27 April 2010

Putnam and Analyticity

As the previous post probably implied, I've been reading more about the analytic/synthetic distinction (and other Quinean themes). Just finished Putnam's essay "The Analytic and The Synthetic". I have mixed feelings about it.

On the one hand, much of what Putnam says against Quine's critics seems to me entirely right and laudable. It's certainly useful for some purposes to have arguments against many philosophical uses of the supposed distinction but which affirm that it exists (since so many people still feel it obviously must have some merit, including the Quine of "The Roots of Reference"). Certainly I'd prefer that everyone agree with Putnam here than with Carnap; if the analytic-synthetic distinction doesn't do any active harm in philosophy then it's rather moot what else we say on the topic.

On the other hand, I'm unconvinced by Putnam's attempt to distinguish "law-cluster concepts" from some other kind (with this other kind being what's susceptible to becoming the subject of an analytic truth). I still think Davidson is right in holding that a concept gets the sense it has by having the inferential connections it has, but that there are no privileged connections here; it's just that if we change too many of them, it's hard to see how we can be working with the same concept we started with. (And it's of a piece with this to not try to make the notion of "same concept" do any heavy theoretical lifting; we can, in general, make that judgement however we like, provided we make appropriate accommodations elsewhere in our story.) So I doubt that there's anything special about the laws connected with a given concept (as opposed to beliefs which make use of that concept more broadly), or that there's a good reason to think any part of language is not like this.

Also, it's worth noting that Putnam is in a sense not defending the analytic/synthetic distinction; he explicitly rejects the notion that analytic-synthetic forms a dichotomy. Putnam thinks there are analytic statements, synthetic statements, statements that are close to analytic, statements that are close to synthetic, and a fifth miscellaneous class. Putnam argues solely in defense of the notion of analyticity: he thinks there are some parts of language which we must not deny as analytic. It was striking how different his defense of analyticity was from Morton White's attack on the analytic/synthetic distinction, reading them back to back like this. (And as a trivial note, Putnam never mentions White in this article, though White was his teacher, nor to anyone else who attacked the distinction; his criticisms are solely directed at Quine.)

I'm not going to try to do justice to everything Putnam has to say in his essay (it's long and dense), but there are a few lines of argument that seem to me to be both central and flawed.

First, a bit I liked, from where Putnam is laying out what he means by way of talking about "law-cluster concepts":

I want to suggest that the term 'energy' is not one of which it is happy to ask, What is its intension? The term 'intension' suggests the idea of a single defining character or a single defining law, and this is not the model on which concepts like energy are to be construed. In the case of a law-cluster term such as 'energy', any one law, even a law that was felt to be defiitional or stipulative in character, can be abandoned, and we feel that the identity of the concept has, in a certain respect, remained. (p.53)
This is very agreeable. Talk of "intensions" carries around the baggage of the "Myth of Meanings": of there being such a thing as The Meaning of a word, and of this as being what a good dictionary entry is supposed to communicate. Dictionaries do not do this, and this is not a flaw; a good dictionary entry gives you some clues as to how a word is used (at least in general, in most cases, by normal speakers), often more by the examples than by the "definitions", and this is often enough sufficient for you to settle any doubts about what so-and-so meant by such-and-such you might've had. Talk of "intensions" or of "meanings" as entities is not a helpful way to understand this.

But, sadly, Putnam does not stop the essay there.
In the case of the terms 'energy' and 'kinetic energy', we want to say, or at any rate I want to say, that the meaning has not changed enough to affect 'what we are talking about'; yet a principle superficially very much like 'All bachelors are unmarried' [the "definition" e=1/2mv^2] has been abandoned. What makes the resemblance only superficial is the fact that if we are asked what the meaning of the term 'bachelor' is, we can only say that 'bachelor' means 'unmarried man', whereas if we are asked for the meaning of the term 'energy', we can do much more than give a definition. We can in fact show the way in which the use of the term 'energy' facilitates an enormous number of scientific explanations, and how it enters into an enormous bundle of laws. (p.53)
I really doubt that it's true that this is the only thing we can say if someone asks what "bachelor" means. Always more than one way to skin a cat, after all. We could, I think, exhibit a great number of sentences in which "bachelor" is used, and trust our hearer to work out the word's significance. And this is plausibly what happens in a great many cases of language-learning; even for terms which Putnam wants to say there are true analytic judgements which take those terms as subjects, it's hardly likely that the use of those terms is *always* taught by explicit statement of an "analytic" definition, or that there's any need for this to be the case. (I imagine this in some more detail in the post on "The Roots of Reference" I linked above.)

And this is connected to my next point: I don't see what's special about the "enormous number of scientific explanations" and "enormous bundle of laws" that "energy" enters into; it just looks to me like a particular case of a word having meaning because it has a use in a form of life (very broadly speaking). Putnam doesn't address this at any point in the essay; I suspect he's privileging laws just because he's using an example from the history of physics to make his anti-Quine's-critics points. Putnam's certainly right that it's this holistic web that gives "energy" the meaning it has, but I see no reason to think the vocabulary of the natural sciences is special in this respect.

I *think* the reason Putnam introduces the notion of a law-cluster concept is just because the notion of a "cluster concept" was already floating around, and Putnam regards this concept as applying solely to "typical general names like 'man' and 'crow'" (p.52). As a philological note, I don't know if this was standard. Putnam attributes the view to Wittgenstein, and the obvious proof-text there is the discussion of "Moses" in PI 79, which is not a general term but a proper name (I'm setting aside for now the nitpick that the "cluster concept" reading of that passage given by Searle is not the best reading). So I'm not sure where the link between "cluster concepts" and general terms is coming from. But as a philosophical matter, I don't see that there's any good reason to distinguish between "cluster concepts" which are "constituted by a bundle of properties" and "law-cluster concepts" which are constituted by "a cluster of laws which, as it were, determine the identity of the concept". Putnam says he agrees with Quine's emphasis on "the monolithic character of our conceptual system", but I think Quine does this monolith more justice by not making the distinctions Putnam here makes. In general, the web of inferential connections a concept is embedded in "as it were, determine the identity of the concept". We can leave out just what those inferential connections are as unimportant; inferential links are inferential links.

Putnam does have more to say about why "bachelor" is not a law-cluster concept:
... 'energy is a law-cluster term, and 'bachelor' is not. This is not to say that there are no laws underlying out use of the term 'bachelor'; there are laws underlying our use of any words whatsoever. But it is to say that there are no exceptionless laws of the form 'All bachelors are...' except 'All bachelors are unmarried', 'All bachelors are male', and consequences thereof. Thus, preserving the interchangeability of 'bachelor' and 'unmarried man' in all extensional contexts can never conflict with our desire to retain some other natural law of the form 'all bachelors are...'. This cannot happen because bachelors are a king of synthetic 'class'. The are not a 'natural kind' in Mill's sense. They are rather grouped together by ignoring all aspects except a single legal one. One is not going to find any laws, except complex statistical laws depending on sociological conditions, about such a class. Thus it cannot 'hurt' if we decide always to preserve the law 'All bachelors are unmarried'. And that it cannot hurt is all the justification we need; the positive advantages are obvious. (p.57)
The first part of this is simply wrong; "all bachelors are nonlobsters" is a counterexample. And Putnam later in the essay gives the definition of "bachelor" as "male adult human being who has never in his life been married" (p.59), which isn't a consequence of "all bachelors are both male and unmarried". And Putnam entertains the (logical) possibility that there are laws like "all and only bachelors suffer psychological trouble Phi" (where Phi is something like "sexual frustration"); he says that if it turns out anything like that is true, then it will have turned out that "bachelor" is a law-cluster concept. His confidence that nothing like this will happen is his ground for saying that "bachelor" is not a "natural kind" and that it cannot hurt to decide to always preserve the law "All bachelors are unmarried men"; if it turns out that "bachelor" is a law-cluster concept then Putnam would reject it as the possible subject of an analytic truth. All of this strikes me as fishy enough to be suspicious that Putnam's "analyticity" is something we'd be better of without. (We can still say that "All bachelors are unmarried men" is something to not give up simply because it's true, and stop there.) But, Putnam says that there are obvious benefits, and he appeals to these as one of the chief motivations for retaining the notion of analytic truths.

So I now turn to the supposedly obvious benefits:
Most important, there is the advantage of brevity. Also, there is the question of intelligibility. If some of the statements in a language are immune from revision and if some of the rules of a language are immune from revision, then linguistic usage with respect to the language as a whole is to a certain extent frozen. Now, whatever disadvantages this freezing may have, there is one respect in which a frozen language is very attractive. Different speakers of the same language can to a large extent understand each other better because they can predict in advance at least some of the uses of the other speaker. (p.56)
Putnam says nothing else on the topic of brevity. I doubt there's any real gain in saying "bachelor" rather than "single man"; both are trisyllabic. Certainly there are advantages to having multiple words with similar meanings (for poetry and to avoid monotony), but we don't need "strict synonymy" for that.

The gain in intelligibility, then, is what I take to be the real supposed benefit. But I don't see that this works, either. For one thing, if we reject Putnam's notion of analyticity we can still say that everyone believes that all bachelors are unmarried men, so we can still predict in advance that any particular speaker will believe this. There's no need to have a "frozen language" when there're frozen beliefs. This I take to be a fully adequate rejoinder; Putnam does not show any good reason to not drop "analytic" from our vocabulary, as Quine would have us do.

But I think there is a real risk that Putnam doesn't address. I think it's not entirely crazy that someone might chaff at "All bachelors are unmarried men"; certainly "married" and "man" are both terms certain people have problems with. (I vaguely recall reading an interview with Judith Butler where she complains about the supposed "necessity" of just this categorical statement, in the mouth of Kripke.) Perhaps there are some intersex persons who feel comfortable self-identifying as "bachelors" but not as "male". Or perhaps some unconventional partnership arrangements lead to men who regard themselves as equally husband and bachelor (perhaps it's an open marriage and they keep separate apartments). If Putnam is right, anything like this involves ceasing to speak English; a "nonmale bachelor" and a "married bachelor" would simply involve equivocation on the term "bachelor". But this seems to me unfair, at least on the a priori grounds Putnam supplies; presumably the intersexed person and the unconventional husband think of themselves as "bachelors" not because of some crazy new meaning they've attached to the word, but because they see themselves as being what is called in English a "bachelor". To say whether their projections of the term are reasonable seems to me impossible to decide without seeing how life works out if we do project with them or we don't. It's not something philosophers have any privileged view on.

So, my conclusion is the inverse of Putnam's: I don't see any gain to retaining the analytic/synthetic distinction, and I see some real possible risks. So I'm happy to go without it.

17 comments:

Evan said...

Thanks for these posts on analyticity, and for turning to Putnam here. I actually found Putnam's essay quite helpful this past fall when I was trying to gain a better understanding of the analytic/synthetic issue for a course paper, so it's interesting to read your critique here. At the time I had really no background in any of this literature (I'm still only just introducing myself to it), and I felt that Putnam was one of the most helpful people for beginning to think alongside Quine. After reading your post, though, I looked back at my paper and it seems I used Putnam mostly in relation to some points about holism, so apparently I didn't see the need to cite him over against Quine on analyticity itself in any substantial way. I'll have to reread his essay when I have the change (although I don't recall it being all that short, despite the fact that "brevity" was supposedly an advantage of his formulation).

Sorry, that's not really much contribution to a discussion, but I'm glad that you're visiting all of this here.

Daniel Lindquist said...

I am glad you like the posts. Don't worry about not contributing to discussion; I'm largely writing these things for myself now. N.N. (from the now-defunct "Methods of Projection") used to be a reliable conversation-partner on these topics (I have at least a half-dozen Quine posts in my archives, all about the same topics), but he seems to have vanished from the internet.

I did like a good bit of the Putnam paper. I think his kinetic energy example is quite a good one; Quine uses the same example later on, and I suspect he took it from Putnam. I just don't think that, in the areas where Putnam opposes himself to Quine, he has good reasons for doing so. The places where he agrees with Quine generally have him helpfully adding to Quine, and not just repeating him. Certainly valuable.

The essay is definitely not short; reading it and taking notes was pretty much all I got done for a few hours yesterday.

N. N. said...

I havn't vanished entirely.

Perhaps there are some intersex persons who feel comfortable self-identifying as "bachelors" but not as "male". Or perhaps some unconventional partnership arrangements lead to men who regard themselves as equally husband and bachelor (perhaps it's an open marriage and they keep separate apartments). If Putnam is right, anything like this involves ceasing to speak English; a "nonmale bachelor" and a "married bachelor" would simply involve equivocation on the term "bachelor". But this seems to me unfair, at least on the a priori grounds Putnam supplies; presumably the intersexed person and the unconventional husband think of themselves as "bachelors" not because of some crazy new meaning they've attached to the word, but because they see themselves as being what is called in English a "bachelor". To say whether their projections of the term are reasonable seems to me impossible to decide without seeing how life works out if we do project with them or we don't.

If A understands a bachelor to be an unmarried man, then he may not understand the statement by an intersex person, B, 'I am a bachelor.' If in response to a query B explains that being an unmarried person is the defining characteristic of a bachelor, then A will likely understand B's statement. Still, A may reply that this is not what he means by 'bachelor.' With respect to the word 'bachelor,' then, A and B speak different languages. This need not stand in the way of communication. And the question, 'Who is really speaking English' seems to me to be completely uninteresting.

What does interest me is the idea that 'All bachelors are unmarried men' is a truth that is revisable in the face of experience. This, I contend, is mistaken (and my views have developed a bit since our initial discussions on this topic). It is not true that bachelors are unmarried men. Rather, it is true that some people mean 'unmarried man' by 'bachelor.' The discovery that other people mean something different is not like discovering that bachelors really do include non-males. It may be discovered that some people use the word 'horse' for llamas. They may even explain their use by pointing out similarities between what we call 'horses' and 'llamas.' Still, this discovery is not like discovering that llamas really are horses. Experience has not shown us that we were wrong in our use of 'horse.' There's no such thing as wrong here, if wrong is supposed to have something to do with the way the world is. All 'wrong' could mean here is, different than a (for some reason) privileged group of speakers. Thus, the idea that all truths, even the analytic ones, are revisable in the face of experience of the way the world is, is a rejoinder to a confusion.

By the way, congratulations on being admitted to Indiana's (that's right, isn't it) Ph.D. program.

Daniel Lindquist said...

Well speak of the devil.

"If in response to a query B explains that being an unmarried person is the defining characteristic of a bachelor, then A will likely understand B's statement."

Right, for my hypothetical to work for me the intersex person would have to not do this. They would have to simply insist that by "bachelor" they meant what anyone means by it, though they regard it as compossible with refusing the label "male". Another way of putting this is that I'm skeptical that there are any "one-criterion words" in Putnam's sense.

Incidentally, in "Word and Object" Quine agrees with Putnam that there are "one-criterion words", that "bachelor" is one, and that this is behind the intuitive labeling of "all bachelors are unmarried men" as analytic. I thus find myself once again defending a stronger position than Quine held.

I'm confused by your example -- are they still using "horse" for horses, or just for llamas?

And yes, Indiana. And thanks.

N. N. said...

They would have to simply insist that by "bachelor" they meant what anyone means by it, though they regard it as compossible with refusing the label "male."

If someone, A, denies that the intersex person, B, means by 'bachelor' what he (A) means — 'But bachelors are male' — will B (on your example) claim that A has not paid sufficient attention to how he uses 'bachelor'? That is, will B say to A: 'Your (literal) use of bachelor doesn't confine its application to males'?

While your example is interesting, I can make my point whether the intersex person is right or not. Whatever the explanation of 'bachelor,' the sentence 'All bachelors are [insert explanation of bachelor here]' is not true (or false). Essentially, it is a stipulative definition, and therefore, it cannot jive or conflict with experience.

With my 'horse' example, I meant that unusual use of 'horse' included horses and llamas.

Daniel Nagase said...

One thing that I never quite understood is how "analytic" changed from meaning "justified by the law of non-contradiction" to meaning "not capable of being revised" (or something to the effect). When Kant introduced the pair "analytic-synthetic", his examples of analytic judgments were "Gold is a yellow metal" and "Body is something impenetrable". That both of these judgments have been proved wrong would probably leave Kant unmoved, as they would still be justified by an appeal to what we mean by those concepts (be they empirical or not). This is entirely in accord with Kant's views about "concepts", which he treated as rules for the recognition of objects: if I encounter an object that is recalcitrant to the rule, either the object is not what I thought it was (e.g. the wall over there could be an illusion) or I should revise my rule (e.g. it turns out bodies are not impenetrable, after all).

It makes me wonder, then, how the distinction came to mean what it meant to Quine when he subjected it to such a violent criticism.

(Incidentally, congratulations on maintaning such a great blog!)

N. N. said...

I don't mind saying so-called analytic truths are able to be revised so long as we understand what revision amounts to here.

Grice and Strawson get it just right on this score:

Now for the doctrine that there is no statement which is in principle immune from revision, no statement which might not be given up in the face of experience. Acceptance of this doctrine is quite consistent with adherence to the distinction between analytic and synthetic statements. Only, the adherent of this distinction must also insist on another; on the distinction between that kind of giving up which consists in merely admitting falsity, and that kind of giving up which involves changing or dropping a concept or set of concepts. Any form of words at one time held to express something true may, no doubt, at another time, come to be held to express something false. But it is not only philosophers who would distinguish between the case where this happens as the result of a change of opinion solely as to matters of fact, and the case where this happens at least partly as a result of a shift in the sense of the words. Where such a shift in the sense of the words is a necessary condition of the change in truth-value, then the adherent of the distinction will say that the form of words in question changes from expressing an analytic statement to expressing a synthetic statement. ("In Defense of a Dogma," 156-7)

Daniel N writes: if I encounter an object that is recalcitrant to the rule, either the object is not what I thought it was [...] or I should revise my rule.

The second option (if I understand Daniel) is what Grice and Strawson are discussing. To revise the rule is to decide to change the meaning of the relevant term(s). We can do this whenever we want and for a variety of reasons. But what we are doing is not judging our previous definition (e.g., 'Bachelors are unmarried men') false. We are changing the meaning of 'Bachelor.' Consequently, 'Bachelors are unmarried men' and 'Bachelors are unmarried persons' (our revised definition) are not in conflict because the word 'Bachelor' in these definitions is equivocal. That is, the two are incommensurate.

Daniel Lindquist said...

Sorry it's taken me a while to respond to recent comments; I suddenly found out that housing in Bloomington starts to get scarce after May 1, so I had to run up and start looking for places very suddenly. And then after I'd found a place (which took several very long days), I had a floorplan to toy around with, so much of this week has been spent furniture shopping. Ikea has left me too tired for blogging of any substance. But at least I have a nightstand of adequate size now!1

N.N.:"If someone, A, denies that the intersex person, B, means by 'bachelor' what he (A) means — 'But bachelors are male' — will B (on your example) claim that A has not paid sufficient attention to how he uses 'bachelor'? That is, will B say to A: 'Your (literal) use of bachelor doesn't confine its application to males'?"

If B did say this, there are a few things they might mean:
(1) That what A means by "bachelor" need not limit its application to males, and it might only be A's additional belief that only males are bachelors that keeps him from speaking as B speaks. So A doesn't change the topic if he speaks as B does; he simply comes to a better understanding of what his use of that vocabulary involves.

(2) That nothing in the world is stopping A from calling nonmales "bachelors" (he can speak as he likes). (This is clearly compatible with Grice/Strawson's position.)

(3) A has been using the word "bachelor" wrong, dogmatically assuming that it can only be correctly applied to males.

I think (1) is the most profitable tack for B to take, here, but I doubt that B would respond like you have them responding at all. B would probably (and more profitably, by my lights) simply deny that "all bachelors are male" is true, either by meaning or by any other way. B would say that A was simply being dogmatic (perhaps under the veil of calling a dogma "analytic"), and that A was wrong to be dogmatic about an issue of such practical import. B would probably then start into a long harangue about what "male" is even supposed to mean, and what a misguided and bigoted web of patriarchal/heteronormative baggage (in theory and in practice) that vocabulary had traditionally been tied in with, (and the same for sexed terms for married/unmarried persons and for the very idea of marriage) etc. I doubt it would even occur to B to think that there's something tractable called "A's use of a word" here that could profitably be looked at in isolation. B wants to change how A gets around with gendered/sexed ideas in a broad sense; I doubt B cares which changes they want to see in A's life are changes of meaning and which of belief. So I don't see that your hypothetical response makes much sense.

"Whatever the explanation of 'bachelor,' the sentence 'All bachelors are [insert explanation of bachelor here]' is not true (or false). Essentially, it is a stipulative definition, and therefore, it cannot jive or conflict with experience."

If it's neither true nor false, then it can't have inferential significance. But "All bachelors are phi; B is a bachelor; therefore phi(B)" looks like it should be a valid inference. So it looks like whatever way you have of filling out phi should give you something which is at least a candidate for truth. So stipulative definitions have to lay out propositions we're supposed to hold true. If they don't, I don't know what role you imagine stipulative definitions playing.

Daniel Lindquist said...

Other Daniel: I am glad you like the blog.

"Justified by the principle of noncontradiction" entails "Not able to be revised", traditionally, since the consensus was that if you dropped the principle of noncontradiction then reasoning just gets shown the door. And "justified by the principle of noncontradiction" doesn't really get you to a claim like "Gold is a yellow metal" by itself; you have to also have the concept "Gold" to unpack in accord with the principal. Analytic judgements for Kant are distinguished from synthetic truths by the fact that you can determine their truth-value with just concepts; synthetic judgements also need intuitions. And since experience is just being provided with a course of intuitions, only synthetic judgements can have their truth-values revised on the basis of experience. So I think that what Frege or Carnap meant by "analytic" was pretty close to what Kant meant.

I think Kant would actually be quite surprised to find that (many) people nowadays would deny that gold is a yellow metal or that body is impenetrable; the latter in particular plays a real role in Kant's phoronomy in the "Metaphysical Foundations of Natural Science", as I recall, and Kant was writing before Lavoisier's revolution in chemistry (or at least before it had spread) -- Kant infamously claimed that chemistry could at best be an "art" and never science. I think that "Gold is a yellow metal" and "Body is impenetrable" are things that Kant found obvious which are now widely judged false. I don't think it's by any means easy to simply say that we've changed our vocabulary here; we still use "gold" and "body" in ways incredibly similar to how Kant would've used their German equivalents, after all, excepting in weird theoretical cases. Kripke has an interesting bit in "Naming and Necessity" where he's just astonished at the idea that "Gold is a yellow metal" might be considered analytic, incidentally.

N.N. on Other Daniel: I think you move too quickly in associating his take to Grice/Strawson's. "I should revise my rule (e.g. it turns out bodies are not impenetrable, after all)" is not the G/S line; they wouldn't say the part in parentheses.

N. N. said...

On your view, are there any significant statements that are neither true nor false?

Think of a new, and completely arbitrary stipulative definition, e.g., "This collection of things is called a 'blorp.'" Under what circumstances would it be (could it be) false?

N. N. said...

If it's neither true nor false, then it can't have inferential significance.

Also, what's the reason for holding this?

Daniel Lindquist said...

I waffle on whether or not there's any utility to having truth-value gaps, and if so then what's the best way to think about them. But the only case that really bothers me there, I think, is "names that don't name", out of some uneasiness regarding McDowell's "De Re Senses".

"Think of a new, and completely arbitrary stipulative definition, e.g., "This collection of things is called a 'blorp.'" Under what circumstances would it be (could it be) false?"

I'm not sure if this is supposed to be a nonsensical stipulation (like asking how many objects can fit in a breadbox), or if "this collection" is just supposed to be made clear by context. I'll assume the latter.

I think you can plausibly treat that sort of claim as false if no one does call that sort of collection a "blorp". That is, I think we can be revisionist about what counts as "stipulative definition", if that phrase is supposed to guarantee the truth of whatever we stamp with it. And I think it does sometimes happen that an author will introduce a "stipulative definition" for an everyday term, and then fail to write in accord with it, so I don't think this is a purely hypothetical line to take. (I recall Hacker having a nice point about this back in that exchange with Dennett and Searle.)

I suppose if it's hopelessly vague what "this collection" was supposed to be, it might be profitable to treat the claim as false there, too. If there's nothing to call a "blorp" then it looks untrue, hence false, I suppose.

"Also, what's the reason for holding this?"

Inferences are supposed to preserve truth; if you have claims which are gaps then those are useless as premises.

Daniel Nagase said...

Perhaps I was a bit sloppy with my original wording. I did not mean to claim that Kant did not adhere to the truth of "Gold is a yellow metal" or of "Bodies are impenetrable". What I did mean to claim is that his theory of analytic judgments is compatible with the aforementioned judgments being (i) analytic and (ii) false, from our point of view. That is because, as I understand him, Kant considered that every empirical concept was always open to revision. Given that, it's not surprising that the corresponding judgment which expanded its marks would also be open to revision. As Kant himself puts it (the quotation is lengthy, but worth of quoting in full):

"As the expression itself reveals, to define properly means just to exhibit originally the exhaustive concept of a thing within its boundaries. Given such a requirement, an empirical concept cannot be defined at all but only explicated. For since we have in it only some marks of a certain kind of object of the senses, it is never certain whether by means of the word that designates the same object one does not sometimes think more of the these marks but another time fewer of them. Thus in the concept of gold one person might think, besides its weight, color, and ductility, its property of not rusting, while another might know nothing about this. One makes use of certain marks only as long as they are sufficient for making distinctions; new observations, however, take some away and add some, and therefore the concept never remains within secure boundaries." (A727/B755-A728/B756, original emphasis)

So, a judgment can be justified (solely) by the principle of non-contradiction, because it merely explains what was already contained in a given concept (e.g. "gold"), but still be open to revision, as the concept itself, being empirical, is open to revision.

I think part of the confusion here is that people are not paying enough attention to two separate issues, namely, that of the acquisition of empirical concepts and that of justification of judgments that explicate those concepts once they have already being acquired. In Kant's eyes, experience is certainly needed for the former but not for the latter, as his example of analytic judgment ("Gold is a yellow metal") readily shows.

I'm not sure, by the way, that Kant would consider that, once the revision is made, the two concepts -- that is, the old one and the revised one -- would be incommensurate. For, while they are in a way equivocal, they are not wholly equivocal. There is a certain continuity between the two (in N. N.'s example, both concepts refer to the unmarried status of persons, be they gender specified or not) which can, perhaps, serve as a basis for comparison. I'm not sure what to make of this, however.

Anyway, hopefully I have been clearer this time around.

Daniel Lindquist said...

Other Daniel: That is a much more interesting point! You've motivated me to reread Kant (though I was inclined to anyway); I've gotten rusty on a lot of the details.

I can see how you think what you think, given Kant's discussion in the first chapter of the Doctrine of Method. Certainly Kant's views on definition complicate the picture in a way that makes treating Kant the same as Carnap & Frege problematic.

But I'm still not sure that that view is wrong (though it might be, or at least be less than the whole truth). Here are some thoughts.

In the passage you quote, Kant mentions marks like "not rusting" as being added & dropped from the concept of gold as experience reveals more about gold to us. It's not clear to me that Kant thinks that *all* marks of an empirical concept are like this. The fact that he uses a fairly narrow roster of examples of analytic judgements suggests to me that the mark "yellow" is supposed to be different in kind from the mark "does not rust" in the concept of gold. Something like essential and accidental marks of a concept. This would then make the association of Kant with the later logicians fairly simple: analytic truths unpack the essential marks of a concept. Your response to N.N. seems to trade on some such distinction (both "bachelors" are concepts of unmarried persons, and so there is continuity), but I maybe you could probably get by with a merely historical continuity (empirical concept A has marks abc, empirical concept B has marks def, but they're not incommensurable because concept B was the result of adding d, e, and f and dropping a, b, and c from A over time -- it just turns out that your initial concept was really, really off-base, I guess).

Kant never seems to relativize the notion of analyticity to the marks we in fact do think in a concept at the moment, so it seems unwarranted to think that "gold does not rust" might be analytic at any point, or for any people, for Kant, even if that mark is in fact one of the ones I happen to think in the concept. On your reading, it would seem that judgements ought to shift from synthetic to analytic (and perhaps back again) as inquiry proceeds, as we add and drop marks from a concept. That strikes me as an interesting (and appealing) idea (and clearly very different from anything like "truth by convention"), but I don't see it in Kant.

Daniel Lindquist said...

Kant's right to say that empirical concepts have no definitions, because a definition is supposed to be "complete" (exhaustive), and Kant thinks that this is not possible outside of mathematical concepts. He does give an exception for this near the passage you quoted, but it's not followed up on much: A729/B757 mentions "concepts thought by choice. In such a case I can indeed always define my concept; for I must surely know what I wanted to think -- since I myself deliberately made the concept and it was not given to me through the nature of my understanding, nor through experience." It seems plausible to me that "bachelor" could've been given by Kant as an example of such a concept (the closest he seems to give to an actual example here is "ship's clock", which he mentions only to note that it's not a proper example of this sort of concept, as it relies on the existence of an object (ships); this is less than helpful).

As a response to your historical question: The poor reception Kant's philosophy of mathematics has had (going back at least to Bolzano, and probably earlier) has probably hurt the reception of this part of the critical philosophy. The idea that mathematics relies on "construction in intuition" has rarely been greeted with open arms, and since this is tied to Kant's treatment of mathematical definitions, his treatment of definitions in general has probably suffered some neglect by association.

The confusion you mention in your penultimate paragraph also seems plausibly responsible; without that distinction in mind, claims like "to base an analytic judgement upon experience would be absurd" seem to demand that "gold is a yellow metal" must be something we can recognize as true before all experience (as opposed to just before any additional experience, once we get the empirical concept of gold).

Again: very interesting topic. You've given me stuff to think about. Thanks for that.

N. N. said...

Inferences are supposed to preserve truth.

Can't we do this while maintaining that stipulative definitions are neither true nor false? Think of stipulative definitions as Wittgenstein does, as substitution rules. Take the true empirical statement, 'Daniel is a bachelor.' Using the definition/substitution rule 'bachelors are unmarried men,' we can substitute 'Daniel is an unmarried man' for 'Daniel is a bachelor.' Given the rule, the truth of the first statement is preserved in the second.

(2) the mark "yellow" is supposed to be different in kind from the mark "does not rust" in the concept of gold. Something like essential and accidental marks of a concept.

This reminds me of Wittgenstein's distinction between criteria and symptoms:

In practice, if you were asked which phenomenon is the defining criterion and which is a symptom, you would in most cases be unable to answer this question except by making an arbitrary decision ad hoc. It may be practical to define a word by taking one phenomenon as the defining criterion, but we shall easily be persuaded to define the word by means of what, according to our first use, was a symptom. Doctors will use names of diseases without ever deciding which phenomena are to be taken as criteria and which as symptoms; and this need not be a deplorable lack of clarity. For remember that in general we don't use language according to strict rules . . . (Blue Book, 25)

(3) Your response to N.N. seems to trade on some such distinction ([1]both "bachelors" are concepts of unmarried persons, and so there is continuity), but I maybe you could probably get by with a merely historical continuity ([2]empirical concept A has marks abc, empirical concept B has marks def, but they're not incommensurable because concept B was the result of adding d, e, and f and dropping a, b, and c from A over time . . .)

(1) appears to be an instance of having your cake and eating it too. On the one hand, there is no essential change in concept. Consequently, the 'two' definitions are commensurable. On the other hand, there is accidental change in concept. Therefore, one and the same concept has undergone change.

Compare (2) to changing the game of chess so that the 'pawns' now move exactly like 'queens.' Let's also say that no one plays the old game anymore. Is the statement about the old game 'Pawns cannot move more than one space (except on their first move when they can move two steps if there's no other piece in their way)' now false? Isn't the right response to whether it's true or false, 'That depend on what you mean by "pawn".' And echoing Wittgenstein, can't we say that this is not really an answer but a rejection of the question?

J said...

It takes some time, but eventually the bright-boy (at least if a non-zombie) realizes that Quine was...wrong in TDOE. Q.'s point's re analyticity (and the end thereof, supposedly) was merely ...theoretical, a somewhat desperate, quasi-Darwinian attempt to undercut the last remaining rationalist aspects of Carnap (or anal.phil as a whole, at least the Fregean sort). Eventually, in a few hundred years, "bachelor" might not be synonymous with a "unmarried male" or "lawyer" might not equal "attorney" (but I wager they will). But for all practical purposes, it's analytical.

And what about say square root of 25 and 5, or ...pythagorean theorem, or the definition of modus ponens, or fundamental theorem of calculus, etc? Semantic synonymies don't seem quite as "fixed" as mathematical identities--ergo, there's a need for pragmatic analyticity, even if the hints of metaphysical realism bother the Uncle meat materialists. QE -f-ing D.