Notes on Kripke's "Naming and Necessity," Lecture I, Part 2 of 2 (pp. 54-70)
1. Kripke on Wittgenstein and the meter stick (pp. 54-57):
Wittgenstein famously claimed that we cannot say of the standard meter in Paris that it is one meter long. This is because it was used to define the length equal to one meter. In other words, the competence to use the measurement “one meter” was nothing more than the competence to measure something against the standard meter in Paris. Since you cannot measure the length of the standard meter against itself, there is no procedural competence that could make sense of the claim that it is equal to one meter.
Kripke doesn’t like this, and he tries to undermine the point by changing the terms. Instead of
(1) Stick S is one meter long,
Kripke claims that the length equal to one meter cannot vary over time. So he changes (1) to
(2) Stick S is one meter long at time t0.
Then he claims there are two possible uses for such a sentence. The first one is to “give the meaning” of “meter.” This is the usage that Wittgenstein refers to in reference to (1). Which is to say, for Wittgenstein, that (1) does not actually say anything about the length of S, but only tells us how the phrase “one meter” is to be used. Wittgenstein would not use (2) in this way, because it would require some independent way of remembering the length of S at time t0.
It’s clear that Kripke does not want to use (2) to fix the meaning of “one meter.” He says (2) can, however, be used to “fix the reference” of the unit of length. In this way, (2) can present the length of one meter as an “accidental property” of S. This allows a person to claim that, if heat had been applied to S, it would not have been one meter long.
This is true, Kripke claims, because there is “an intuitive difference” between the phrases “one meter” and “the length of S at time t0.” This is intuitive to us, of course, because we don’t define “one meter” this way. If, however, we regard (1) as fixing the meaning of “one meter,” then (2) does not function as Kripke intends.
Kripke does not want to accept this. Instead, he says the phrase “one meter” is “meant to designate rigidly a certain length in all possible worlds, which in the actual world happens to be . . .” Yet, if (1) gives the meaning of “one meter,” then “one meter” refers to the length of S come what may. Perhaps we can say that “one meter” refers to the length of S in all possible worlds. Kripke's only basis for rejecting this is apparently the fact that we define “one meter” differently, making it intuitively hard for us to think of it as the length of any particular stick.
For us, the meaning of “one meter” should not change depending on the length of any stick. But that was not true for Wittgenstein and others who lived at that time. For us, the length of “one meter” is defined differently.
As with the case of the meaning of “Aristotle,” Kripke’s discussion of “one meter” focuses on names and descriptions of things which we cannot identify ostensively. Yet, his general arguments are about things that we can and do point to in our daily lives. There is an incongruity here.
Now, on p. 56, Kripke tips his hand: What he’s really after in all of this is a way of driving a wedge between a priori truth and necessary truth. He claims that a person who uses (1) to “fix the metric system” will know a priori that (2) is true. Yet, he claims, (2) only expresses a contingent truth about the relationship between S and one meter. Therefore, he says, there can be contingent a priori knowledge. He says the “metaphysical status” of S is not that it is equal to one meter . . . even if we define “one meter” that way? Ugh . . . He is using our definition to say that another definition is unimaginable.
He says “Water boils at 100oC” is another contingent a prior truth. But how is this a priori? We do not define the celsius system by the boiling point of water. If we did, then it would not be a contingent truth.
2. Kripke returns to Hesperus on p. 58:
Imagine if there was an original baptism, of the following sort:
(B) I shall use “Hesperus" as the name of the heavenly body appearing in yonder position in the sky.
He observes that the location of Hesperus in the sky is an accidental feature, and that Hesperus would still be Hesperus if it had occupied a different celestial position. This requires that the use of “the heavenly body appearing in yonder position in the sky” is not given as the definition of “Hesperus,” of course. Rather, that description is being used to point to something which is determinable in some other way—namely, by sight. It is that experiential communication which fixes the meaning of “Hesperus,” and that is what allows us to say that it would still be Hesperus if it occupied a different position. What we mean is that it would have been the same experienceable body (it would have allowed for roughly the same experiences) if it had occupied a different celestial position.
On p. 59, Kripke accuses Frege of confusing two different senses of the term “sense”: first, the sense of a designator is its meaning; second, it is a way of fixing a reference. Frege says both (the meaning and reference-fixing of a designator) are given by a definite description. Kripke says no, “A description may be used as synonymous with a designator, or it may be used to fix its reference,” but these are not the same thing. He says this is evident by the ordinary distinction between two kinds of definition. I suppose he means descriptive and ostensive definitions. But then, this is just to appeal to two kinds of knowledge: by acquaintance (direct experience) and by description (indirect experience).
Kripke says that the meaning of a designator can be to fix its reference without a definite description. This raises an interesting issue. The ostensive definition found in B relies on experiential knowledge. The question, then, is whether any definite description can fully replace that experiential knowledge. Could we have descriptive knowledge of Hesperus such that no experiential knowledge of the planet was necessary to say anything about it? Or is experiential knowledge required to ground our talk of Hesperus? If descriptive knowledge cannot fully replace our experiential knowledge, then the meaning of “Hesperus" can never be reduced to a description.
Would we say that experiential knowledge of Hesperus allows us to rigidly designate the name “Hesperus”? Memory is fallible. This was Wittgenstein’s argument. If I am relying on my experiences of Hesperus to regard the name as referring to the same thing in all possible worlds, I might wonder if I know what I’m talking about. This is why Wittgenstein says there is no such thing as a private language: a way of fixing a meaning without external regulation.
We want to say that our ability to refer to Hesperus transcends our ability to remember any particular experiences of it—except we must be able to connect the name to some experience (even an indirect one). So we can ground our naming in indirect experiences . . . But if it is an indirect experience, it can be described. The possibility of naming via indirect experiences means that the meaning of such a name is reducible to a description. It is only names of intrinsically direct objects of experience that cannot be reducible to descriptions.
3. On establishing a distinction between a priori and necessary truth based on the function of fixing a reference:
Kripke states this plainly on p. 63, assuming (for the sake of argument) that the cluster theory of names (associated with Searle and Strawson) is correct: "A man might know [that “Aristotle" had a particular disjunction of properties] a priori in some sense, if he in fact fixes the reference of 'Aristotle' as the man who did one of these things. Still it won't be a necessary truth for him.” So fixing a reference is an action A establishing a relation between a phrase p and an object O, where p can be a disjunctive description of properties P, such that the attribution of P to O is known a priori; and yet, A does not make the attribution of P to O necessary.
This does not seem to make sense. If the attribution is known a priori, then it does not depend on anything being the case. So how could we say it was metaphysically contingent?
If “metaphysics" is just a matter of semantic meaning, and the meaning is established by fixing a reference, then there is no difference between a priori truth and metaphysical necessity. Kripke wants to say there is a sense of meaning—a metaphysical sense—that is ostensively fixed without necessary attachment to any particular description. But this is to regard meaning as empty of content. It seems only empty terms, like “this" and “that,” function this way. (Presumably that is why Russell said only such terms were proper names.)
Kripke’s argument could be seen as approaching a general argument about the arbitrary nature of signs. Regardless of how a word is defined (ostensively or by description), it is always the case that it could have been defined differently. The relation between a word and its referent is never necessary. Therefore, we might say that analytic truth (truth by meaning alone) is always contingent. However, this is obviously not where Kripke wants to end up, because he regards analytic truth as both a priori and necessary. He wants to say that proper names are not arbitrary designators in some way: Once a baptism is made, it refers to that object no matter how the object is otherwise named. So, Kripke would say, Aristotle would still have been Aristotle even if he had been called Thanos. Such a claim amounts to saying
(R1) An object O designated by a proper name is O under any description.
This makes sense as a basic tenet of realism, and it is what Kripke might call a definition of metaphysical necessity. Kripke thinks R requires making a distinction between a priori and necessary truth, but that just doesn’t seem to make sense. Kripke wants to reject a second basic tenet of realism:
(R2) For any property P and any object O, such that "O is P” is knowable a priori, then "O is P" is true in all possible worlds.
Maybe Kripke resists R2 because he thinks it is possible to have a priori knowledge of objects without knowing their necessary properties. We must be able to fix a referent without knowing its necessary properties, but in that case we cannot have a priori knowledge about its properties.
4. A formal version of the cluster concept theory:
Kripke’s prefaces a formal treatment of the cluster concept theory by presenting an explicitly anti-philosophical attitude on p. 64: “[The cluster concept theory of names] really is a nice theory. The only defect I think it has is probably common to all philosophical theories. It's wrong. You may suspect me of proposing another theory in its place; but I hope not, because I'm sure it's wrong too if it is a theory.” So Kripke does not claim to be promoting a theory of names!
(1) To every name or designating expression 'X', there corresponds a cluster of properties, namely the family of properties 𝞅 such that A believes '𝞅X'.
He is willing to accept this as a definition. However, he says the following two theses are false:
(2) One of the properties, or some conjointly, are believed by A to pick out some individual uniquely.
Not sure yet why Kripke will reject this, but I think it is true. The family of properties 𝞅 must include properties which relate X to particular objects of experience. Otherwise 1 would be true only of universals, and not particulars. (Though even then, 2 would be true in so far as universals are individual concepts.)
(3) If most, or a weighted most, of the 𝞅's are satisfied by one unique object y, then y is the referent of 'X'.
This is a statement of sufficiency, so that there are sufficient but not necessary properties. This still can work, so long as we are still including properties that relate X to particular objects of experience.
He criticises Strawson for taking a democratic approach, and not giving some properties more weight than others. Kripke prefers to weight properties, comparing them to holdings in a corporation. He then claims the stockholders can vote on whether a name refers:
(4) If the vote yields no unique object, 'x' does not refer.
(5) The statement, "If X exists, then X has most of the 𝞅's" is known a priori by the speaker.
(6) The statement, "If X exists, then X has most of the 𝞅's" expresses a necessary truth (in the idiolect of the speaker) .
Kripke then says we can reject (6) on the grounds that the reference of X can still be affixed to the same individual even if they did not have most of the 𝞅’s. He mentions Aristotle again (of course), but never considers those properties which should be given the most weight: such that he was born of particular parents with more or less the same DNA. Is there a possible situation in which Aristotle does not satisfy this description? I think not.
I might be accused of attaching necessary, and not merely sufficient, conditions on the designation. However, all I am doing is naming two properties (one rather strict, concerning the identity of his parents; the other less so, as indicating some but not all of the same DNA) as having a majority of the weight. Certainly that is allowed, given that Kripke advocates a weighting system.
Kripke might object that the name “Aristotle" was used long before people knew about DNA. That is true, though they knew of heredity, and so they knew of DNA indirectly (through other descriptions). And they presumably would have observed the same rule that I am indicating, in so far as they would not believe that Aristotle could have been a mouse.
Lewis discusses something like this in his paper on counterpart theory. Let’s say Aristotle had a brother, Thanos. In that case, there is a possible world in which Aristotle and Thanos did not exist, but in which their parents had a different child, Dionysus, who has as much in common with Aristotle as with Thanos. Does “Aristotle" refer to Dionysus? Does “Thanos” refer to Dionysus? Or should we say that neither Aristotle nor Thanos is Dionysus?
If the scales are slightly tipped in Thanos’ favour, Lewis would say that Dionysus is a counterpart of Thanos, and not of Aristotle. Kripke doesn’t like the idea of counterparts, so he would say that Dionysus is Thanos in this world. Yet, even though Dionysus is more like Thanos than Aristotle, Dionysus still uniquely exhibits most of the properties that define Aristotle. Therefore, Kripke might have us say that Dionysus is also Aristotle. Lewis avoids this problem by claiming, first, that the relation between Dionysus and Thanos is that of a counterpart, and second, by defining counterparts such that a counterpart cannot be of two different individuals within one world (unless they are identical). We do not necessarily need counterpart theory to avoid the dilemma. We could just make an adjustment to (3), so that
(3*) If most, or a weighted most, of the 𝞅's are satisfied by one unique object y, then y is the referent of ‘X’, unless y uniquely satisfies most (or a weighted most) of the conditions for a referent of 'Z', where X ≠ Z.
On p. 67, Kripke returns to Wittgenstein’s example of Moses, considering the possibility that all of the descriptions are false, and observes two possible interpretations of the biblical story: one is that the story is pure legend, in which case no such person existence; the other is that the story is a false account of a real person. He switches to the example of Jonah, because (he says) Biblical scholars agree that Jonah was a real person, even though the account of him in the Bible is not accepted as true. He says “there are reasons for thinking this was about a real prophet.” But in that case, the stories in the Bible are simply not given the weight given to those other reasons. This does not pose a challenge to the cluster theory.
Kripke says there is a problem related to how the name “Jonah" is referentially fixed in the Bible. But that is a problem for historians, isn’t it? How do historians establish such things? It is not clear that this poses a metaphysical challenge.
5. On non-circularity
Kripke ends with a brief discussion of the following criterion, which seems unobjectionable:
(C) For any successful theory, the account must not be circular. The properties which are used in the vote must not themselves involve the notion of reference in a way that it is ultimate' y impossible to eliminate.