Notes on Kripke, "Naming and Necessity," Lecture 3, Part 1 of 2 (pp. 111-136)
Kripke says we can imagine a possible world in which the Queen of England was actually born of the Trumans (in the usual manner) and passed off as a member of the Royal Family. But, he asks, can we imagine that this woman the Queen was actually born of the Trumans?
He says no, but why not? What if he met her. Could he not wonder if she were really a Truman? He says no. He can imagine if some other woman had been born of the Trumans and passed off as the Queen, and perhaps that woman could greatly resemble the actual Queen, but they are not the same.
In short, he is saying that when he imagines the possibility that Queen Elizabeth is actually a Truman, he is not imagining anything about Queen Elizabeth at all, but about somebody else. Now, remember that this is exactly what he said he did not want to do back in Lecture 1. He criticised Lewis’ counterpart theory because he (wrongly) thought it prevented him from making just the sorts of counterfactual considerations he should be able to make about the Queen of England.
Now, in trying to justify this predicament, he refers to the importance of origins. Parents are an origin, and “anything coming from a different origin would not be” the same object. Okay, but that is a description! If the reference of a name is fixed by the object’s causal history prior to naming, then that is part of the meaning of the name. The name “Queen Elizabeth” means a person with a particular origin, according to Kripke. But, as Kripke would observe, it is not meaningless to say that Queen Elizabeth was born of the Royal Family, and we can get as specific as we want in that regard. But for Kripke, this is as necessary as saying that Hesperus is Phosphorus.
He refers to the relationship between identity and matter. Would a wooden table be the same table if it were made from a different block of wood, or of a different substance? He says no, and in a very obvious way he is right. The question is, can I imagine a possible world in which this table was made out of a different block of a wood? Of course I can. I just imagine the same process that created this table, but with a different block of wood. If Kripke will insist that I am not imagining the same table then I will refer him to Lewis’ counterpart theory: I am referring to a possible world in which this table is made out of a different block of wood! That is something he says we should be able to do. That was a key part of his first lecture.
Of course, with a person, the substance is connected to their biological structure, and so—as I noted in my previous notes, and as Lewis notes in his paper on counterpart theory—issues with biological identity are central to our ability to refer to people. But this is an argument for a descriptionist view of naming, not against it.
Kripke returns to Kant’s analytic/synthetic distinction with respect to gold (p. 117). Specifically, Kant says sentences are such that the predicate of an analytic judgment is already “contained in” the subject of the sentence. So, for example, “Gold is a yellow metal” is analytic in so far as the concept of a yellow metal is contained in the concept of gold.
Kripke’s discussion of this makes reference to the periodic table, which was not invented until well after Kant’s lifetime. So we should recognise that Kant’s use of "gold" was likely not the same as ours. Kripke also brushes aside any notion that we should distinguish between a phenomenological image (perhaps Sellar’s manifest image) and a scientific image of the world, but acknowledges that metal is not a good hook for that argument.
The question he asks is, could we imagine that gold is actually blue instead of yellow? He goes full Cartesian and asks us to imagine a trickster demon which has been tricking the world into thinking that gold is yellow when it is in fact blue!
If the truth came out, and we started to see that gold is actually blue, we wouldn’t conclude that gold does not exist. We would conclude that “yellow" is not an intrinsic or necessary property of gold. Now, I am not sure this would have worked for Kant, because maybe he really did take “gold" to mean (among other things) “a yellow metal.” Maybe any yellow metal was gold for Kant, and he did not take it to refer to a particular substance at all.
Anyway, Kripke’s argument is bizarre, because of course color is dependent on various factors that affect perception. To say “gold is a yellow metal” (today, not in Kant’s time) is to say something about how gold appears in standard lighting conditions. To imagine that gold was actually blue would be to imagine that it was a different substance, and Kripke says that is impossible. (I am rejecting the trickster demon scenario, because it is ill-defined.) Indeed, how is “Gold is actually blue” any different from “Queen Elizabeth is actually not of Royal Blood?”
Kripke goes on for a bit about tigers and the meanings of common words, and it’s not very interesting. His point is to reject the cluster concept approach to understanding things. He says we should rather go about the normal business of assuming things like tigers for a natural kind. Of course, he appeals to a cluster of concepts to justify that move. He refers to “past experience” in which things which fit such and such a cluster of descriptions to determine that tigers probably also form a natural kind. He is speaking counterfactually here, because—he says—we already know that they do form a natural kind. He says it is up to science to find out for sure. Science could determine that there are many kinds of tigers—which is to say, it is possible (we can imagine it, or stipulate counterfactually) that the name “tiger" refers to different kinds of things. Indeed, can we ever rule out that possibility?
Kripke’s general claim is a reformulation of Locke’s distinction between primary and secondary properties. Primary properties are scientifically measurable, and secondary properties are how they are perceived. Color, according to Kripke, is a secondary property: we can be deceived about them because our perceptions can be mistaken (or we can be mistaken about our perceptions). Material substances (whatever might be necessary to the identity of a person or object, such as DNA, hunks of wood, etc.) are primary properties, and they are discoverable by science. Kripke’s theory is that names refer to objects by virtue of their primary properties, not their secondary properties, even if the people giving the names cannot make the distinction.
What is to prevent Kripke’s trickster demon from tricking us about primary properties?
Kripke says it was possible (for people who did not know that tigers are a natural kind) to stipulate counterfactual worlds in which there are many kinds of tiger. It was possible for them to imagine that tigers were robots or angels or anything in between. So it was possible for them to imagine that some of the things they called “tiger" were robots while others were alien species from another planet, while others were varieties of kangaroo, and others were hallucinations, and others were . . . The list is as long as our imagination will allow. For people in that epistemic situation, each tiger they see could be a different kind of thing. They use the term “tiger" because, despite that possibility, the word “tiger" picks out a qualitatively similar experience. Now, a person in such an epistemic situation, thinking that each tiger could be an entirely different sort of thing, would not say, “There could be many different kinds of tiger.” Of course they could say that, but that is not the thought I am after. I’m rather after the thought that says, “The word ‘tiger' might refer to an indefinitely heterogenous set of things that have nothing in common whatsoever.” That is not the same as saying that there are many kinds of tiger.
To say that there are many kinds of tiger is to say that “tiger" picks out a particular category which includes a variety of kinds. We can say there are many kinds of pepper, or many kinds of book, in the sense that we can categorise our knowledge of peppers and books into classes. But in the case I’m talking about, tigers cannot be so categorized, because the person in the right epistemic situation has no way of doing so. They can only imagine that each tiger they experience is a different sort of thing. In other words, they can imagine that each tiger they experience is different with respect to all of its properties, not just its secondary properties. And in that case, they would not say “there are many kinds of tiger.” They would rather say that “tiger" is a purely subjective term that does not pick out any particular kind of thing in the real world.
Let’s say a person in that epistemic situation develops a science and discovers that what they call tigers can be grouped into two sets: those that are biological descendants of a common ancestor to us, and those that are animatronic vehicles that an alien species has created to spy on our planet, and which just happen (pure coincidence) to strongly resemble the biological species.
Would a person in this new epistemic situation say that there are evidently two kinds of tiger: one a biological species and the other an animatronic vehicle for an alien species? Maybe so, but it could depend on the ratio. If the two sets (biological and animatronic) were of similar size, then probably yes. But what if millions of tigers were biological creatures, and only one or two were animatronic vehicles? In that case, we would likely say that tigers were biological creatures, and the animatronic vehicles were not really tigers at all. They only looked like tigers. I am not claiming that we have to put it that way, but only that it would make sense. The reason is because most discussions of tigers will be of the biological variety. People who discuss the animatronics could still use the word “tiger,” but they would want to qualify it that so people did not think they were talking about the much more common, biological creatures. On top of that, people in this epistemic situation could reasonably conclude that they had been wrong to refer to the animatronic vehicles as tigers, and that they had been tricked!
Why is this important? Because in both epistemic situations—the first, with no scientific knowledge of things going by the name “tiger”; and the second, with a scientific distinction between biological creatures and animatronic vehicles—the use of the word “tiger” is determined by what people think about the referents. We cannot say that people have to use the word “tiger" to refer to whatever objects (or whatever set of objects) they referred to when they learned the term. If references were fixed the way Kripke says—in an unbreakable chain from referent to users throughout a community—then people would have no right to claim that they had been wrong about those animatronic vehicles being tigers. They would have no right to say that those animatronic vehicles were not simply a different kind of tiger.
It does not matter that some people can regard those animatronics as a variety of tiger. What is important is that nobody has to. This makes Kripke’s approach profoundly counterintuitive. That is, I think, the best thing one can say about Kripke’s theory.
Since Kripke cannot make infallible claims about our own epistemic situation with respect to the natural world, his approach makes it impossible for us to determine what we are or are not allowed to say (or imagine) about the natural world. His notion of reference requires that we adhere to a standard that is set by properties of the world that are beyond the grasp of our epistemic situation. On Kripke’s view, we simply cannot know if we are using words correctly or incorrectly.
Kripke wants to say that science has the job of making our use of terms more efficient and precise, but his theory does not make that possible. For, on Kripke’s view, a scientist can never say that a particular name was ever inefficient or imprecise. For Kripke, names always and only refer by virtue of the facts that it is up to science to determine. Therefore, science can never disagree with how things are named. All science can do is explain to us how it is that our names refer to the things that they refer to. In the process, science gives us more things to name. What science cannot do is tell us that a particular use of a name is right or wrong.
Consider Hesperus and Phosphorus. In Kripke’s view, those are the same no matter what our epistemic situation happens to be. Therefore, while we might think we can imagine a possible world in which they are different, we are not really imagining that. This is not a function of our epistemic situation, but a function of the fact that Hesperus and Phosphorus are actually the same thing. Science, Kripke would say, can correct our false beliefs about this. But scientific explanations are limited.
What of people who use the names to refer to the qualitatively different experiences (as I mentioned in my comments on Lecture 2)? Science cannot say that they are wrong. It is a fact that the experience in the morning is dissimilar to the experience in the evening. One can imagine possible worlds in which the distant causes of those experiences were the same or different without fixing a reference for those causes.
Indeed, the perception of a planet requires, among other things, light. What is perceived is the light, and not the thing itself. We can therefore imagine that a different celestial object reflected the same light. And since the light is different in the morning and in the evening (because the sun’s light interferes more in the morning), we can say that we are actually naming two physically distinct phenomena. The fact that the light in both cases has (to some extent) reflected off of a far away celestial body is of minor concern. For that reason, we can imagine that the far-away body could have been very different. What is important is the light, not the far away reflective surface.
Can this be right? When we refer to a planet, we are not referring to the light it reflects. Well, that is true in most cases. However, for somebody who does not know that Hesperus is a planet (let alone the same planet as Phosphorus), they think they are referring to a light source.
Imagine a person looking at a mirror that reflects a lamp. They think, “oh look, a light source!” Are they wrong? Not exactly. They don’t know if the light source is shining directly or indirectly on them. They might suppose that it could be a mirror, but their working assumption is that they are seeing the light source directly.
How is this different from people who see light reflected off of Venus and think it is a star? They are seeing a light source indirectly. Is the light the same in the morning and in the evening? Only partially, for in the morning the light is blended with light from the sun. That explains the qualitatively different experience, and it could (hypothetically) justify naming the experience differently. "Hesperus" could identify light that is mostly reflected off a faraway planet. "Phosphorus" could identify light reflected off a faraway planet combined with light from the sun refracted through our atmosphere. In such a case, telling people that Hesperus and Phosphorus are really not stars at all, but one and the same planet, would not have the consequences Kripke wants. For one, the people will wonder why the planet shines like a star. When they understand that they are seeing starlight indirectly, they will reconsider how they think about these things. It is possible but certainly not necessary that they will decide to use "Hesperus" and "Phosphorus" to refer to one and the same planet. They could just as easily use those names to refer to the different experiences of starlight--especially if they are told that the planet is already called Venus, and does not need another name.
Let’s return to the text. On p. 122, Kripke gives a model of how concepts are formed. “The original concept of cat is: that kind of thing, where the kind can be identified by paradigmatic instances.” In what sense can we say that paradigmatic instances identify a kind? Isn’t it redundant to call them “paradigmatic”? What makes an instance paradigmatic?
Kripke has us imagine a possible world where all the places in the world where gold has been found never actually contained gold, but rather a different substance—say, iron pyrites (fools gold). He says that, since iron pyrites is not a chemical element, we would be able to say that what has been found is not a chemical element. However, we would not say that gold is not a chemical element, and this is because what has been found is iron pyrites, not gold. This is per his instructions, of course: He has stipulated a world in which a substance other than gold has been found on earth. Therefore, it is incorrect to conclude from the fact that iron pyrite is not an element that gold is not an element. Fair enough.
Kripke’s initial conclusion is benign. He says, “And so, it seems to me, this would not be a case in which possibly gold might not have been an element, nor can there be such a case (except in the epistemic sense of ‘possible’). Given that gold is this element, any other substance, even though it looks like gold and is found in the very places where we in fact find gold, would not be gold”(p. 125).
In other words,
(C) Given that gold = such and such element, there can be no possible world where gold is not such and such element.
That is a basic definition of gold as a natural kind. As such, I see no problem with it. We should note that (C) is not a scientific discovery. It is not clear that science is in the business of defining natural kinds as such. So Kripke’s next step is curious. He says, "such statements representing scientific discoveries about what this stuff is are not contingent truths but necessary truths in the strictest possible sense”(p. 125).
It is not clear exactly what scientific discoveries Kripke has in mind. There have been many scientific discoveries about gold: its melting point, boiling point, density, relative atomic mass, key isotopes, and other characteristics. These are all scientific discoveries about “what this stuff is.” Would Kripke say that these are all necessary truths? And what would that mean, except to say that we define “gold" as referring to whatever substance has all of those properties?
Kripke says that these are necessary properties of gold, but not a priori knowable properties of gold. And yet, they are a priori knowable in so far as this is how gold is defined. I can look up a scientific definition of “gold" and find this information, and this information is not contingent on any experiences. If it were contingent, it would not be necessary. So, to the extent that the information is necessary to the concept of gold, it is a priori knowable. Why does Kripke say otherwise?
Now Kripke raises the possibility of discovering “fools cats”: cats that resemble mammals on the outside, but have a reptile-like internal structure. These would not be cats, he says. But why not? What if we name them as cats—they are baptised as such. They are cats!
Kripke could appeal to our previous decision to regard “cat" as a natural kind term denoting mammals, but that is to appeal to a description. If reference is fixed by baptism alone, then calling a reptilian cat a “cat" is all that is needed to make it a cat. Kripke has not given us any possible way of doing it differently.
Talking about the molecular nature of a table, he says, "once we know that this is a thing composed of molecules—that this is the very nature of the substance of which it is made—we can't then, at least if the way I see it is correct, imagine that this thing might have failed to have been composed of molecules”(127). Notice he says, once we know that such and such, we cannot imagine it being otherwise. This is just to say that, once we have found a necessary condition for fixing the reference, we cannot use the term to refer to something that does not satisfying that necessary condition. That sounds like an appeal to descriptions.
Regarding water, Kripke makes a sort of utopian conjecture about how water was originally identified: “by its characteristic feel, appearance and perhaps taste.” Interesting that Kripke does not distinguish between freshwater and saltwater. He also does not consider that whatever was originally identified as water might also have included urine, tears and a number of other fluids that we do not consider as water today. Do we need to know what, exactly, was identified as water first? Does it matter what language we are talking about? Presumably the word “water" has a complex origin that cannot be so easily grasped.
Kripke’s analysis requires too much, and offers too little. First, there must have been an initial baptism in which the word was tied to a particular natural kind. Second, there must have been a scientific discovery such that the true essence of the natural kind was revealed. Thus, there was an original birth of the concept of water via baptism, and then it was scientifically discovered that water is H20. Thus we have a rigid designator.
All of the things that went by “water" (and that went by whatever other terms pre-dated the term “water" and helped determine its referential use) may have had H20 molecules, but they also had others—and not simply impurities. There is salt in saltwater, urea in urine and tears and sweat . . . the word “water" could have originally referred to all of these things. So in that case, shouldn’t Kripke say that “water" is not really a natural kind after all, because there are many kinds of water?
Does it matter if all of the kinds of water have H20 molecules? It might, if we want to define water as H20. But we do not have to define it that way. We can imagine a world in which different chemical elements produced the same relations between bodies as occur on earth (whether or not this is physically possible), and in that world, the people whose bodies are like ours (but without H20) interact with a substance that is like water (but without H20). Should we say that they are interacting with water?
But this is just a matter of how we want to define our terms. If we use Lewis’s counterpart theory, then we can say they are acting with water’s counterpart. If we want to regard objects in possible worlds as somehow the same as the objects in the real world (only imaginary versions, I guess), then it’s just a question of how we define “water.” We have no original baptism to refer to, and there’s no reason to think we need one. We can define “water” as H20, or we can define “water" as whatever serves the water role for people like us. Our choice in this matter does not seem to have any philosophically interesting implications.
For pages after this, Kripke goes on about how we fix referents through perception. There is nothing of interest here that I can see. Finally, starting on p. 134, he recapitulates his main points. The first point is to situate his view of names in relation to both Mill and the Frege-Russell tradition. The second point is to warn against confusing "the a priori yet possibly contingent properties carried with a term, given the way its reference was fixed, and the analytic (and hence necessary) properties a term may carry, given by its meaning"(135).
On p. 136, Kripke makes a comment about observational errors. He describes a general situation in which we can believe we have found a new kind, and we baptise it with a new general name; later, we discover that the items so baptised are actually identical to a different, previously named natural kind. We would then say that there was no new kind after all. We would most certainly not proceed to use the new name to refer to the old kind.
How could Kripke explain this? There is no de-baptism rule that says the link between a general name and an object is severed if it is discovered that the object is already linked to another general name. That would make it impossible to create multiple general names for the same kind, and we obviously do not follow such a rule. According to Kripke’s rules of naming, once a group of objects were baptised with the new general name, then that name referred to that kind in all possible worlds. The result of the observational error, once it is revealed, is that the previously named kind now has two names, and we have no principled reason to choose one over the other.
Kripke might accept this, saying that we really have no reason to choose one name over the other, except perhaps force of habit. But that is counterintuitive. We believed that we were naming something new, and once we realised that was not the case, the new name no longer meant what it was supposed to mean. We want to drop the new name, because it has a sense. But Kripke does not accept this fact about general names, and so his view does not reflect how people actually use and think about them.