the name, or whether or not that source was an educated one. All that Kripke’s theory requires is that there really be a (not necessarily causal) chain linking the user of a name to the object that the name names; the name-user need not know or recall the chain, so long as there is one. 

 
Note: the thing that is baptized need not be causally connected with the dubber, so calling this a ‘causal chain’ theory of names is not quite correct. We can, for example, name abstract objects such as numbers and functions and theories and what-not—things that are abstract, or are not causally efficacious.
 

Rigid Designation: Once a name has been introduced, it is then and thereafter what Kripke calls a rigid designator. That is, it picks out the same individual in all possible worlds (in which that individual exists). In fact, the test for whether a term is a rigid designator is to see whether it will survive modal scrutiny. For example, (using Kripke’s own example) someone other than the U. S. President in 1970 might have been the U. S. President in 1970, but no one other than Nixon could have been Nixon. So, the descriptions ‘the U.S. President in 1970’ is not a rigid designator, while ‘Nixon’ is. In general, then, the test goes like this: if we are wondering whether a certain term ‘X’ is a rigid designator or not, we ask the following counterfactual: “Is it the case that someone other than X could have been X?” If the answer is ‘yes,’ then ‘X’ is not a rigid designator; if the answer is ‘no,’ then ‘X’ is a rigid designator.  


Note: Kripke’s endorsement of rigid designation is not the same as, and does not imply a commitment to, the principle of substitutivity of singular terms. This principle states that singular terms can everywhere be substituted, salva veritate. Kripke does not think this. For one thing, he thinks that a sentence involving two distinct names—two rigid designators that rigidly pick out the same individual in all possible worlds—could express an empirical truth, whereas a sentence involving only one name would not. Compare, for example, ‘Hesperus is identical to Phosphorus’ and ‘Hesperus is identical to Hesperus’. He thinks that the former can be an empricial discovery, whereas the latter cannot. (Or, at least, so he says in N&N. See “A Puzzle About Belief” for (what I take to be) a revision of this claim.) One of the reasons Kripke gives for why co-referring rigid designators may not be everywhere substitutable for each other is that each rigid designator is associated with it’s own historical chain.  

 

Kripke’s hedge: Kripke claims (in Naming and Necessity) that he does not mean to be endorsing a full, positive view. Rather, he thinks it is enough that he merely gestures towards a better picture than the one that the descriptive theorists (or the direct reference theorists, for that matter) have offered up until now.

 

Why Unicorns Could Never Have Been

At the start of his first lecture in Naming and Necessity, Kripke claims that he has a rather surprising view about unicorns—namely, that he thinks that there could not be, nor never could have been unicorns. Even if scientists were to uncover fossils deep within the layers of the highest peaks of the alps, that had the shape of large horses with one horn protruding from their foreheads, this would not, Kripke thinks, be evidence that there were or had been unicorns. Just why this is so, Kripke explains in his Addenda to Naming and Necessity…

Kind terms such as ‘tiger’, Kripke thinks are rigid designators. This means that ‘tiger’ tracks tigers across counterfactuals, and across possible worlds. Moreover, like all rigid designators, they are not equivalent to some identifying description, or a cluster of descriptions. So, we can track tigers across counterfactual situations, completely independent of any physical characteristics. Tigers could have looked very different from how they do now, and some other species could have had the external appearance of tigers (e.g., some other animal could have been cat-like, with orange fur and black stripes, etc.).  But since there actually are tigers, our kind term ‘tiger’ picks them out as they are in the actual world, and this is what fixes the species across possible worlds.
 
In contrast, the term ‘unicorn’ doesn’t pick anything out in the actual world. ‘Unicorn’ picks out a mythical—e.g., merely possible—species, just as ‘tiger’ picks out an actual species. The only description we have of unicorns is just that—a description. And, as Kripke’s arguments against the descriptive theory of names supposedly shows, descriptions are not enough to track individuals (or individual species) across possible worlds. So while there may be creatures in other possible worlds fitting the description of a one-horned horse with magical powers, etc., this description is insufficient for determining whether a certain animal is a particular species or not. In other words, many different species with many different internal structures could have satisfied the external appearances that we associate with unicorns. Kripke claims: “If we suppose, as I do, that the unicorns of the myth were supposed to be a particular species, but that the myth provides insufficient information about their internal structure to determine a unique species, then there is no actual or possible species of which we can say that it would have been the species of unicorns.” [N&N: 157]

The above is a metaphysical thesis: that no counterfactual situation could properly be described as one that involved unicorns. His second point (he claims) is an epistemological thesis.
 

Since the myth only contains descriptions, we cannot say whether something fitting the description is really what the story was about. To use his own example, if there were a story that was about an object with the external, physical characteristics of gold, we cannot conclude whether the story was about real gold or ‘fool’s gold’ or something else entirely that only looked like gold under certain circumstances. Unless there is an actual, historical ‘chain’—a link with the object and the first dubbing ceremony and a transmission of the name down through the ages—there can be no guarantee that the name is connected with any object whatsoever. He claims, “…we cannot say that the unicorns of the myth really existed; we must also establish a historical connection that shows that the myth is about these animals.” [157]   

 

The above reasoning generalizes to proper names as well as kind terms. So, to use Kripke’s example, it never could have been the case that Sherlock Holmes existed. This is so for two reasons: (i) the story is underspecified; the descriptions in the Sherlock Holmes novels are insufficient to track just one individual across possible worlds, and (ii) there is no connection between the name ‘Sherlock Holmes’ and an actual individual whom the detective stories were about. Kripke says: “…granted that there is no Sherlock Holmes, one cannot say of any possible person that he would have been Sherlock Holmes, had he existed. Several distinct people…might have performed the exploits of Holmes, but there is none of whom we can say that he would have been Holmes had he performed these exploits. For if so, which one?” [158] 

 

So, given that stories of non-actual things such as unicorns and Sherlock Holmes are merely descriptive, there is an under-determination problem when we try to see whether there could have been either unicorns or Sherlock Holmes. Since counterfactuals, Kripke thinks, are not underdetermined in this way, it must be that there could not have been and could not be unicorns or Sherlock Holmes.   

 

 

Worries about Kripke’s View on Unicorns and Sherlock Holmes

 

Kripke has apparently two arguments for the claim that neither unicorns nor Sherlock Holmes could have existed: one metaphysical, the other epistemological. However, in each case it seems that his argument rests on contingent features of the stories involving unicorns or Sherlock Holmes.

 

Imagine, for example, that the myths involving unicorns not only gave us external descriptions about magical one-horned horses, but also gave us detailed descriptions of the internal structure of unicorns. The stories may have provided us with a genetic map, say, detailing the DNA and molecular make-up of these creatures. Also, let us suppose that these descriptions of the internal structure of these mythical creatures were simply composites of actual and known molecular particles. We supposedly had enough imagination to combine the idea of a one-horned animal with the idea of a horse, so let us just repeat this creative process at the molecular level. Now, given Kripke’s comments on the essentiality of the internal structure of individuals and individual species, wouldn’t we have enough to ‘rigidly designate’ a fictitious species? His complaints in his Addenda at the end of Naming and Necessity are seemingly addressed by this fix, anyway. For it is not the case that different species with different internal structure could satisfy this detailed description across possible worlds. Given that species (even according to Kripke) are identified by their internal structure, then it seems that once we’ve been given a description of this—even if it is fictional—then this will be enough to get Kripke to admit that it might have been or could be the some fictional creatures existed (again, given that the fictions provided enough information).

 

And we could do the same for Sherlock Holmes. Imagine that we found a locked trunk in Conan Doyle’s attic, wherein we found elaborate, but fictional, descriptions of Sherlock Holme’s genetic code. Would this then be enough to specify across possible worlds a single individual? This would seemingly be enough, anyway, to say that Conan Doyle was writing a story about this [pointing to the DNA code] man, even if the genetic code was pure fiction. At least, we seem to be able to do this with real people (think of a police investigation, where they have the genetic information of the criminal). So it looks like Kripke is in need of an argument of saying why this wouldn’t be enough in a fictitious case. 

 

Also, since in each case (as I’ve set it up) the underlying elements, or the fundamental molecular or genetic make-up, is a composite of things that actually do exist and are furthermore things that are known to exist, then Kripke’s epistemological argument seems circumvented here. For the historical chain of ‘unicorn’ can just be the combined chain of the molecular elements that allegedly constitute unicorns. Historical chains, then, can be inherited from the chains of the various elements that combine to compose the individual in question. We seem to be able to do this with things that actually exist, anyway. For example, suppose I know what Hydrogen molecules are, and how they behave, and how they can bond with other molecules, etc. And suppose I know similar things about Oxygen molecules. But suppose (for whatever reason) I have no idea about H2O. Couldn’t someone say: ‘H2O’ is just two Hydrogen molecules and an Oxygen molecule bonded together in such-and-such a way?
 
Or, more pertinent to the fictitious example, it seems that we could dub a certain molecular combination that has never existed, ‘Bob’. We could be very specific about what would have to happen in order for Bob to exist. Yet since we are so specific, and since all of the elements that Bob is (will be?) composed of already exist, can’t we then use the name Bob?  

It seems we do something similar with sufficiently detailed descriptions of fictional characters. And if so, then it seems we can avoid both Kripke’s metaphysical argument and epistemological argument for the claim that fictional characters could not have existed. In this way, his thesis is not as radical as it might at first seem. For the only real complaint he seems to have against the possibility of unicorns and Sherlock Holmes is that the fictions describing them are not specific enough to determine the truth value of counterfactuals involving the terms ‘unicorn’ and ‘Sherlock Holmes.’ But if the above considerations are right, then the stories of unicorns and Sherlock Holmes could have been more specific, in which case it looks like Kripke would have to admit that both unicorns and Sherlock Holmes could have been.


Final thoughts (inconclusive and unfinished): Underlying these last comments in response to Kripke’s arguments for the claim that there could not have been either unicorns nor Sherlock Holmes is the worry that Kripke’s criticisms of the descriptive theory of names is in tension with his claim about the essentiality of individuals (and individual species). Given what he says about a posteriori identities—e.g., water = H2O, heat = molecular motion, pain = firing c-fibers, etc.—it seems that he needs to maintain something similar about, say, individual humans. So, for example, it may be an a posteriori discovery that you = your genetic code. But given this identity, Kripke will think that it holds necessarily. But then, wouldn’t a description of your genetic code be enough to track you across possible worlds? It certainly seems to pass the rigid designator test (see above). Moreover, it seems we could have these identities stipulated in fictions as well, and then it’s not clear that his metaphysical and epistemological arguments hold much water.

 

[Ok, that’s all for now. More later maybe…]

Page Last Updated: Feb. 7, 2008
Back to Phil 330 Page
Back to Meg's Teaching Page
Back to Meg's Main Page