Philosophy 155: Carnap and Quine Professor Warren Goldfarb Lecture 26 (Final Lecture): December 3, 2012 Inscrutability of Reference

Philosophy 155: Carnap and Quine Professor Warren Goldfarb Lecture 26 (final lecture): December 3, 2012 Inscrutability of reference

Note. Material in square brackets is additional to what was in the lecture as delivered.

I have spent the last lectures going closely over what I take to be Quine’s argument for the indeterminacy of translation in chapter 2 of Word and Object, and some re- sponses to it. A part of the argument is the consideration as to what could possibly be relevant to translation by way of behavioral evidence. Quine wants you to think about what the task is—how you will be able to guard against the idiosyncrasies of the individual linguistic informant, and how you will eliminate collateral informa- tion as affecting the way you translate. All that will restrict what the objective evidence could be.

Indeterminacy has the force that it does for Quine because he takes it that there are no further physical facts that could be relevant, over and above the behavioral things that we have isolated in thinking about what the project of translation is. This is where Quine’s view of language learning, and of us as being socialized (cf. the image of the elephantine bushes), comes in and plays an argumentative role. Once one sees that there are no further physical facts that are relevant, Quine will take it that there are no further facts at all. For there are no further facts: physics tells us what the stuff of the world is. There is no need to think that there might be more, unless you have some sort of explanatory breakdown that leads you to expand your physics. But Quine doesn’t think that we have any such breakdown.

Let us turn now to the 1968 paper “Ontological Relativity.” This is probably the most controversial article that Quine published. In it he focuses attention on a particular aspect of indeterminacy, namely, what he calls “inscrutability of reference”. The thesis is that it is indeterminate what to ascribe as the extensions of the predicates of the person you are translating, and what to ascribe as the referents of his singular terms. He illustrates this with the familiar example of rabbits vs. rabbit stages vs. undetached rabbit parts. But he has some new examples too. E.g., he talks about distinguishing alpha as a type, an abstract sign, from alpha as a token. He claims that we could change our translation so that any occurrence of “alpha” could be parsed one way rather than the other way.

Indeterminacy generally has to do with pairings of sentences with sentences. In- scrutability is what happens when you go to the level of the subsentential parts. When you look at the framing of the thesis of indeterminacy of translation, you are just talking about the existence of different sentence-sentence mappings that are equally consonant with all of the evidence. But when you talk about inscrutability you are asking how it is that you ascribe extensions and referents to subsentential parts. Quine’s idea is that you could even keep the sentence-to-sentence pairings fixed, and that would still not determine how you ascribe extensions and referents to the subsentential parts. For you need to fix the apparatus of singular reference, “apparatus of individuation,” as he puts it: i.e., the apparatus of identity, the quantifiers, and plurals. All of that is additional to just the sentence-sentence pairings.

But it turns out that even that much is not enough. Quine argues in “Ontological Relativity,” in the case of abstract objects, that even if you fix the apparatus of individuation you still get inscrutability of reference. For you can substitute one abstract object for another in certain sorts of ways (e.g., Gödel numbers for formulas). But later he streamlines the argument into a more general argument. This is the argument from “proxy functions.” In “Things and their Place in Theories,” he talks about reinterpreting the language:

the sort where we save nothing but merely change or seem to change our objects without disturbing either the structure or the empirical support of a scientific theory in the slightest. All that is needed in either case, clearly, is a rule whereby a unique object of the supposedly new sort is assigned to each of the old objects. I call such a rule a proxy function. Then, instead of predicating a general term ‘P’ of an old object x, saying that x is a P, we reinterpret x as a new object and say that it is the f of a P, where ‘f’ expresses the proxy function. Instead of saying that x is a dog, we say that x is the lifelong filament of space-time taken up by a dog. Or, really, we just adhere to the old term ‘P’, ‘dog’, and reinterpret it as ‘f of a P’, ‘place-time of a dog’…. The apparent change is twofold and sweeping. The original objects have been supplanted and the general terms reinterpreted. There has been a revision of ontology on the one hand and of ideology, so to say, on the other; they go together. Yet verbal behavior proceeds undisturbed, warranted by the same observations as be- fore and elicited by the same observations. Nothing really has changed.

By permuting the universe and reinterpreting your general terms, you can always give yourself another interpretation of the language where the ontology is different—the items referred to are different. This seems to be an argument that Quine is completely happy with. He writes in Pursuit of Truth, p. 31: Reference and ontology recede thus to the status of mere auxiliaries. True sentences, observational and theoretical, are the alpha and omega of the scientific en- terprise. They are related by structure, and objects figure as mere nodes of the structure. What particular objects there may be is indifferent to the truth of observation sentences, indifferent to the support they lend to the theoretical sentences, indifferent to the success of the theory in its predictions. The point can be accentuated by invoking what I have called proxy functions.

And he gives the same argument. This seems to be his streamlined argument for the inscrutability of reference. The whole proxy-function argument, notice, keeps the truth-conditions of whole sentences the same. The argument is an (attempt at) underlining the artificiality of analyzing sentences further down, i.e., of isolating and interpreting the subsentential parts.

(I have no settled opinion on whether the argument for inscrutabililty should invoke the indeterminacy-like considerations, as in “Ontological Relativity”, or whether the quick proxy-function argument is really sufficient.)

Well, inscrutability of reference upset people greatly. The indeterminacy of translation upset them too, of course; but inscrutability upset them far more. It is one thing, you might say, to attack meaning (the target of the indeterminacy thesis). Quine is quite open about that, and you can barely turn two pages of Quine from very early on without finding yet some other attack on the notion of meaning, or on the “mental myth,” or on propositions, etc. So it is no surprise. But here, in inscrutability, he is attacking the objectivity of reference. And that seems more difficult, and difficult to make compatible with Quine’s own distinction between theory of meaning as bad and theory of reference as good. From early on, e.g., the paper “Notes on the Theory of Reference” in From a Logical Point of View, Quine distinguishes theory of meaning from theory of reference; he finds the theory of reference to be in acceptable shape, while the theory of meaning is not. (The paradigms “—” is true iff — and “—” refers to — give the notions of truth and reference “a peculiar clarity”, Quine says.)

The problem, we might say, starts with this. First, you must realize that indeterminacy and inscrutability are not just about foreign languages. They hold in the home language as well. In fact, even though you speak English I can variously translate you in ways completely consonant with all of your dispositions to speech behavior, variously translate you so as to ascribe to you different ontologies. Ordinarily, Quine says, we adopt the homophonic translation, i.e., equating your words with my words, although there may be special reasons to depart from this. For example, you might on occasion want to invoke the principle of charity, where you depart from homophony in order to render me as saying something more credible. (Twenty- seven years ago, I spent a semester teaching at Berkeley. The next-door neighbors were a bunch of Berkeley women soccer players. They spent a lot of their time on the phone, often outside the apartment, since they didn’t want their roommates to hear. (No cell phones back or even wireless landline phones — they had long exten- sion cords.) So I heard a lot of California-speak in 1984, in particular many re- countings that parties they had attended were “radical.” So I had to translate that in a non-homophonic way. They weren’t attending left-wing events; “radical”, for them, was to be equated with “awesome”, or some such.)

You have to translate the home language in order to specify an ontology. And that seems to threaten a collapse, when you claim that there is inscrutability here. And so we get probably the most dramatic three pages of Quine that he ever wrote, pp. 47-9 of “Ontological Relativity.”

We seem to be maneuvering ourselves into the absurd position that there is no difference on any terms, interlinguistic or intralinguistic, objective or subjective, between referring to rabbits and referring to rabbit parts or stages; or between referring to formulas and referring to their Gödel numbers. Surely this is absurd, for it would imply that there is no difference between the rabbit and each of its parts or stages, and no difference between a formula and its Gödel number. Reference would seem now to become nonsense not just in radical translation but at home.

The answer is to note in what terms we actually do distinguish referents. And so Quine goes on to say,

In these terms [of our own language] we can say in so many words that this is a formula and that a number, this a rabbit and that a rabbit part, this and that the same rabbit, and this and that different parts. In just those words. This network of terms and predicates and auxiliary devices is, in relativity jargon, our frame of reference, or coordinate system. Relative to it we can and do talk meaningfully and dis- tinctively of rabbits and parts, numbers and formulas.

So it is in terms of an already fixed apparatus in the home language that we would distinguish the different possible referents. Quine concludes,

It is meaningless to ask whether, in general, our terms “rabbit,” “rabbit part,” “number,” etc., really refer respectively to rabbits, rabbit parts, numbers, etc., rather than to some ingeniously permuted denotations. It is meaningless to ask this absolutely; we can meaningfully ask it only relative to some background language.

Is there a regress, then? If we can only refer to rabbits rather than to rabbit parts relative to some background language that we specify, what then, we might ask, fixes the referents of the background language? And you raise the question about the referents of the terms in which you’ve given the difference between rabbits and rabbit stages. Suppose I say: I’m going to take a term of his language to refer to rabbits and not to rabbit parts. To do that is to specify the referent in my background, or translating, language. But you might then ask me: What about your term ‘rabbit’ in which you have specified his ontology? Isn’t there a question about whether that really refers to rabbits or to rabbit stages?

Quine concludes, “In practice we end the regress of background languages, in dis- cussions of reference, by acquiescing in our mother tongue and taking its words at face value.” A little later he says, “What makes sense is to say not what the objects of a theory are, absolutely speaking, but how one theory of objects is interpretable or re-interpretable in another.” That concludes what I consider the “dramatic stretch” of “Ontological Relativity:” we end the regress by “acquiescing in our mother tongue and taking its words at face value.” And it is that that people have found very puzzling, if not downright inconsistent.

Among those who have found this the juncture of Quine’s philosophy that simply doesn’t make sense is Hilary Putnam. He says, for example, in “A Comparison of Something with Something Else,”

What Quine says here may seem, however, not to meet the problem. What is it to “acquiesce” in my “mother tongue”? And how can “acquiescing” and taking my mother tongue’s words at “face value” bring it about that there is a relation between my sentence “Snow is white” and a substance out there in the world being a certain color, if no such relation is to be discerned by an impartial observer who speaks a different “mother tongue” (say, an ideal linguist from Mars)?

How can it be that acquiescing in the language brings it about that there is a relation between my sentence ‘Snow is white’ and the stuff out there? That is the problem for Putnam. He goes on for a couple of pages into the difficulties here, and about how it is not analogous to the situation in relativity theory and coordinate systems (as Quine suggests it is). This difficulty … was pointed out a number of years ago in an important paper by Stephen Leeds, but Quine has not, to my knowledge, responded to it. The only solution consonant with Quine’s general position that I can see would be to abandon the geometrical analogy [of coordinate systems] and to say that in the case of my own language calling a sentence “true” is doing no more than reaffirming the sentence. I am not ascribing any property, not even a relative property, to the sentence when I say that it is true; I am just assenting to the sentence. … On this interpretation, to say, as Quine does, that there is only “immanent truth” is—as close as makes no difference—to say il n’y a pas de hors texte.

(Quine as Derrida, in short.) This is his solution, consonant with Quine: “calling a sentence ‘true’ is doing no more than reaffirming the sentence. I am not ascribing any property, not even a relative property, to the sentence when I say that it is true; I am just assenting to the sentence.” Well, this talk of whether one is ascribing a property or not, is very un-Quinean. Putnam’s point can be rephrased: Is there something real about the truth ascription? He wants to make this into the issue: is this a real property? But I find it amazing that Putnam is amazed by this. For Quine has been insistent, all along, that what is involved in thinking about truth is that the Tarski paradigm tells us what truth is—that the only purchase we have on the notion of truth is given to us by the Tarski biconditional. Quine says that over and over again. The idea is that truth is not a transcendent notion—i.e., we cannot define ‘true in L’ for variable ‘L’. Rather, when we understand a particular language, e.g., English, our knowledge of what truth is consists in, or is based on, our appreciation of the paradigm Tarski biconditionals. In that way Quine is saying that to call a sentence true is simply to reaffirm that sentence. That is why I find this a rather curious tone Putnam is adopting; for it seems quite clear from the ear- liest get-go that that is one of the things involved in Quine’s talking about the Tar- ski paradigms in the way that he does. That is Quine’s answer to those who think that the Tarski paradigms are “factual,” or are things we should find out by investigation, or something like that. Quine is taking them, if you like, as definitional of the notion of truth. You have no other purchase on the notion.

Now to make matters more difficult yet—if one is worried about inscrutability applying at home, what it is to acquiesce in one’s mother tongue—you have to notice too that indeterminacy and inscrutability apply to oneself. Quine says,

If there really is no fact of the matter, then the inscrutability of reference can be brought even closer to home than the neighbor’s case; we can apply it to ourselves. If it is to make sense to say even of oneself that one is referring to rabbits and formulas and not to rabbit stages and Gödel numbers, then it should make sense equally to say it of someone else. After all, as Dewey stressed, there is no private language.

In short, what Quine is saying is that we have no special access to a sort of rarefied knowledge of what we ourselves are really referring to. We certainly can’t appeal to some mental act of dubbing that attaches this name to this object; we have to take indeterminacy of translation and inscrutability of reference to apply to ourselves. There is nothing special thing, nothing we have that would cut against this. That is, of course, quite shocking: it looks as if Quine is going all the way down to some sort of solipsism of the moment. If I have to understand what I said five min- utes ago, then I have to be seen as translating my earlier words. Doesn’t this suggest that this does make communication incomprehensible?

Not really. At the key paragraph on p. 48, Quine says that “we can say in so many words that this is a formula and that a number, this a rabbit and that a rabbit part, this and that the same rabbit, and this and that different parts. In just those words.” It is not, if you like, that we cannot answer questions about reference; rather, we need our language to do it. That’s a rather remarkable move on Quine’s part. For then when you see that, you see that all of the old philosophical questions about reference lose their point. The self-knowledge, you might say, is just that ‘rabbit’ refers to rabbits. It’s just disquotational.

Well, the coherence of what Quine is doing here, and of the dissolution, you might call it, of the problem here, has been attacked by many people. There is an article by John Searle in the Journal of Philosophy twenty years ago, which particularly takes up the problem of self-ascription (see the list of Paper Topics). But one also finds even a general question: Has Quine gone too far in dissolving reference? And Put- nam still can’t accept it. In Putnam’s reply to Dreben’s “Putnam, Quine—and the Facts,” in the issue of Philosophical Topics devoted to his philosophy, he writes,

I believe that there is a relation called “reference” between words and parts of the world; if there is no such relation, then equating some of my words with other of my words (or, in the homophonic case, with themselves) cannot create one. It can, however, do something else. Equating my word “cat” with a word W either in my own language or in someone else’s language can make it the case that the indeterminate reference of “cat” is by stipulation to be the same as that of W in each model. The reference of “cat” and W can be so linked that they will float together. And this is exactly how Quine himself at times puts it, e.g.: To say what objects someone is talking about is to say no more than how we propose to translate his terms into ours; we are free to vary the deci- sion with a proxy function. The translation adopted arrests the free- floating reference of alien terms only relatively to the free-floating reference of our own terms, by linking the two. In such a view, there is no fact of the matter as to whether any given object is “really” a quark, or France, or myself! Even the supposed definiteness of “surface irritations” is spurious; there is no fact of the matter as to which objects are really my neurons!

Putnam is taking it that to adopt the inscrutability of reference in this form is to make it the case that “there is no fact of the matter as to whether any given object is ‘really’ a quark, or France, or myself.” I think that that is an extraordinary con- clusion. And it is hard to reconstruct the argument. I can have some vague feeling of why he wants to say that. For if you look at a stretch of my discourse, and there is no fact of the matter as to whether I am referring to rabbits, but here I am occupy- ing a discourse talking about rabbits, then there can’t be a fact of the matter about whether that’s a rabbit. That seems to be the argumentative flow Putnam has in mind.

But is there a real argument that can be developed from this? I don’t think so. The argument seems to take it that there must be some account of your talking, such that the lack of such an account undermines your talking. But Quine is, in a sense, urging that we can talk—we can make assertions—without the pretensions of there being such an account. He said, to a seminar of Harvard faculty members on 11 February 1993: “The ontology we attribute is relative to a translation manual. The ability to claim chairs are as real as can be, despite indeterminacy, is just naturalism.” That is what acquiescing in you own language means: to talk without there being, as a precondition of your talking, some idea that there has to be a philosophical account of what you are saying. Quine went on, “To undercut that [this naturalism] is just to go back to some idea of first philosophy, of Dinge an Sich, of inscrutable objects that yet somehow we want to scrute.”

Gary Ebbs, in his book Rule-Following and Realism (which treats Carnap, Quine, and Putnam at length), puts it in an interesting way. Ebbs is one of the few writers on Quine who actually understand this point. On p. 58 he writes,

Despite the compatibility of alternative reference schemes, many philosophers still feel certain that we cannot use our words to state reference schemes for our own idiolects, unless the references of our words are somehow uniquely determined. But in Quine’s view, this feeling should be dismissed. For there is no scientific perspec- tive from which to question whether we can use the sentences of science to make assertions. Here again we reach the bedrock of Quine’s scientific naturalism. Putnam’s argument appears to be that it wouldn’t be the case that something, x, is a rabbit unless that can be explained by way of explaining the content of the sentence ‘x is a rabbit’. The truth or falsity of x’s being a rabbit relies on first going through an account of the reference of my terms. And that is what I take him to mean when he says, “I believe that there is a relation called ‘reference’ between words and parts of the world; if there is no such relation, then equating some of my words with other of my words (or, in the homophonic case, with themselves) cannot create one.” But that is precisely what Quine is denying. Quine sees it as just part of his naturalism. Naturalism says that we talk and theorize fine, as we do; but Quine’s naturalism also yields a skepticism about accounting for our words in the traditional ways. We can talk of reference via disquotation, but there are no objective facts that give us a reference relation. When I say ‘rabbit’, I refer to rabbits. All that comes to is a certain kind of stipulation. He is deflating the notion of reference. To repeat, Quine says that one is not to expect a factual basis for every putative difference in mental states that can be couched in old-fashioned mentalistic terminol- ogy.

[Note. Putnam’s thinking here also relies on a liberal and unquestioned use of the notion “no fact of the matter”. See Ricketts’ “Roots of Ontological Relativity” for why this is not coherent in this context.]

It is from the standpoint, then, of acquiescing in our language, of taking it for granted, that Quine does his positive work—what I call (following Dreben) “Chapter 3 Quine”, since it reflects the sorts of projects Quine does starting in Chapter 3 of Word and Object. One thing always going on in Quine are the two stances: the criticial stance of doing naturalistic epistemology, which includes the criticism of traditional notions (the sort of thing we see in chapter 2), and the positive stance of chapter 3, in which we go on to do these positive projects. It is important to see which stance he is adopting when he says something. And the positive project is to talk about what things we should admit into our ontology, and what we should leave out; how we should analyze belief statements; how we should regiment such- and-such ways of talking. For Quine this is no different from saying that the physi- cists can do their work.

The coherence of Quine’s philosophy depends on the coherence of this double aspect of what he’s doing. We might say that the dissatisfaction expressed in the feeling that inscrutability applied at home leads to some kind of incoherence, or is a reductio, makes very clear what it is to give up the notions of meaning and reference—or rather, to give up the notion of meaning and to say that the notion of reference is given only by disquotation. For the dissatisfactions with that, or the allegations, of incoherence, are always framed in a language that assumes that there is something more to those notions.

What one finds in Quine's talk about belief and the other attitudes are two things at work. On the one hand, there is the attack on the status on these things coming with the turf of indeterminacy. He even says, in “On the Reasons for Indetermi- nacy,”

To expect a distinctive physical mechanism behind every genuinely distinct mental state is one thing; to expect a distinctive mechanism for every purported distinction that can be phrased in traditional mentalistic language is another. The question whether … the foreigner really believes A or believes rather B, is a question whose very significance I would put in doubt. This is what I am getting at in arguing the indeterminacy of translation.

So you have that argumentative flow, for the scientific disreputability of belief ascription. But you have a whole raft of different arguments appearing in the chapter 3 Quine. In Word and Object Quine embarks on an extended investigation into how we would regiment belief sentences if we really wanted to do so: what objects we would require to make plausible logical sense of them.

We note things about belief statements like their curious behavior with respect to logical inference (e.g. the fact that we can substitute co-referential terms within a sentence and change its truth-value, which makes the positions inside the sentence not open to quantification-thus, they are referentially opaque). This makes regimen- tation difficult. In later essays he is rather shorter with belief sentences; he stresses the ways in which the criteria for how much you can vary the belief- ascribing sentence and still be said to have a correct ascription of the subject's belief, is something that is essentially context-bound. It depends on your purposes in ascribing the belief on that occasion. So one should not expect that you could regiment them. In the later writings Quine often ends up comparing belief statements to indexical expressions like “I”, “here”, and “now”. It is on that basis that he urges that the logically developed form of our theory of the world should not contain such idioms.

So here he is not looking to see whether you can find belief in behavior. We already know that he thinks you can't do that. This is another project, albeit a related project. It is just one of trying to find clear enough uses of the language of belief ascription as a first step towards explication, and as I say he finds it wanting. He there- fore relegates the language of belief ascription to a lesser status-not to be thought of as forming part of science, or of our scientific theory of the world, but useful for everyday purposes. This is what is expressed in §45 of chapter 6, “The Double Stan- dard.” He says, on p. 219,

In the strictest scientific spirit we can report all the behavior, verbal and otherwise, that may under- lie our imputations of propositional attitudes, and we may go on to speculate as we please upon the causes and effects of this behavior; but, so long as we do not switch muses, the essentially dramatic idiom of propositional attitudes will find no place.

Quine is —to use current jargon— an eliminativist. He believes that in the full scientific theory of the world there will be no place for notions like belief. Indeed, I think that he is probably the original eliminativist: the first philosopher to suggest that our ordinary “folk psychology” of belief will wither away in the eventual scientific theory. There is a very nice statement of some of the problems he finds with discourse about belief in his reply to Hintikka in his Schilpp and Hahn volume in the Library of Living Philosophers series. )

About modality, i.e., the philosophical notions of necessity and possibility. Quine has been opposed to those notions from the very start. From his correspondence with Carnap we see his opposition to them. He wrote some famous papers in the 1950s (e.g. “Three Grades of Modal Involvement” in Ways of Paradox). Basically he thinks that they are unclear, and that the philosophical use of them is based upon a mistake. He also thinks that if you use necessity with variables, you are committed to form of essentialism, which again he cannot make sense of. And finally, to show that the modalities are in even worse shape than the propositional attitudes, he emphasizes that there is no need for the notions in everyday life, much less in science.

[Here is a place where the strategy that Quine identifies with the fictitious positivist Ixmann (from “Carnap and Logical Truth”) appears. Quine says: Suppose you want to oppose a claim that science raises certain metaphysical problems. Then what you do is to propose a language that can handle all of the science, but in which those metaphysical concerns cannot be framed. This, he says, is a Carnapian strategy that he applauds: it is the best aspect of Carnap. This is the strategy he will employ with the philosophical modalities. You see that you have languages that are perfectly adequate for science, but which do not employ the modalities.

So much for belief and the modalities. I have said little about the philosophy of logic, except for what we already know: for Quine, the truths of logic have no special status. They are not in a different category from any other scientific truth. In Phi- losophy of Logic he spells out how you specify the truths of logic. (I recommend you read particularly chapters 1, 2, and 7.) As he puts it, “Logic chases truth up the tree of grammar.” What you do to obtain the class of logical truths is to distinguish the logical particles (things like 'and' and 'there exists' and 'not') and then specify the logical truths as those truths that remain true under arbitrary lexical substitutions. So anytime you take a true sentence, and it turns out that whenever you replace any non-logical particle with any other suitable lexical item you get another truth, then the original is a logical truth (as of course are all the truths obtained by such substitution). This amounts to the same thing as the definition given in that masterful textbook, Deductive Logic: a schema is valid just in case all substitutions (interpretations) are true; a sentence is logically true just in case it can be schema- tized by a valid schema. That is the same definition. So a sentence is a logical truth if all sentences with that structure are true.

But this definition, Quine is quick to point out, gives no support to any view of the special status of logic as in some way different from other truths. They are in one sense “free of subject matter,” for you are talking about truths that remain true no matter how you replace the lexical items; but there is no privileged definition of the logical particles. You might as well say that the subject matter of the logical truths is the logical particles. But given that he doesn't think you can do anything except simply to point to a bunch of constructions: there’s no principled way of picking them out. It is true, he says, that the truths of logic “feel” special. But that is expli- cable by the fact that they are “potentially obvious” (a notion we saw as far back as “Carnap and Logical Truth”): every logical truth is either obvious or can be reached from an obvious truth by obvious steps. And of course this relies on the completeness of first-order logic, the fact that there is a complete proof procedure. And here of course he means “obvious” in the garden-variety sense; they are obvious in the sense that “There are dogs in Massachusetts” is obvious. The other thing that makes logic seem special is its “ubiquity,” as he puts it: they are truths that are useful in all subjects. They are the most central to our conceptual scheme because they are used in all subjects. That explains why we feel very reluctant to give up a logical law.

But, of course, there is for Quine no sharp boundary between logical truth—as far as epistemology goes—and any other truth that is important in science but very far from observations. The gradualism or holism in that view makes it clear that these earmarks can bear no special philosophical weight.

Now what of mathematics? Quine is famous for distinguishing first-order logic from set theory. He was the first analytic philosopher to insist that there is a distinction to be made, whereas Carnap (and everyone else, including Frege and Russell) had a capacious sense of logic, which included enough of what we now think of as set theory to found all of mathematics. But Quine distinguishes the two, on grounds of ob- viousness (since of course once you have any amount of set theory you no longer have completeness), as well as on grounds of ontology. That is to say, in set theory you actually have to assume the existence of various items-sets-whereas first-order logic is free of that. But of course, mathematics is also fairly ubiquitous; once again, however, that gives it no special logical status.

If asked what the warrant for mathematics is, where its grounding is, all Quine can come up with is its indispensability to science; it is just part of the overall package. So we have our scientific theory of the world; as a whole, that theory is confirmed by observations; mathematics is part of the science; so it gets whatever confirmation it has from that whole package.

That is really all we have to say about mathematics, though one could ask, What about those parts of mathematics that are not used in science? There are, after all, some parts of mathematics —higher set theory, for example — that are not used. (No one has ever applied wooden cardinals to science.) What Quine says about that is most nicely put in his latest book, From Stimulus to Science:

A word, finally, about the higher reaches of set theory itself and kindred domains, where there is no thought, or hope, of applying in natural science. When I liken mathematical truths to empirical ones, on the score of their helping to imply observations, I was disregarding these mathematical flights. As empiricists, how should we view them? They are couched in the same vocabulary and grammar as applicable mathematics, so we cannot simply dismiss them as gibberish unless by imposing an ab- surdly awkward gerrymandering of our grammar. Tolerating them, then, we are faced with the question of their truth and falsity. Many of these sentences can be dealt with by the laws that hold for applicable mathematics. Cases arise, however-notably the axiom of choice and the continuum hypothesis-that are demonstrably dependent on prior theory. It seems natural at this point to follow the same maxim that natural scientists habitually follow in framing new hypotheses, viz., simplicity, economy of structure, and ontology. That is really all he has to say about higher set theory: just treat it as part of science, even though it isn't really part of science; otherwise it is too awkward.

It is in the philosophy of mathematics that Quine's view has seemed philosophically very vulnerable. It can just seem philosophically mad to think that number theory and analysis have their justification because of their applicability and physics (and experiments confirm physics). There is no registering, if you like, of the inconceiv- ability of another arithmetic. But here again, Quine is going to dismiss those things as data. You can't work up the philosophical point of view from which Quine's position on mathematics looks so weak and unlike our intuitive view of mathematics, unless you use the philosophical vocabulary that Quine, throughout his whole system, is dismissing.]

Let us now review Quine by going through a song. Some autobiographical details here. Quine was a great fan of Gilbert & Sullivan. He was a patron of the Harvard Gilbert and Sullivan Players, and he attended their shows. In 1970, when I was a first-year graduate student and enrolled in the course Word and Object with him, I appeared as the title character in the Harvard Gilbert & Sullivan Players production of The Mikado. So Quine came to this production with his family, and came to the green room afterwards to wish me well; and he was apparently taken by my performance. I didn't realize it, but five years later, when I was looking for my first academic job, Quine put in his letter of recommendation for me that I had appeared in The Mikado. So apparently he was quite taken with my performance!.

Quine retired in 1978. As part of the festivities I decided to write a song of homage to him in Gilbert & Sullivan style. So I penned lyrics to a “patter song” to the tune of “I am the Very Model of a Modern Major General,” from The Pirates of Penzance. The lyrics give a fairly good overview of Quine’s philosophy.

I am the very model of a modern neo-posit’vist The only word I take is that of the atomic physicist Who tells us what ontology, that is, the range of quantifiers Meets up with our demand for truth (and more than truth we can’t require)

That of course reflects Quine’s view that it is the job of physics to tell us what the stuff of the universe is. And, when you are talking about what the stuff of the universe is, what you are talking about is ontology, which is captured by the range of the quantifiers (that is the notion of ontological commitment). “Meets up with our demand for truth (and more than truth we can’t require)”: this reflects Quine’s dismissal of the philosophical modalities and his general extensional approach to everything.

’Tis theory-bound I know full well—it is the state that we are in There is no first philosophy, we can’t be Archimedean Thus my concerns are all within a framework that is naturalized What’s natural here is given us by what our science has surmised

I think that that is self-explanatory. It’s what I’ve been going on about: the idea that we begin in mediis rebus, to use Dreben’s expression; there is no position prior to the results of science from which to criticize or judge science. That leaves us in the position of epistemology as looking just at what goes in and what comes out.

Good views are hence continuous with news from the laboratory It’s only mental myth that makes us think there is another story Upon a raft of Neurath’s sort, in open sea we are adrift And so I am the model of a modern neo-pragmatist

“Good views are hence continuous with news from the laboratory:” i.e., there is no distinction between philosophy and science. We just have gradations of things that are close to observation and things that are further away. “It’s only mental myth that makes us think there is another story:” that, of course, is Quine’s diagnosis of why we think there are meanings. It is just the remnants of this old myth of ideas. And then there is Neurath’s image: we are refitting our boat in the open sea as we go along, without ever getting into dry-dock. “Neo-pragmatist:” reflects Quine’s re- mark, “I espouse a more thoroughgoing pragmatism,” as he put it in “Two Dogmas.” Although when I wrote this I didn’t realize that Quine was making no reference whatever to the views of the classical American pragmatists. He was thinking more of the sort of things that Carnap called pragmatic considerations, and was arguing that they are more pervasive than Carnap thought. I asked him about this in the early 1980s. And apparently, despite those nice words about Dewey in the Dewey Lectures, he really wasn’t a very big fan of the American pragmatists at all.

On language, too, I’ve thought a bit—to most it was a bitter pill There are no grounds I said to talk of truths called “analytical” To back this up I wrote some words on reference inscrutable And on the different options that can make translations suitable This holds at home in what we do, not just in projects radical Homophony and charity are opposites that matter still, Our manual is settled just by some amount of wantonness And minds are shaped by practices like bushes elephantinous Here we have “Two Dogmas”, indeerminacy of translation, and inscrutability of reference. (Remember that ordinarily in assessing the home language we take the homophonic translation; but we can be moved to depart in order not to ascribe weird beliefs to people. That would be the principle of charity.) Finally, we have that wonderful topiary image from chapter 1 of Word and Object, made to rhyme in a way I thought clever. Now we move to chapter 3.

With structure fixed we can research transparence and opacity The latter leads to total loss of reference-capacity But attitudes and modals here I’d like to excise from my list And so I am the model of a modern neo-posit’vist

That is material we’ve just been talking about. You can’t get any clear enough usage of belief talk that would lead to a clear explication.

Now I go to more technical matters, which we haven’t talked about in class.

I’ve put some work into those fields that are more purely logical Though ML fell NF may do, despite its aspect magical

A brief note: ML and NF are the two systems of set theory that Quine has proposed. ML comes from the original edition of the book Mathematical Logic (hence ‘ML’); but Rosser showed that the originally proposed system of ML is inconsistent. (You can develop a version of the Burali-Forti paradox in it.) NF is New Foundations, which comes from the paper “New Foundations for Mathematical Logic” in From a Logical Point of View. It is an extremely odd set theory. It has absolutely no intuitive backing whatsoever, and it leads to all sorts of weird consequences:

Ax Inf implied, full choice denied, and for the rest we’ll wait and see On whether there’s a rel’tive proof that gives us its consistency

The only axioms of NF are those that come from a comprehension schema, and ex- tensionality. But in fact you can get from that that there are infinitely many objects, and that the Axiom of Choice fails. (Rather mysterious.) But it has never been proved consistent relative to any of the standard systems of set theory. (Perhaps this has just changed: I just saw a note on the web claiming to have proved this.)

More recently I’ve spent some time on quantifiers objectual And whether substitution might be just as much effectual In “Existence and Logic” (a paper in the Ontological Relativity collection), Quine toys around with the idea of substitutional quantification, where ‘(∃x)Fx’ does not mean that there is an object x of which ‘F’ is true; it means, rather, that there is some name that you put for ‘x’ that makes the sentence true. He wonders a bit about how much you can do with substitutional quantification. He likes substitutional quantification, because it doesn’t carry ontological consequences; and of course he is interested in ontological parsimony.

These logic matters mean a lot, they’re at the center of our scheme And so a neo-positivist had better look past what they seem

And then just a last little stanza about some of Quine’s other interests:

My interests extend yet more, to things not philosophical I’ve toured throughout the world in places temperate and tropical.

Quine was a great traveler. His autobiography, The Time of my Life, was filled with accounts of where he’d gone; he used to keep a life-list of the countries he’d visited. And he was particularly proud of the countries he’d visited that no longer exist. He told me that there were two: one was the Free City of Danzig, and the other was Newfoundland (which was then a separate crown colony, not part of Canada until 1939). One local wag actually dubbed his autobiography Moving Van.

I know the roots of words and all the op’ras penned by G & S And so I am the model of a modern neo-posit’vist.

Quine had a serious interest in etymology; in the 1960s he reviewed various dic- tionaries for the New York Review of Books.

[What about Quine’s impact? Quine is generally acknowledged to be the most important and influential analytic philosopher of the second half of the twentieth cen- tury. The oddity is, though, that despite his enormous influence, almost nothing he propounded is widely accepted. Indeterminacy and inscrutability are things that people think to be more reductio’s of Quine’s position than philosophical positions it would be desirable to take. Of course, there are also things that simply could not be accepted by anyone who thinks that there is something going on in cognitive science. His skeptical attitude towards the propositional attitudes and the modals is certainly not widely accepted. In fact, if anything, since he started writing it has been going the other way, particularly with the recrudescence of a notion of necessity in philosophy associated with the work of Saul Kripke, which also brings with it this idea that somehow your intuitions about necessity can be used as data in philosophy. But of course that is precisely the sort of thing that Quinian scrutiny is meant to undercut.

So what is the source of his influence? What he did was to unleash what could go on; or as Putnam once put it, “It was Quine who made it OK to talk of ontology again.” Quine went against the Carnapian view—the distinction between the framework and what is within the framework, and the idea that you set out your framework clearly and find that the so-called philosophical problems have to do only with framework issues and hence are not real questions—and in so doing Quine made those things questions again.

The history of analytic philosophy since Quine is the result of that unleashing, or that opening, of the questions again, but without the scrutiny and skepticism and control that Quine’s austere naturalism provides. Quine is heir to one very strong tradition in analytic philosophy, a tradition that starts with the early Wittgenstein. This is a tradition of scrutinizing questions and arguing that what looks for all the world as if it is a real question, is not a real one at all. It doesn’t make sense. One finds that kind of view in Wittgenstein, early and late; one finds it in Carnap; one finds it in the British ordinary language philosophers; and one finds it in Quine. A question can look as though it is real, but it isn’t.

Now of course in all of these different philosophers, what it is to make sense—how it is diagnosed that something makes sense—is different, as we have seen in this course. It is very different in Carnap. He will try to go against traditional questions in a different way from Quine. And those of you who attended Philosophy 141, my course on Frege, Russell, and the early Wittgenstein, will see that it is quite different from the way that the early Wittgenstein went about it; and those of you who take Philosophy 137 will see that it is quite different again from the way that the later Wittgenstein did it.

The oddity is that subsequent analytic philosophy took up Quine’s liberalism about what questions can be asked, but without his skepticism about how those questions are to be clarified (or the vantage point from which he raises the problem of clarifi- cation). Contemporary analytic philosophy has lost this strand in the analytic tradition—what you might call the “suspiciousness” about the nature of philosophical questions. I think that it has lost that suspiciousness completely and entirely. I do not think that it is a healthy development to have lost it completely and entirely; I think that the kind of methodological anarchism that reigns in contemporary analytic philosophy has not produced much of great value. That is the very odd paradox of Quine’s influence.]