A lot of people have felt that Kant's analytic/synthetic divide has been "dead" since Quine and his "Two Dogmas of Empiricism." While that's true for many of us, even those who accept Quine often still end up making distinctions that are almost the same as Kant. Of course the traditional example of an analytic truth is "all bachelors are un-married." I've been thinking of that today and trying to figure out how to use that example in a fashion that breaks the divide. It's hard, but here's my guess. You can tell me if I'm off my rocker.
Consider the situation where a bachelor is partying it up in Vegas. He has an alcohol related blackout. During this period he and some stranger in Vegas go to a wedding chapel and get married. Neither remember it and the paperwork is lost. Is this man still a bachelor?
Well, if I am born a Jew but adopted by Jehovah's witnesses who go out of their way to hide all details of my origin, am I still a Jew? I think literally minded Orthodox Jews would say yes I am, by the definition of "Jew".
I'm not sure that's as good an example simply because the meaning of Jew to an orthodox rabbi would be purely about who your birth mother was. (I'm not sure how they handle adopted children; but I suspect a rite of adoption)
The interesting bit about the bachelor example is that the meaning of bachelor doesn't appear to be so pure. Bachelor doesn't appear to be quite as pure a concept. That is the concept can't be captured by the definition and thus it isn't really analytic.
Of course given the Quinean charge presumably we ought be able to show that even ones Jewishness isn't purely analytic. So I guess I have to be able to handle that one as well.
So consider a surrogate mother who is Jewish but the genetic mother is not. (Or vice versa) Of course I'm sure Orthodox Rabbis have ruled on such cases. But up to the point of the ruling (which extends the definition) it would seem to be synthetic and not analytic. One might also say that if definitions thought to be analytic end up "expanded" by empirical situations are they still analytic?
That is doesn't the mere fact a Rabbi has to rule and set a convention entail that this isn't analytic at all?
And of course even if we come down on the side that says the child is analytically Jewish we can merely adjust the thought experiment. Say we take a gentile egg and start removing fragments of DNA and add in Jewish DNA. At what point does the egg become Jewish? (Yeah, a science fiction thought experiment, but one probably scientifically relevant within the next 100 years)
Who knows the facts of your situation? By definition, your very hypothetical doesn't permit us not to know it, so the guy who got married is not a bachelor, if only because we know it.
But doesn't the example rest on the fact that his knowledge is more significant than our knowledge? That is it is ambiguous whether the marriage is really a marriage.
Now one could attack this example and argue the problem isn't with the definition but whether whether they are married. But clearly they are married. But I think it clear that in terms of meaning folks could still call him a bachelor and not be wrong.
The example we always use in cognitive psychology as an illustration of the problem with examples is a Catholic priest. Catholic priests are unmarried males, but you won't find many people willing to call them bachelors.
Clark,
I think your example goes not so much to show that the sentence "all bachelors are un-married" is not analytical, but that the concept of bachelor is not simply "unmarried male".
The similar examples are "Is Pope a bachelor?", "Is a muslim with 2 wifes bachelor?", or "Is a baby boy bachelor?".I think that what such examples show is that the concepts (or I would rather say phenomena) are sometimes complex and to be aware of them one needs to be aware of the wider context - i.e. phenomena of bachelor is connected to the phenomenon of there being society in which male children on certain age are expected to be married (and as pointed in the example, it doesn't cover lot of complications).
BTW, I'm not sure that Quine's attack is against Kantian distinction, but towards the rephrased distinction of analytical (logical empiricist) tradition before Quine. (At least that's why I heard somewhere, but I don't remember where I read that, and also I don't quite remember the argument in "Two Dogmas Of Empiricism", and can't give any arguments)
Also if we will accept that "all bachelors are unmarried" is analytical, I guess it all depends on, as Chris says, on what we are willing to call bachelor.
If we are willing to call a Muslim with two wives a bachelor, then the sentence is not true.
But Tanasije, isn't the point of the sentence "all bachelors are unmarried" as analytic possible only because of the meaning of "bachelor"?
Your point about Quine is interesting. I'd not heard that before. Now I may be completely wrong in all this (thus the post - something for me to learn). However I thought Quine's distinction was between truths of meaning and truths of facts and showing the distinction didn't hold up. Now I'll be the first to admit my Kant isn't as good as it should be. But I don't see the distinction between Kant's "predicate concept contained in subject concept" and the positivists' "true in terms of meanings and not facts." Could anyone clarify this for me?
Chris, I should have thought of that one. Although I suppose it might fall prey more for the meaning vs. fact distinction. That is it is about meanings more than facts whereas the two examples I gave (the forgetful unknownly married bachelor and the genetically engineered "jew" appear more about facts)
Clark,
Right, as I understand it, to be analytic "all bachelors are unmarried" should be true in virtue of the meaning of "bachelor", and meaning of "married".
But it seems to me that for "all bachelors are unmarried" to be analytic, all that is needed is that meaning of "bachelor" includes "unmarried" as a necessary feature.
So to say, "unmarried male" might not be sufficient for determining somebody as a bachelor (as in your example, or the example of Pope, or young boy), but as long those are necessary features, "all bachelors are unmarried" would be analytic.
On other side I think the example with the Muslim with 2 wifes, as long one accepts him as a bachelor, is where the "being unmarried" is negated as necessary feature, and hence the analiticity of "all bachelors are unmarried" is negated.
As for Quine/Kant relation, I was thinking more but I can't figure out where I read what I said. Wish I did. (BTW, I don't mean that Kant's distinction is good. Also I wrote a post inspired by this your post, you might want to take a look.)
But doesn't the issue of facts interfering with the truth of "all bachelors are married" as true by meaning entail the failure of the analytic/synthetic distinction?
Note that Tanasije post is here
True,
As far as we can point to a possibility that there can be bachelors which aren't be married, we can show by example that "all bachelors are married" is not analytical, and that its truth depends on the state of the world (of course, as far as the bachelor is problematic, it also opens the issue of how to do the tests... should we test just those bachelors which are male and unmaried? if we do, we will get to the answer that the proposition as analytic proposed).
In my opinion, what is more clear here is that the analytic/synthetic distinction can't be grounded in the way that Kant was proposing (mainly because I think it is based on the classical view of concepts, which seems problematic to me), and also probably doesn't hold in all different ways that Quine tried to reformulate it. However, I don't think that this shows that analytic/synthetic distinction in general is shown false. I would still think that math, logic and "good" metaphysical propositions can be true in virtue of their meaning.
BTW, as for Kant / logical empiricist distinction, I guess one obvious one is that according to Kant's distinction math was synthetic, and according to empiricists' distinction math was analytic.
O.K. I'm going to barge in here on Kant's side of the distinction. I'm going to bracket the fact that Kant spoke in terms of judgments, and thus implicitly focused on the moment of applications/acts, rather than logical propositions/inferences, (though I think that was his key basic move in response to Hume). What's of further interest is that Kant, in taking over Leibniz' concept containment account of logical meaning and giving it a twist to address empirical issues or the issues raised by empiricism, thought that it was a significant possibility that one could deploy a concept and not realize fully all that was or might be entailed by that concept. For example, one could define "human being" as "rational animal" without realizing it entailed "finite", in the sense of dependent on sense reception for the content of experience, which is precisely what he accused his rationalist predecessors of missing. (I'll bracket whether "rational animal" is synthetic or analytic). In other words, "functionally", an analytic judgment is an explicative remark, whereas a synthetic judgment is amplicative, while both,- or perhaps the according of "analytic" a priori status to certain synthetic judgments,- are required for the conceptual synthesis of experience/knowledge. The oddity here is less the knowledge-constitutive role of synthetic judgments a priori than the possible ignorance of analytic entailments, which raises a question about the practical dimension of concept holding/acquisition. (As an aside, I think that Kant's view of mathematics as "synthetic" has to do with the projective-constructive role of the "forms of intuition" of space and time in his scheme).
An analytic judgment would be the equivalent of a "grammatical remark", as a philosophical reminder, in later Wittgenstein. "Sensations are private" is o.k. as a grammatical remark, just in case one somehow failed to recognize that you don't experience the pain of my toothache or vice versa,- (though I suspect for W. it's more the ethical solecism than the cognitive default that's at issue in such an oddity),- but, if treated as a referential statement, such that some mysterious property of "privacy" attaches to ordinary sensations- ( rather than, perhaps, "privacy" adhering to the bearers of sensations)-, metaphysical trouble is invited. Wittgenstein developed his conception of grammatical remarks in his notebooks from the late 1930's partly in terms of a distinction between "symptom" and "criterion", with the possibility that the two could change places and then back again. AIDS, say, could be identified by the symptom of clusters of unusual opportunistic diseases, until it's recognized that the syndrome is one of immune system collapse, so that the symptom becomes criterial, except that the recognition occurs that its manifestation is different in well-to-do Westerners and malnourished Africans, until antibodies to the HIV retrovirus become the test for its causative factor, in which case non-symptomatic cases can be recognized, and so forth. The upshot is that the analytic/synthetic distinction is neither intrinsic, nor absolute, but varies both diachronically/historically and synchronically/topically in terms of which conceptual features of "our" experience are held constant/"certain" and which are treated as variable in our dealings with the world and each other.
Quine inherits the analytic/synthetic distinction in a reduced form from logical positivism, where the question is which features of a theory belong to its formal framework and which to its empirical content, "protocol sentences". There is some loss there of engagement with the world and practical need,- (as opposed to its mastery by theory). And Quine is wary of the notion of "necessity", which, if taken in a generalized and more-or-less metaphysical sense, he's right about. But relativized to specific conceptual frameworks or complexes, it's doubtful that the analytic/synthetic distinction can be quite dispensed with, and that the "correct" criticisms should be directed at how conceptual syntheses/theories can be adjusted to the world, rather than at how they open up (or close off) the world and its aspects within broader forms-of-life. It's not just a matter of the conventionality of "our" meanings, but of how "deep" those conventions go and of their cross-implications and of their riddling contingencies and the constraints on their modifications/transformations. And, of course, of the needs in terms of which they are "justified" and of our need for "justification".
At any rate, maybe the only thing I want to say definitively or offer up here is that the original Kantian notion of analytic judgments was more pointed and more shrewd than has been often subsequently taken to have been the case.
Clark,
It's a synthetic proposition. Your hypothetical Vegas drunk, let's call him Fred, is still a bachelor based on the definition that a married person has a certificate of marriage on record with a county registrar. It's synthetic because in the proposition "Fred is a bachelor," "is a bachelor" amplifies the concept of "Fred." Further it is a posteriori since verifying it would required empirical investigation.
The phrase "All bachelors are unmarried men" would be analytical and a priori because the predicate is contained in the subject and is defined without appeal to experience. It doesn't matter that to apply it to specific cases one has to appeal to empirical facts, the phrase itself need not refer to anything empirical.
Of course you knew all that, and likely have some reason why the previous analysis is wrong.
The issue David is whether that phrase's meaning really is analytic though. The example of the drunk bachelor is just to raise questions about the meaning of the phrase. i.e. to provide intuitions that the analytic/synthetic divide in this case is off.
Of course we can always just affirm analyticalness and deny Quine's argument.
Part of the problem is that you need to be more clear as to what you are referring to. You have conflated two things. First, there is the definition of marriage, which you did not address. Second, there is the proposition of Fred's marital status.
For my post I used the proposition "Fred is a bachelor" as the proposition, using the having a marital certificate at the county registrar as the definition of marriage. Done that way I think it is very clear that the proposition is synthetic. I don't think you can call the definition synthetic or analytic because it is not a proposition, it isn't true or false.
Let me take a different tack, since my sense is you're rejecting the whole approach. Do you think the synthetic/analytic divide is absolute? That is, do you think Quine wrong? If so, why?
I don't mean to reject your whole approach, I was simply trying to reach clarity as to what you were proposing. In my mind there are two separate things at work here, as I said in my previous post. After those two things are made clear feel free to argue one way or another.
I do think that the analytic/synthetic divide is absolute and I do think that Quine is wrong. Part of the problem is that Quine is a pure empiricist so when he goes looking for analytic statements he does so as an empiricist. Hence the only thing that for him would prove something as analytic would be empirical evidence. However, that's impossible, it's like asking for scientific proof of God. By design, science says nothing about God, so anything you do scientifically will leave you with no proof.
I have read "Two Dogmas of Empiricism" and find it curious that in looking for analytic statements Quine several times introduces artifical languages for his arguments. I don't find that persuasive because he has not shown much about the properties of the artifical languages themselves. Is there an exact equivalence between these made up languages and the languages of everyday usage? It's as if he makes up languages with the properties he wants and then uses that as proof for the conclusions he wants to make.
The reason I ask about whether you think the divide is absolute is because your approach to the question pre-supposes we can separate out definitions from "the world" whereas my approach is designed to undermine that. So your attempt at clarity is, to me, missing the fundamental issue.
To me definitions are always in terms of meanings that aren't quite as stable as I think the analytic taxonomy requires. But I recognize that to most who accept the divide I'm just not going to convince them of that.
My attempts at clarity are purely pragmatic. If there is not clarity in the issue then 1) I have no idea what you are saying and 2) You can pretend that my response means something to you, but it really doesn't as you have no clue what I was responding to.
Clarity is fundamentally about acknowledging that if you are going to communicate with people you have to communicate. Unclear communication prevents that as neither party has a clue what the other intends. Unclear communication is an admission of solipsism.
If you are not going to be clear then how am I ever to understand? The only option would be that I somehow, independently of experience, intuit what you are saying.
Well, an analytic proposition, such that "all bachelors are 'necessarily' unmarried and male", but not vice versa, is an odd notion because, as an identity statement, rather than a criterion for identification, it's virtually tautologous, such that the assertability conditions for such a proposition are dubious. As I said earlier, I don't think Kant's original distinction quite has that problem, since it's in contact with "common sense" in implying such assertability, and since Kant's interest is really in synthetic judgments, which he wants to claim must in some instances be accorded a priori status, in order for experience/knowledge to be possible. Which is what motivates reflection upon analytic judgments, in order to distinguish them from synthetic ones, showing the "necessity" of the former to be insufficient for understanding. (Again, I'll leave out of consideration the odd logical status of a transcendental argument, whereby something stipulated is "shown" to be not otherwise conceivable, except upon the projection of said conditions of possibility as "necessary"). What bothers me about the notion of an analytic proposition in Analytic philosophy is its relatively unmotivated status, except insofar as a formal-logical framework, whether construed as "syntactic" or "semantic", is to be distinguished from an actual empirical content or claim. (My sympathies are neither with Carnap, nor Quine: both seem to me to rather paradoxically hold to an instrumentalist notion of meaning, tied together with an entirely disinterested notion of truth/knowledge, by means of a formal notion of inference).
But what I really want to remark upon is how odd the example of "all bachelors are unmarried and male" is to illustrate an analytic proposition. "Bachelor" is virtually a kinship term, like "uncle", rather than something referent to a natural reality. In other words, a term that is partly constitutive of the nexus of relations and practices which determine what it is "about". Just as it's possible that maternal and paternal uncles might be identified by different terms/categories, or that maternal uncles and senior second cousins might be identified by the same term/category, so it's possible that, say, young/eligible bachelors might be distinguished from old/superannuated bachelors. And, by the same token, bachelors might be subject to non-standard or "deviant" usages. What of annulled marriages, under certain systems of rules, morganic marriages, or common-law relationships? With respect to that last, what's the difference after 6 years or after 8 years? A better example might be "cats are caniverous mammanls", except that there's a contingent discovery there, since all species of cats have a functionally disabled gene, such that they can't digest vegitable proteins, such that cats are necessarily carniverous, in contrast to other mammal groups. The upshot is that the analytic/synthetic distinction can't really be construed without considering its applications and purposes.
But aren't we really thrown back upon that ancient question of Platonic anamnesis, of concept acquisition/possession and how experience/knowledge is possible without it? Isn't the really relevant distinction not that between analytic/synthetic propositions, but between constitutive/regulative rules with respect to the uses/applications of words/sentences. It might well be that a given complex of concepts and their concommitant experiences, relations, and practices involve the holding constant of certain rules as constitutive sine qua non, while allowing for/rendering possible more flexible regulative usages elsewhere. This says nothing about their supposed "necessity" per so, nor about any "necessary" conceptual order, and it does not deny that their recurrent usages might result in change through the drift of those uses/applications, or through empirical discoveries. But, at the same time, such complexes might have a relative stability precisely through their recurrent connections between meanings, experiences, practices and relations.
Dave, I guess what I'm saying is that I think what you are taking as a matter of clarity is, in my eye, a matter of pre-conceptions. Sometimes what makes something clear is a metaphysical assumption that is incorrect. The premise of the post is Quine's position and how one would deal with common analytic statements given Quine's claims. So I'll turn things around and ask you how you'd take an analytic statement and give an example of how it isn't really purely analytic.
The problem, as I see it, is that for Quine (following the positivists rather than Kant necessarily) is asking about meanings. So one need only demonstrate that for a given analytic statement its meaning isn't really entailed by definitions. (That is there is something in the context, if changed, would change our judgement) At least that's how I understand Quine. (I'll confess to getting Quine wrong at times so I may have him wrong here)
Apologies in advance if this is entirely misguided...
But from this discussion, it sounds like the problem with the analytic/synthetic divide is connected to the vagueness of our concepts.... like in the bachelor example, we are not really sure if you're still considered a bachelor if the paperwork is lost, you don't remember anything etc.
Now I always thought Quine's point had more to do with the way we adjust our beliefs and concepts. In connection with the "meanings just ain't in the head" line of thought I guess. The scenario I would come up with to prove "all bachelors are unmarried" as synthetic would then be something like:
Let's say that after reading about this discussion, I hear the word "bachelor" used more and more often to refer to men I know to be married. So, I look it up in a webdictionary and the definition for bachelor there is: "Young man" or something like that. In that case I'd probably judge, based on the empirical evidence, that "bachelor" doesn't mean "unmarried male".
Pragmatism is the key word here, methinks.
John, the issue of whether Quine holds to an Dewey-like instrumentalist view of meaning is an interesting one. Sadly I don't know enough about Quine to really speak there. Certainly he has behavioralist tendencies.
Regarding your other points, I agree.
The issue is then what would count as an analytic statement. Most turn to mathematics, I know. I'm not sure that ends up working any better. But that then gets into the whole foundations of mathematics debate where counter-examples are harder to give.
Jonas, aren't you conflating Davidson and Quine a bit? It seems to me that the "where" of what grounds meaning is very different from both. For Quine it is something close by whereas for Davidson it is something closer to the objects themselves. But clearly this issue of externalism probably does have some bearing.
If I recall correctly, the upshot of his (Quine's) discussion comes in the last section about "Empiricism without dogmas".....
I'm quite confident that he says things like "No statement is immune to revision" and that the decision on how to adjust your "web of belief" in the face of new experience is a pragmatic question. Therefore, my thought that the most sensible way to adjust my beliefs when faced with the problems I described concerning "bachelor" would be to say that it means something different from what I thought. (I could, of course, insist that the bachelor-unmarried male connection is analytic, and everyone around me simply wrong, but I could eventually end up having no reason to other than stubbornness)
Another way to put it would be: almost ANY statement could be called "analytic", if that means that it's true independently of experience...... but this just means that no matter what experiences you have, you'll rather change some other part of your "web" of beliefs. Of course, describing it this way also robs the "analytic" label of any philosophical power.
Not quite sure what you mean concerning Quine and Davidson, perhaps you could elaborate a bit? Those two do seem quite alike in my mind, so you might be right about the conflating.....
I was more thinking about the "meanings aren't just in the head" bit.
You are right that many aspects of Quine's thought are quite pragmatic. There's still a debate whether to call him a pragmatist proper. But the term itself is fairly wide. I mean Dewey is quite different from James and Peirce. Of course Putnam and Rorty, who often are called neo-pragmatists, are even more different.
The issue of the analytic/synthetic divide in pragmatism is an interesting one I'd not really considered. (Surprisingly)
I can't speak to Peirce and James too much in that regard. Although Peirce appears to accept the a priori category whereas Dewey seems to attack it. (Probably arising out of Dewey's starting point as a Hegelian) When Quine is called a pragmatist in this regard the idea is more Hume than Peirce though. The idea that meaning is tied up in observable consequences rather than origins is what is called pragmatism. Certainly that's a big part. And I think Peirce's notion of meaning would get wrapped up in that. However Peirce was also a formalist with respect to logic (unlike Dewey), thus the a priori. But I think one has to separate logic from meaning for Peirce.
So I guess Peirce would say we have to separate out the logical (analytic) aspects of entailment from the meanings themselves. Put an other way, garbage in garbage out. Just because you think a logical argument applies doesn't mean it actually does.
But that ends up getting to I suspect David's critique. The issue isn't whether all bachelors are unmarried men is analytic but whether it is correct.
Sometimes what makes something clear is a metaphysical assumption that is incorrect. I agree. If am clear then I can be shown to be wrong, whereas if I am unclear I can always hide behind the excuse that my respondent didn't get it.
So I'll turn things around and ask you how you'd take an analytic statement and give an example of how it isn't really purely analytic. This was a unclear to me, so I assume that we are to proceed as follows 1) I am to give an example of an analytic statement, 2) I am optionally to defend it as an analytic statement, and 3) You will show that it is not purely analytic.
I guess the best example of an analytic statement would any purely formal proposition, such as the conclusion of a valid syllogism, a tautology, or contradiction. I guess one could argue that they are trivially analytic, but they are still analytic. Come to think of it the proposition "Trivially analytic truths are analytic" is also analytic because it relies on the meaning of the words. Unless of course you change "trivially" to mean "not analytic" in which case it becomes a paradox. However, in that case since the meanings are different I would not assert it as analytic.
By the way, this whole exercise has really got me thinking about analytic/synthetic, a priori/a posteriori, and necessary/contingent. It is much more complicated that I originally thought, but I still think that the analytic/synthetic distinction exists.
Except that a purely logical statement independent of meaning isn't what is at question. That is something like
All A are E
All B are A
B
therefore E
isn't really what's being discussed.
Except that a purely logical statement independent of meaning isn't what is at question. You asked for an example of an analytic proposition and I obliged.
Here again, clarity is needed. There are two ways of going about defining analytic propositions, that of the logical positivists and Kant's way. I assumed that you have been talking about the logical positivists' way. On that view I gave a perfectly valid example of an analytical proposition. A purely formal proposition is something that is true solely by the conventions of language, which is one way the logical positivists defined analytical statements.
Another way of looking at analytical statements, used by the logical positivists is that the proposition whose truth depends solely on the meaning of its terms. In that case, yes, there is room to argue. I still think that these exist, but it makes a much easier target for saying that it involves a synthetic judgement in arriving at the meaning of the terms, which is the tactic I presumed you would take in refuting it. This is much more open to attack because the logical positivists lumped analytic and a priori together, so to attack one aspect of analytica/a priori is to attack the whole.
I've closed comments in order to avoid spam since I don't check this older blog as much anymore.
and Theology
Number of unique visitors:
Blogged by Clark Goble