Leiter Reports: A Philosophy Blog

News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.

  1. Fool's avatar
  2. Santa Monica's avatar
  3. Charles Bakker's avatar
  4. Matty Silverstein's avatar
  5. Jason's avatar
  6. Nathan Meyvis's avatar
  7. Stefan Sciaraffa's avatar

    The McMaster Department of Philosophy has now put together the following notice commemorating Barry: Barry Allen: A Philosophical Life Barry…

What Kripke Should have Said (J. Stanley)

Jason Stanley

In a recent paper, one of Kripke’s many empirical arguments against the description theory in Naming
and Necessity has been challenged. This example concerns a fictional
scenario in which someone named “Schmidt” in fact was the first to prove
Godel’s incompleteness theorems; Machery, Mallon, Nichols, and Stich argue that speakers in Hong Kong think that “Godel” then refers to Schmidt. If correct, this is a good contribution, since it would give us information about which claims in a certain branch of semantics are most robust. But it is hard to see any kind of broader significance here. First, the case is one that
Kripke himself regarded as pretty shaky (Kripke writes “But it may seem to many
of you that this is a very odd example”), which is why he immediately follows
up the fictional case by providing an actual case – the case of “Peano”
(Machery, Mallon et. al. do not even attempt to argue that
mathematicians in Hong Kong think that “Peano” refers to Dedekind). And indeed even if we
abandon all cases, fictional and actual, of the “Godel”-“Schmidt” variety,
Kripke still provides an overwhelming case against the description theory of
names in Naming and Necessity. In forthcoming work, Mallon, Machery et. al. go on to argue
against what they call “The Method of Cases” (basically appeals to speaker
intuitions about truth-conditions) solely on the basis of their one experiment
about “Godel”-“Schmidt”. This is an argument directed against the entire
methodology of linguistic semantics. To base a case against a branch of
science, practiced in the main by people completely ignorant of philosophy, on
a single experiment about a rather insignificant claim made by a philosopher thirty
years ago seems a bit odd, a bit like rejecting the entire methodology of
chemistry because of a single false claim made in a largely true article in one area of the subject.

Update: As Justin Sytsma points out below, I messed up a bit in the original post, because in the original paper the study was of an English-speaking group of Hong Kong residents – so the experiments were uniformly in English. I have adjusted the post accordingly.

Update: To see the newer comments, please press ‘next’ at the bottom of the thread.

Leave a Reply to Simon Rippon Cancel reply

Your email address will not be published. Required fields are marked *

50 responses to “What Kripke Should have Said (J. Stanley)”

  1. Jussi Suikkanen

    Here

    http://experimentalphilosophy.typepad.com/experimental_philosophy/2008/06/michael-devitt.html

    is also a link to a very illuminating paper by Michael Devitt on the same issue which argues along similar lines (and a good discussion about it).

    Reply
  2. jonathan weinberg

    "And indeed even if we abandon all cases, fictional and actual, of the “Godel”-“Schmidt” variety, Kripke still provides an overwhelming case against the description theory of names in Naming and Necessity."

    Well… isn't that a perfectly good reason to give up on the "method of cases", then? That is, if it's methodologically troublesome (for MMNS reasons), _and_ it's not really needeed in the first place, why not chuck it?

    I'm also very unsure about the attempted assimilation of the philosophers' method of cases to the linguists' method that might go by that name. For starters, in linguistics, diversity of intuitions is very easy to accommodate, as one can relativize to language. (I'm a Southerner, and my intuitions tell me that many double modal constructions are fine; I'm guessing they don't strike your ear the same way; but linguistics feels no need to rule one of us right and the other wrong.) Is the philosophical literature on reference so unproblematically free & easy in its relativizing?

    Reply
  3. dan kaufman

    Amen!

    Reply
  4. Jonathan Ichikawa

    I think this is right. Michael Devitt has a recent paper draft saying many of the same things.

    Reply
  5. Jon Cogburn

    The first, eponymous, essay in Stich's "Deconstructing the Mind" gets at the heart of Kripke's case most charitably. Stich presents a number of cases where we could rationally go with something more causal or something more descriptive. [This is, I think, a consequence of what Dummett calls this "the inextricability of reference and belief" in his essay on Quine; unfortunately while Dummett endorses the thesis he fails to draw the conclusion that his own "molecularism" is undermined by it.] For example, we tend to say that we are talking about the same thing Bohr was talking about by "atom" even though the descriptions are radically different (something like the causal theory here). But we say that "phlogiston" doesn't exist, even though (as Putnam points out) we could with some justification call negative valence electrons incidence of "phlogiston" (shoot, maybe that was caloric). Here the failure of the description associated with phlogiston was sufficient to say that it did not exist.

    The big point is that even if reference is (in Mark Wilson's "Predicate Meets Property" sense) this underdetermined, descriptive theories are still false. But so is the causal theory, in part for the reasons Kripke gave. The very fact that there are actual cases of referential history where we have gone with the causal intuition, and that communication would be impossible if we were always prevented from rationally doing so, undermines the descriptive theory.

    I think Richard Rorty had it correct when he said that a disagreement about specific cases of referential meaning are usually just agreements to halt the conversation. To decide that such a thing is "merely semantic" and "not empirical" is to decide that there is no good point to arguing about it any more. As long as one is not a Rortyan relativist generally about there being good and bad things to argue about, nothing of import is lost by this reversal in explanatory goals. One can contrast the Rortyan position with Russell's in "The Problem of Philosophy." Russell argues that universals must exist to insure agreement of meaning, because the ideas in our own minds are too different. The Rortyan perspective says that this is exactly backwards. Yes the stuff in our minds are radically different, but universals are reified projections of our normative agreement to agree that we're talking about the same stuff.

    That is, the converse of Rorty's point is that an agreement about meaning is a normative agreement to agree. But then (insert Wittgensteinian argument here) theories of reference (and perhaps some lexical semantics) maybe commit something like the naturalistic fallacy. However, none of these worries about the semantics/normative pragmatics of referential terms undermines compositional semantics of the Montogovian kind covered in Shalom Lappin's "Companion to Formal Semantics" (Blackwell). Likewise as long as one is sufficiently objective about the normative pragmatics of argumentation, no concession to Rortyan relativism has been made.

    Reply
  6. Josh Dever

    All good things for Kripke to have said, Jason. But he could also have said this: "If you'll read carefully what I said in N&N, it goes like this: 'Suppose Gödel was not in fact the author of this theorem etc.'. So a certain counterfactual scenario is being picked out in Gödel terms. Thus there's just no interesting question about who the name 'Gödel' refers to in this scenario. Trivially, the name 'Gödel' refers to Gödel, and I've told you who Gödel is in the scenario. Everyone has to agree with me on that, whether they're Millian or descriptivist. The real question is whether the scenario is metaphysically possible in the first place – the rhetorical strategy is to distract you with the (in fact trivial) question of how the names refer so that you forget to object to the very possibility – and then your tacit acceptance of the possibility reveals the Millianisn which underwrites your own referential practice. So when someone runs an experiment claiming that some people think that 'Gödel' picks out Schmidt in the example, that at most shows that some people get confused easily. But in fact, it doesn't even show that, because the experiment uses the prompt 'When John uses the name ''Gödel'', is he referring to etc.', which pretty flagrantly runs together questions of speaker reference and of semantic reference, which I could have sworn I wrote a paper on back in the day.

    Reply
  7. Jason Stanley

    In response to Jonathan Weinberg:

    The 'method of cases' in this case is simply empirical data gathering, of exactly the sort semanticists do in linguistics. If Machery et. al. are correct in their claim that the "Godel" type cases are not convincing, this has nothing whatsoever to do with the method of cases. It just would show that this particular empirical argument is not convincing. If Machery et. al. are right, then Kripke was wrong to base arguments upon this particular case. But why does that supports any kind of general conclusion about anything is beyond me [I am not, however, endorsing the experimental set-up of their case, and indeed find it problematic]

    Kripke's arguments in NN are empirical arguments about natural language. As I write in the first sentence of my 1996 paper, "Names and Rigid Designation", "The fact that natural-language proper names are rigid designators is an empirical discovery about natural language." Kripke uses empirical premises to draw interesting philosophical conclusions. We naturalists are fine with that methodology.

    As to language relativity – it could be that Kripke's conclusions about reference are language-specific, and in some languages, what look like proper names function instead like definite descriptions. Nothing in NN precludes this, and it would be an interesting discovery if true. Or it could be that the non-descriptionality of proper names is a linguistic universal, like certain model-theoretic properties of determiners (e.g. conservativity). But even if Machery, Mallon, Nichols, and Stich were right about "Godel", they wouldn't have shown this, since Kripke has a battery of arguments against the descriptionality of proper names.

    Reply
  8. Jason Stanley

    Right, Josh. That in fact would be a better thing for Kripke to have said, since it shows that the Machery, Mallon, Nichols, and Stich paper actually doesn't raise any problems for him. I regret the implicature of my original post, which was that I thought it did (I certainly didn't intend that implicature, and am glad for the opportunity to cancel it explicitly). I basically was focussed on making the point that even if the case didn't turn out to be robust, no general moral would follow, either about Kripke's argument against descriptivism, or about the methodology of linguistic semantics.

    Reply
  9. Jeffrey C. King

    What Josh Dever said!

    Reply
  10. Justin Sytsma

    I find Kripke’s response to be reasonable (excluding his offensive attempt at a joke). He seems to be saying that people’s intuitions should be tested, but that in this case he would be very surprised if there was significant variation in their intuitions and that he therefore would want to look closely at the study.

    This follows one basic line of response that people have taken with regards to MMNS’s work: You can accept (perhaps just for the sake of argument) that if MMNS’s study shows significant variation (cross-culturally or intra-culturally) in semantic intuitions about Kripke’s Gödel case, then MMNS make a compelling case against a standard practice in modern philosophical work on theories of reference; one can then deny that their results show what they claim. Ludwig (2007), for example, suggests that there is a speaker’s reference/semantic reference ambiguity in MMNS’s probe that might explain the variation they found (and as Josh Dever notes). Jonathan Livengood and I have been conducting some follow-up experimental studies exploring whether there is a (somewhat broader) perspectival ambiguity in the probe (we find that there is and we argue that this explains much of the variation that they found).

    Another line of response is illustrated by Jason Stanley’s post, as well as Devitt (ms). Devitt argues (as I understand it) that while MMNS’s study does provide evidence against people’s semantic intuitions being largely uniform for one prominent fictional case, and while standard philosophical methodology often relies on the assumption that people’s semantic intuitions are largely uniform (whether they are well advised to do so or not), uniformity with regard to intuitions about this one fanciful case is not as important to Kripke’s case as MMNS suppose.

    I find both of these to be reasonable lines of response to MMNS’s highly provocative work. As such, I’m not sure that Kripke should have said something along the second line, as opposed to saying something along the first line (as he did).

    Reply
  11. Greg

    In defense of MMNS, one might say that they are not throwing into doubt the entire "methodology of linguistic semantics," i.e. "appeals to speaker intuitions about truth-conditions." Rather, they are only casting doubt upon appeals to speaker intuitions about reference, or the truth-conditions of sentences of the form X refers to Y. This still might be unacceptable, but it's (to use Jason's analogy) more like throwing out a particular sub-field of chemistry, instead of all of chemistry. And speaker intuitions are not sacrosanct (I'm no expert on this, but I do know that many people working on presupposition now (e.g. Mandy Simons and Kai von Fintel) think that speakers' intuitions about truth-value gaps are too stable and unreliable to serve as data for a theory of presupposition).

    But I personally did wonder, after reading both papers Jason discusses, whether our toolkit for investigating reference would be overly impoverished if we took MMNS's claims to heart.

    Reply
  12. Janice Dowell

    I couldn't agree more with Jason's response to Jonathan. It's pretty hard to see how the results about Cantonese tell us anything about whether the method of cases is sound. At most, it would seem to show that Cantonese is a language for which descriptivism is true. And that's fully compatible with Kripke's arguments in Naming and Necessity. He never argues that there couldn't be a language for which descriptivism is true, he merely argues that our language isn't one.

    But setting Kripke aside–is it just me or is anyone else finding it hard to imagine an efficient language that failed to have some Millian device for picking out individuals? That would seem to make the communication of information about specific individuals between speakers with different background assumptions about them needlessly difficult. The conversations I'm imagining in such a language would run like bad comedy routines. Such a language does seem possible, but it would be pretty unfortuante for its speakers if it were actual. That makes me inclined to suspect any experiment that purports to show of some langauge that it doesn't contain such devices.

    Reply
  13. Ron Mallon

    A range of really good points and criticisms here. Here’s a pass at a few of them:

    Jason, I’m not sure I understand your remarks regarding Mallon et al.
    With regard to Machery et al. The target of that second paper is “Arguments from Reference,” not the “Method of Cases.” Arguments from reference are arguments that presuppose a substantive theory of reference, and then go on to draw philosophically significant conclusions from it. We suppose that the Method of Cases is important to determining what the correct theory of reference is, an assumption that you (and others like Devitt) have lately attacked.

    I think you also overstate our reliance on the empirical result in the second paper (perhaps because you're not interested in the discussion of Arguments from Reference). Our use of the empirical result in the second paper is primarily to show that there’s a real question about whether there is variation in intuitions, but we the argument of the paper is explicitly marked as condition upon the assumption of variation is true.
    What we say about our results in the second paper is:

    “We have no illusions that our experiments are the final empirical word on the issue. This is a newly emerging type of research, and obviously it is too early to draw any definite conclusion about the variation of intuitions about reference. “

    And then just after we set up the argument of the paper:

    “While the empirical results reviewed in Section 2 are still preliminary, they constitute a strong prima facie case that intuitions about reference used to construct theories of reference might vary from culture to culture and person to person. In the remainder of this paper, we will assume that such variation does exist, and we will explore its implications both for the theory of reference and for arguments from reference.”

    The point of the second paper was to explore what variation in intuitions about cases would mean for “arguments from reference” on the assumption that the Method of Cases had a role to play. (It, for example, suggests that relativization to language is not philosophically innocuous in arguments from reference.) It provides no new arguments that such variation exists, but rehearses the empirical finding to show that it might. Even the final sentence of the paper recalls the conditional nature of the discussion.

    I’m (unsurprisingly) sympathetic to Jonathan Weinberg’s thought that we should abandon the method of cases (for philosophical purposes), and I’m curious about why that would leave us with an impoverished toolkit (as Greg suggested, if I’m interpreting him correctly).

    Josh Dever’s remarks are very clever, and perhaps Kripke should have said them. I also agree with him also (and with a number of others including David Chalmers, Max Deutsch, and Kirk Ludwig) that the original experiment does not determinately show that it’s the theory of reference that drives the different responses rather than differences assumptions about speaker vs. semantic reference.
    On the other hand, the original result was surprising, and it was found by forming a hypothesis on the basis of empirical data that existed before and finding variation where and in the direction we predicted it. That’s not easy or trivial to do, and we don’t believe that someone with understanding of the speakers reference vs. semantic reference would have predicted finding what we found. So, while it would be nice to have eliminated it, the failure to do so does not (we think) deprive the empirical result of interest. There’s always another interpretation of the data (in this case an interesting and independently plausible one), but the finding is surprising in light of the alternate interpretation. Still, it would be good to empirically test for it, if we could.

    Michael Devitt’s recent paper is very interesting, and raises the question about the variety of lines of evidence in favor of the causal theory (a point raised with me previously in conversation by Jonathan Adler), and it dovetails with the question as to whether the Method of Cases is really necessary as a source of evidence. This obviously needs to be answer, and with more care than I am able to do so here.

    In response to Justin, I also thought Kripke’s philosophical remarks were interesting, in part because it didn’t seem like he wanted to disavow the use of intuitions as evidence, nor did he think the empirical test was irrelevant. (He just thought that it was wrong because it got what he took to be the wrong result, which is an acceptable form of inference provided further argument follows). Anyway, I’d be interested to see your further empirical results. What Machery et al. did was certainly preliminary.

    Reply
  14. Justin Sytsma

    It is worth noting that MMNS's study was conducted in English (not Cantonese). It was conducted on two groups of undergraduate students, the first group at Rutgers and the second at the University of Hong Kong (an English-speaking University).

    Reply
  15. Jason Stanley

    Justin,

    Yes, you're right. My post should have been titled "One of the things Kripke could have said". I'm new to this debate, and what you report Kirk Ludwig having said, which is one of Josh Dever's point, seems exactly right to me about their experimental set-up. I was just (I guess echoing Michael Devitt) pointing out that even if one accepted their result, it doesn't show anything general, either about the argument for the description theory of names, or even this style of argument for the description theory of names (since we still have the "Peano" case), or semantics generally.

    Reply
  16. Jason Stanley

    Ron,

    Yes, you're right, you're not really arguing against the method of cases. You are arguing that the method of cases doesn't support arguments from reference. But I disagree with this, since I do think the method of cases in fact supports arguments from reference. First, I agree with you guys that 'The Method of Cases' underlies Arguments from Reference. You raise the possibility that the method of cases does not leave us with a good argument for arguments from reference. But the evidence is just the one piece of data about the "Goedel" case. But the "Goedel" case is just one of Kripke's many arguments against the description theory of names (most of which *do* involve this or that employment of the Method of Cases – so I'm not challenging the claim that the Method of Cases is important to Arguments from Reference). At most, even if we were to accept the experimental set-up of the original paper, what follows is one of the many instances of the "method of cases" fails to be convincing (a result made somewhat problematic and mysterious by the fact that the "Peano" case presumably is convincing). This shows nothing general about employing the method of cases in support of arguments from reference. For that, one would have to take on the whole battery of Kripke's empirical arguments.

    So, I took you to be employing the earlier result to show that there is a problem in using the method of cases in favor of arguments from reference. And then I generalized to take you to be casting aspersions on using the method of cases more generally in drawing conclusions about linguistic meaning. I see no general problem with using the method of cases in drawing conclusions about semantic content, though there may be bad applications of it (as I guess Ludwig and Dever point out, the "Goedel" case isn't one of them).

    Also, on the interpretive point, I can't be totally amiss in reading you guys as wanting to reject basing arguments on the method of cases, since as you say in your comment you are "unsurprisingly sympathetic to Jonathan Weinberg's thought that we should abandon the method of cases". So you start out your comment by saying you have no problem with the method of cases, it's just arguments from reference that bother you, then later you say that we should find it unsurprising that you are sympathetic to the view that we should abandon the method of cases, at least for philosophical purposes. So you must think that one lesson of your paper is that philosophers shouldn't appeal to the method of cases. That's what I was responding to. There is no worry here for the method of cases, even in support of arguments from reference.

    Reply
  17. Josh Dever

    Ron, of course the mere fact of alternative explanations of the data (such as an explanation running on speaker's reference) does not obviate the interest of the explanation of that data in terms of variation in theory of reference. But on the line I was suggesting, the worry about speaker's reference/semantic reference is an incidental point – my main thought was that an explanation in terms of variation in theory of reference is simply unavailable, because the scenario has been described as a Gödel-involving one in a way that guarantees that the word "Gödel" refers to Gödel. The prompt says that Gödel is the person who got hold of the manuscript and claimed credit for the work; how could any other answer about the referent of the word "Gödel" be available?

    So before invoking considerations of speaker's reference vs. semantic reference, there's already a trivial right answer available to the question at the level of semantic reference – and trivially right independent of one's views on the semantics of names. What remains to be explained is why people sometimes give the wrong answer to the trivial question – here the thought is that, due to the ambiguity of the prompt, some people read it as a question about speaker's reference, and give an arguably correct answer to it so read. (That there would be cultural variation in the pragmatic devices governing the choice of interpretation of an ambiguous prompt would on no one's view, I take it, be a surprising or disturbing result.) As you say, the hypothesis that some survey participants are engaging a notion of speaker's reference is only one hypothesis, and other hypotheses should be considered. But what that hypothesis, and the possible alternative hypotheses, are explaining is not an apparent divergence from Kripkean semantic theses.

    Reply
  18. jonathan weinberg

    I honestly am just not getting Josh's argument here, and I don't know which one of us is confused. Here's how things seem to me.

    Consider the possible world in which I, Jonathan Weinberg, am born into circumstances where my parents decided to follow through with the joke they kept telling my grandparents, and they give me the name "Aloysius". And suppose that they then named my younger brother (who in the actual world is Stephen Weinberg (no, not _that_ Stephen Weinberg)) with the name "Jonathan".

    Now, we can ask: in that possible world, who does the name "Jonathan Weinberg" refer to? The intuition I get (though I often have abnormal intuitions about Kripke-related cases) is that, in that world, "Jonathan Weinberg" refers to the person who, in the actual world, is my brother Stephen, and I am only referred to by the name "Aloysius Weinberg". I take it that this is the result that most standard theories of reference would give as well.

    However, if we run a _mutatis mutandis_ on Josh's argument (swapping the description "was given the name 'Aloysius Weinberg'" for the description "got a hold of the manuscript, etc."), we end up with the following: "…the scenario has been described as a Jonathan Weinberg-involving one in a way that guarantees that the word "Jonathan Weinberg" refers to Jonathan Weinberg. The prompt says that Jonathan Weinberg is the person who was given the name "Aloysius Weinberg"; how could any other answer about the referent of the word "Jonathan Weinberg" be available?"

    The problem, it seems to me, is that when I use my name in telling the story, those tokens take on their actual linguistic properties from our world; but when I only mention my name in the story, I am thereby referring to a linguistic object in the world of the story, and that object thereby referred to may well have different linguistic properties in that world than it does in our actual world. It's a possible world where, though Jonathan Weinberg inhabits it, "Jonathan Weinberg" doesn't pick him out. And so "Jonathan Weinberg" is just not going to be a trivially correct answer to the question, "To whom does 'Jonathan Weinberg' refer?" Such are the vagaries of discourse about possible worlds with entities from the actual world in them.

    What am I missing?

    Reply
  19. David James Barnett

    The number of people agreeing with Josh Dever has got me wondering if I've totally misunderstood Kripke. But my recollection is that the Godel case was presented only after Kripke had rejected the strong descriptivist view that the meaning of a name can be captured by a definite description–which he refuted with modal arguments about how Aristotle could have been a fisherman instead of a philosopher. It is only then that Kripke turns to the weaker descriptivist view that a description serves merely to "fix the reference" of a name. The Godel case was intended to refute this weaker form of descriptivism, which is soft of like an early version of two-dimensionalism.

    Josh Dever claimed that the Godel case was supposed to work like this:

    "'Gödel' refers to Gödel, and I've told you who Gödel is in the scenario. Everyone has to agree with me on that, whether they're Millian or descriptivist. The real question is whether the scenario is metaphysically possible in the first place…"

    I agree that Kripke has told us who Godel is, but this just means he has told us which person in the example is the one *we* call 'Godel'. The question of what fixes the referent of a name turns on who the people *in the example* refer to by 'Godel'.

    If Kripke's argument worked in the way suggested, then couldn't it be used as an argument against the obvious platitude that it would have been possible for people to have called Schmidt by the name 'Godel', and to have called Godel by some other name?

    Reply
  20. David James Barnett

    Sorry, it looks like Jonathan Weinberg made a similar point while I was writing my post.

    Reply
  21. Jason Stanley

    Josh made several points. One is about the fact that MMNS ask "who is John talking about", which is a question about who John has in mind, rather than what the referent of the name 'Godel' is in his mouth. It actually seems quite clear that in the scenario described, John had Schmidt in mind rather than Godel (which makes me wonder about my Rutgers undergrads…). But this vitiates the experimental design as a test of what the semantic content of 'Godel' is in the relevant situation.

    As to the point currently at issue – at the very least, Josh has shown that the experimental design of MMNS's set-up guarantees that anyone who answers (A) or (B) is treating "Godel" as not synonymous with "the man who discovered the incompleteness theorems" (since anyone who treated "Godel" as synonymous with "the man who discovered the incompleteness theorems" should reject the description of the possibility as incoherent). So in fact, by showing that speakers in Hong Kong treat names as not being synonymous with definite descriptions, MMNS's experiment confirms a good deal of Kripke's argument.

    Reply
  22. Josh Dever

    Jonathan and David, what's at issue in the Godel example is Kripke's Thesis 3 of descriptivism: (simplifying slightly) if there is a unique F, then it is the referent of N. Now Kripke admittedly makes things harder than they need to be by starting off with a fictional example, since the thesis itself is non-modal. For this reason, as Jason has emphasized, the subsequent non-fictional examples make the same point more straightforwardly.

    When considering the fictional example, we thus have to figure out how to bring it into engagement with the non-model Thesis 3. What is needed is that the language of the fiction simply be our language – without that assumption, the fiction is argumentative inert with respect to Thesis 3, and Kripke's evaluation of the scenario is groundless (without continuity of language, the thesis is underdescribed – there are fictional scenarios like that described by Kripke in which "Godel" (in the language of the fictional community) picks out Godel, and others in which it picks out Schmidt, and others in which it picks out Dedekind, or anything else we like).

    I see two immediate ways to enforce the linguistic continuity requirement. One is in effect to build it into the scenario, by saying "Imagine that we had all thought that the name "Godel" picked out the discoverer of etc, but in fact it didn't, because Godel had stolen the theorem from Schmidt". This is, as in my original line on the thought experiment as given, in a sense question-begging, but it achieves its purpose when we agree that that's a real possibility (rather than when we make the right evaluation about reference).

    The second way is to treat the fictional scenario as a counteractual description of our very own linguistic practice. But in this case, my original line applies unchanged – since this just is us, using our language, the word "Godel" refers to Godel, and we're told who Godel is. (So, picking up on David's description, we are the people in the example. This then doesn't tell (fortunately) against the possibility of Godel being named "Schmidt", since that possibility would be counterfactually realized in a community with a different linguistic practice).

    (Any obscurities in this post are entirely the result of a toddler clinging to my ankles as I type.)

    Reply
  23. Kevin Schutte

    Since no one else is doing it, I've got to put in my "ditto" for what Jon Cogburn said above. I think he got it exactly right.

    Reply
  24. jonathan weinberg

    Jason, your comment of 5:04pm seems to me to fall prey to the some of the same worry that some of us have raised about Josh's comment: you don't seem to be taking there to be any room for a difference in meaning between "Goedel" as used to describe the scenario, and "Goedel" as a linguistic item talked about in that scenario. It may be clear that the people answering the survey do not treat "Goedel" as synonymous with "the man who discovered the incompleteness theorems" _in their own actual-world discourse_. But the question at hand is what meaning they assign, when asked, to that name _as uttered by John in the scenario_. And there's no reason to think that these two "Goedel"-uses should have the same meanings.

    Reply
  25. jonathan weinberg

    "What is needed is that the language of the fiction simply be our language – without that assumption, the fiction is argumentative inert with respect to Thesis 3…" I'm sorry, but I'm just not seeing this. What it needs to be is a language _that works like our language_; I agree that that much continuity is absolutely necessary. But there's no reason for each & every linguistic item in the world of the fiction to have the same meaning as it does in the actual world. For example, one could have made do with entirely made-up fictional names that have no referents in the actual world whatsoever (like Calvino's "Qfwfq"). I'm just not seeing where anything in this part of the argument needs to turn on the particular names that we actually happen to use, let alone their being used in the fiction in exactly the way that we do use them in actuality.

    Reply
  26. Jason Stanley

    Jonathan,

    I hope it falls prey to the same worries, since the worries are misplaced. If we imagine a scenario in which the meaning of "Goedel" is say the same as "the book on my coffee table right now", that doesn't show anything about the meaning of our term "Goedel". For Kripke's argument to have any weight at all, the scenario needs to be about *our* term "Goedel". No conclusion whatsoever about the meaning of names in our language could follow from scenarios involving names in other languages (imagine someone arguing that "Goedel" isn't a rigid designator by arguing that we can imagine a fictional scenario in which someone else was given that name – that would be a very bad argument).

    Kripke's argument is supposed to work like this. Imagine a fictional scenario in which our term "Goedel" exists, with the meaning it actually has. In the scenario, things went a different way, and Schmidt had discovered the incompleteness theorems. What would the referent -the extension- of "Goedel" be, relative to that fictional scenario? Kripke claims that it would be Goedel, or else it wouldn't be our term "Goedel" – it would be a term with a different meaning.

    I've long felt it's an excessively complex thing for people to imagine. It's hard not to confuse the question of what some other linguistic community would mean by a similar sounding word with the question of what the meaning of *our* term would be as used there. Only the second question is at issue in this debate, not the first. The non-fictional "Peano" case therefore is much better. It's just obvious that a mathematician whose only belief about Peano is that Peano discovered the Peano axioms for arithmetic nevertheless speaks truly when she utters the public language sentence "Peano was Italian" (even though Dedekind was German). It's somewhat less obvious, but still quite robust, that if the knowledge is lost that Dedekind originally formulated the Peano axioms, and everyone only associates "Peano" with the information that he first formulated the Peano axioms, the statement is still true (though "Madagascar" type worries enter in when the name starts to have other causal associations).

    Be that as it may, let's distinguish the claims about the strength of the intuitions from the purely structural point that it had better be *our* term "Goedel" in the fictional scenario, with its actual meaning, or no conclusion can be drawn from intuitions about that scenario. But if the meaning of the term "Goedel" was synonymous with "the person who discovered the incompleteness theorems", then the description of the scenario would be totally incoherent.

    (As D.J. Barnett points out, the description of the scenario isn't obviously incoherent, if we think of the description as being a reference-fixer, rather than giving the meaning of a name. But these aren't subtleties attended to in the MMNS paper. The fact that the Hong Kong subjects find the description of the scenario perfectly fine shows that they do not take "Goedel" to be synonymous with the definite description "the person who discovered the incompleteness theorems").

    Reply
  27. Ron Mallon

    Really interesting…

    In response to Jason’s earlier post of 2:37pm: Okay, good. That seems much clearer. You are right that I am skeptical of the method of cases, but that doesn't mean that the one finding by Machery et al. is the only reason to be so! Your original post charges us with making a philosophical mountain out of an evidential molehill, something I don’t think we do.

    In response to Josh Dever’s point, and following on Weinberg’s and D.J. Barnett’s posts:

    It also seems to me that the following interpretation of our cases is possible. When a subject hears the name “Godel” introduced in the story, the subject fixes the referent in some way consistent with choosing the descriptivist answer to the probe. For example, to use the real material from our probe, she might attach the description “the guy who John learned proved the incompleteness theorem.”
    Because the question asks about how a mentioned name refers in a possible world (in which a mentioned description is widely believed), it doesn’t require subjects to actually endorse the description that the person in the possible world does. (The experimental subject can have her own, different description attached to the name). So, as D.J. Barnett suggests, the word "Godel" uttered/thought by the experimental subject might then have a different referent than "Godel" for the person in the other possible world, but that seems okay. You’d get a case where the person could say that, for a person in that possible world “Godel” doesn’t refer to Godel (or “Jonathan” to Jonathan”). As Jonathan pointed out, that seems perfectly coherent.

    If this is right, then, while Jason is right to say that:
    “the experimental design of MMNS's set-up guarantees that anyone who answers (A) or (B) is treating “Godel" as not synonymous with "the man who discovered the incompleteness theorems."
    He’s wrong to infer that that shows that “speakers in Hong Kong treat names as not being synonymous with definite descriptions.”
    They could be treating it as synonymous with a different definite description than the one widely believed in the world described by the vignette.

    Jason says “these are not subtleties attended to in the MMNS paper.” I’m not sure what this means. The obvious description for the experimental subject to attach to the name is not “the discoverer of the incompleteness theorem” but rather “the person who John believes discovered the incompleteness theorem.” But if that’s right, the problem doesn’t arise. It seems like we can answer beliefs about how "Godel" refers for John's community given our theory of reference just fine.

    Reply
  28. Colin Phillips

    This discussion is a little off my home turf (I'm a psycholinguist rather than a philosopher, and I happened upon this discussion by accident), but it inspired me to go and take a look at the Machery et al. study that started the discussion. I would be reluctant to read too much into the students' judgments on the 'Goedel-type' items, which appear to be a focus of the controversy. The mean and standard deviation for the Rutgers students in the study are very close to what one would expect if they had no idea how to respond, and hence were simply giving random answers to a question that struck them as odd. (Frankly, I can sympathize with them.) This does not, of course, explain why the Hong Kong students managed to give slightly more self-consistent answers, but it would seem premature to read too much into the philosophical import of the group contrast.

    Reply
  29. Eric Schliesser

    For present purposes, I am agnostic on the merits of Kripke's arguments or the MMNS paper. But I as struck by one of Jason Stanley's claims in his original post that has not received comment. He writes "This [MMNS] is an argument directed against the entire methodology of linguistic semantics. To base a case against a branch of science, practiced in the main by people completely ignorant of philosophy, on a single experiment about a rather insignificant claim made by a philosopher thirty years ago seems a bit odd, a bit like rejecting the entire methodology of chemistry because of a single false claim made in a largely true article in one area of the subject." Now, let me grant Stanley's main point that, of course, no group of scientists will give up their methods under these circumstances. Nevertheless, there are three things at stake here: 1) what authority do mere philosophers have to challenge the methods of a *science*; 2) what authority do the methods of one branch of science have to challenge the methods of another branch? 3) how divorced from philosophy can any of the human sciences ultimately be even if its practitioners wish/claim to be ignorant of (the various branches of) philosophy. The third point is of particular interest to me because I just wrote a paper on economists' claims to being value-neutral experts (or doctors) in the context economists's entanglement with Pinochet in the 70s and 80s. It's forthcoming in a volume on Chicago Economics, but a draft can be viewed here: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1142741)
    I think most folks commenting on Stanley's post are skirting around the second issue–can the methods of experimental psychology (as practiced by philosophers?) challenge the methods of linguistic semantics? Historical precedent suggests that the methods of one human science can replace the ones in another (look at the use of economists' tools in political science and sociology), but perhaps some form of co-existence between intuitons and experiment will evolve.
    Of course, the first issue is most tricky. Philosophers may wish to challenge all kinds of methods in branches of science; some sciences (and its members) are more welcoming to constructive criticism than others (although one should not underestimate the potential fierceness of the response), but (in general) unless one offers a better toolkit they will be unreceptive to even very sound concern.

    Reply
  30. Eric Schliesser

    I apologize, but several of your readers noted that the link to my paper should be: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1142741

    Reply
  31. Jason Stanley

    Judging from our discussion, I think I would now like to add another criticism of the experimental design to the ones familiar from the literature: it also fails to distinguish between (a) intuitions about what some other speech community would refer to by a similar sounding word from (b) intuitions about what our word would refer to, had the facts been otherwise.

    To make matters worse, there is also, as D.J. Barnett points out, the 2-dimensionalist reading of (b) – where we are considering another world as actual, but keeping the meanings of our words fixed- which also needs to be distinguished from (a). So that's three different questions that would need to be distinguished in a good experiment of this sort.

    Granted, for the reasons Josh pointed out, these are particularly difficult in the fictionalized "Godel"-"Schmidt" scenario; one has to be particularly sensitive to the distinction to do so, and perhaps only trained philosophers could be. So using an actual case would be better, as Kripke goes on to do in NN.

    Reply
  32. Jason Stanley

    Colin,

    Thanks for your input. Yes, MMNS should have noticed that their data does not support the conclusion that "Westerners are more likely than East Asians to report intuitions that are consistent with the causal-historical view". Their data just supports the conclusion that Westerners are extremely confused by the experimental set-up.

    Reply
  33. Kirk Ludwig

    (1) I think Josh (Dever) is right that, given way the probe is presented in the Machery et. al. 2004 paper, there is only one right answer to the question who 'Gödel' refers to as used by John, the protagonist of the story. This is something I also pointed out in the 2007 paper in Midwest Studies. However, I think Kripke's own discussion is more subtle. It is true that Kripke uses 'Gödel' in setting up the discussion to refer to someone who was not the author of the theorem. But he asks not about his use of 'Gödel' in describing the situation but about that of "our ordinary man." Our ordinary man in this case is someone who associates with 'Gödel' (aside from descriptions that are parasitic on the reference of others) only the description 'the discoverer of the incompleteness of arithmetic'. Here is the relevant passage.

    "Imagine the following blatantly fictional situation. (I hope Professor Gödel is not present.) Suppose that Gödel was not in fact the author of this theorem. A man named ‘Schmidt’, whose body was found in Vienna under mysterious circumstances many years ago, actually did the work in question. His friend Gödel somehow got hold of the manuscript and it was thereafter attributed to Gödel. On the view in question, then, when our ordinary man uses the name ‘Gödel’, he really means to refer to Schmidt, because Schmidt is the unique person satisfying the description, ‘the man who discovered the incompleteness of arithmetic’. … So, since the man who discovered the incompleteness of arithmetic is in fact Schmidt, we, when we talk about ‘Gödel’, are in fact always referring to Schmidt. But it seems to me that we are not. We simply are not." (pp. 83-4).

    It is open relative to this description of the circumstance that our ordinary man uses 'Gödel' to refer to Schmidt rather than to his friend, though Kripke uses 'Gödel' to refer to Schmidt's friend. For it is not a presupposition of the description of the situation that our ordinary man uses 'Gödel' to refer to the person Kripke uses 'Gödel' to refer to. Kripke does go on to use 'we' in saying 'we, when we talk about 'Gödel' [on the view in question], are in fact always referring to Schmidt', but it is clear here that the intent is to pick out those of us in the imagined situation who are like our ordinary man. So it seems to me that Kripke is not guilty of setting up the scenario in such a way that it begs the question.

    (2) Jason suggests that Kripke thought the Gödel/Schmidt case was "pretty shaky" because he wrote "But it may seem to many of you that this is a very odd example." But what he says after this doesn't suggest he thinks it is not a good case. He goes on as follows: "… or that such a situation occurs rarely. This is also a tribute to the education of philosophers. Very often we use a name on the basis of considerable misinformation. The case of mathematics used in the fictive example is a good case in point. What do we know about Peano? …." As I see it, Krikpe is not granting the claim that if such cases were rare or odd or even non-existent that it would not show descriptivist Thesis (2) was mistaken. He is just pointing out that there is nothing odd or rare about the type of case, so that the complaint gets no traction because inter alia it rests on a false assumption. And he shouldn't grant that the argument would be pretty shaky if such cases were rare because Thesis (2) entails that our ordinary man in that situation would be referring to Schmidt rather than to Gödel (as we use the term), and so (if Kripke is right) it entails something false. And that is a good reason to reject it.

    Reply
  34. Ron Mallon

    I suspect subjects do perfectly well understand the distinction between asking who John’s use of the name is about as opposed to their own. The vignettes and questions seems perfectly comprehensible, as does, e.g., Jonathan Weinberg’s story that raises similar issues.

    Jason’s more general point that there are lots of distinctions that could be in play is definitely right. But that doesn’t mean the experiment isn’t "good." To return to an earlier point: while it is possible that one of these other distinctions could explain the finding (for example, I think Josh Dever is right that it wouldn’t be surprising to find cultural differences giving rise to pragmatic differences leading to differential speakers vs. semantic reference choices), I take it that these other distinctions would not have led one to predict this difference in this direction in advance in the way we did before finding the difference.

    Colin is right that the high standard deviations are of concern. We’ve wondered whether this indicated within group diversity (since, when I give the cases in class, my undergraduate students do not converge on consensus, even after discussion), but we don’t know.

    Still, if Westerners are answering at chance, and Easterners are answering (consistently with the C-H view) significantly less often, then doesn’t it follow that “Westerners are more likely than East Asians to report intuitions that are consistent with the causal-historical view” which is, after all, a pretty weak empirical claim?

    One more point: Jason (and Josh earlier), rightly, raises the question about the probe using the "'who is John talking about', rather than what the referent of the name 'Godel' is in his mouth." He's right again. We asked the question the way that we did specifically so that it would be comprehensible to our subjects (who we thought might not understand the quasi technical term "reference"). Of course, it accentuates the difficulty of interpreting our findings as being about reference, and raises the question of the extent to we can get at the question we want with the methods we have. Does anyone have a better idea about how to probe questions of reference without using the term "reference"? I'd be interested to hear it.

    Reply
  35. Jason Stanley

    Ron,

    Ask about the truth of sentences in which the term occurs. So imagine a mathematician who only associates "Peano" with the information "the discoverer of axioms of arithmetic A1..A9" uttering the sentence "Peano was an Italian mathematician". If we think what he said was true, then "Peano" refers to Peano and not Dedekind.

    Reply
  36. Justin Sytsma

    Colin and Jason,

    I think that this is right, but that more needs to be said. As I find MMNS's scoring scheme confusing, let me rephrase their results for Westerners. They found that 56.5% of 31 Rutgers Undergraduates answered (B). (That number doesn't quite work out, as it suggests that 17.515 Undergrads answered (B). The reason for this is that each subject was actually given two probes (a Western name and a Chinese name version) and the percentage is aggregates across those two probes. As a result, I take it that on the 62 probes given, there were 35 answers of (B).) So, 56.5% (B) answers compared to 43.5% (A) answers (I'm not sure what more the standard deviation tells us here and think that it might just be a mistake which is neither here nor there).

    Given the low number of subjects, I doubt that the percentage they found is significantly different from chance. Even supposing that there are a few possibilities: (1) it might be that subjects are individually confused and just guessing (as Colin seems to suggest); (2) it might be that the question is ambiguous and subjects are split in their interpretation of the question (and not individually confused). I suspect that the latter is the case. As I suggested earlier, I think there is a perspectival ambiguity in the question with some subjects answering from John's perspective and others from their own perspective being privy to information in the story that John is ignorant of. In fact, Jonathan Livengood and I have run three experiments testing this (for those who are interested, we should have a draft of the paper reporting these results ready to post on the experimental philosophy blog, as well as my own blog (http://mymindis.wordpress.com) in the next week). To steal some of our own thunder, in our second experiment, we found that if we clarified the test question to emphasize the subject's perspective, native English speakers were fairly uniform in their answers. The text of that probe is the same as MMNS's, but the question reads as follows:

    "Having read the above story and accepting that it is true, when John uses the name 'Gödel', would you take him to actually be talking about: (A) the person who (unbeknownst to John) really discovered the incompleteness of arithmetic? Or (B) the person who is widely believed to have discovered the incompleteness of arithmetic, but actually got hold of the manuscript and claimed credit for the work?"

    The study was run online and we found that 73.8% of 84 non-philosophers answered (B); in comparison, 75.9% of 58 philosophers answered (B). In contrast, when subjects in our first experiment were instead asked who John thinks he is talking about, only 22.0% of 50 native English speaking undergraduates answered (B). I take this to suggest in favor of “confusion” (2) rather than (1).

    Reply
  37. Simon Rippon

    This isn't really my area, but I find myself very sympathetic to Jonathan and David's points here. I honestly can't see why we should accept Jason's claim that: "No conclusion whatsoever about the meaning of names in our language could follow from scenarios involving names in other languages". I understand Kripke's argument as expressible in the following way, in which it seems to do exactly what Jason says is impossible:
    —
    Thesis :(3) A name refers to the person who uniquely satisfies the description that users of the name have in mind when they use the name.

    Imagine a possible world where people use the name 'Ledög', having in mind a description ('the discoverer of the incompleteness of arithmetic') that is uniquely satisfied, in their world, by someone called 'Schmidt'. In that world, a man who we shall call 'G' murdered Schmidt and claimed credit for his discovery; these events were covered up and long forgotten, with the exception that G became known as the discoverer of the incompleteness of arithmetic. In all other respects their world is just like ours. Who does their later use of the name 'Ledög' refer to?
    If we answer that their name 'Ledög' refers to G, then it must be because we think that (3) is not true in their world. But there was nothing in the description to suggest that thesis 3 should hold in the hypothetical situation but not in the actual world. Therefore, our answer indicates our implicit rejection of (3) in the actual world.
    —
    What am I missing?

    Reply
  38. Anibal

    And testing for the intuitions about empty or fictional names (e.g. dragons,play characters etc.) as a case that lay in a grey-zone between causal theory of reference and descriptionists approach to reference.

    Reply
  39. Jeff Maynes

    As a graduate student working on the MNNS article and related topics, I am happy to see the topic discussed here, and to read all of the thoughtful comments. I do, however, have a question about one of Professor Stanley's initial comments. He wrote:

    "Mallon, Machery et. al. go on to argue against what they call “The Method of Cases” (basically appeals to speaker intuitions about truth-conditions) solely on the basis of their one experiment about “Godel”-“Schmidt”. This is an argument directed against the entire methodology of linguistic semantics."

    Professor Mallon has since claimed that the focus of their article really was on arguments from reference, and not the method of cases itself. My question, however, is whether the method of cases really is the methodology of linguistic semantics. It seems that the crucial bit of the method of cases, as formulated in "Against Arguments from Reference" is that the (paraphrase) "correct theory is the theory which is best supported by the intuitions of competent users." Not that intuitions might provide some evidence, but that you can read the correct theory off of intuition consistency. If the results of the first MNNS paper hold and generalize, then there are cases where intuitions will not consistently determine the correct theory, and we need to look elsewhere to adjudicate theory selection (and with regard to arguments from reference, MNNS hold that there is not sufficient evidence elsewhere).

    It is then a distinct, but important, question whether or not certain theories of reference are capably supported (besides appeals to intuition) – such as the case made in N&N by way of Kripke's modal arguments, etc. This would be a much more muted take on the conclusions, with the moral being that appeals to intuitions, as empirical claims, deserve empirical scrutiny rather than that the method of appealing to intuitions is hopelessly muddled. Is this a fair characterization of the MNNS argument, or are they making (or committed to) the further claim that appeals to intuition (and thus the methodology of linguistic semantics) are not sound methodology?

    Reply
  40. Max Deutsch

    Kirk says that he agrees with Josh that “given way the probe is presented in the Machery et. al. 2004 paper, there is only one right answer to the question who 'Gödel' refers to as used by John, the protagonist of the story.” This seems wrong to me. In Kripke’s supposedly subtler presentation, it is open, given K’s description of the case, that although he, K, is using ‘Godel’ to refer to Godel, “our ordinary man” uses it to refer to Schmidt. Why is it not likewise open that John, the protagonist in Machery et al.’s story, is using ‘Godel’ to refer to Schmidt even though Machery et al. are using ‘Godel’ to refer to Godel? Here is Machery et al.’s probe:

    "Suppose that John has learned in college that Gödel is the man who proved an important mathematical theorem, called the incompleteness of arithmetic. John is quite good at mathematics and he can give an accurate statement of the incompleteness theorem, which he attributes to Gödel as the discoverer. But this is the only thing that he has heard about Gödel. Now suppose that Gödel was not the author of this theorem. A man called “Schmidt” whose body was found in Vienna under mysterious circumstances many years ago, actually did the work in question. His friend Gödel somehow got hold of the manuscript and claimed credit for the work, which was thereafter attributed to Gödel. Thus he has been known as the man who proved the incompleteness of arithmetic. Most people who have heard the name ‘Gödel’ are like John; the claim that Gödel discovered the incompleteness theorem is the only thing they have ever heard about Gödel. When John uses the name ‘Gödel,’ is he talking about:

    (A) the person who really discovered the incompleteness of arithmetic?

    or

    (B) the person who got hold of the manuscript and claimed credit for the work?"

    Putting aside the very real worry about the semantic reference/pragmatic reference ambiguity, why, exactly, is (B) “the one right answer” to the question: What is the semantic reference of John’s uses of ‘Godel’? There’s a minor problem with saying in the first sentence of the probe that John “learned in college” that Godel is the proof discoverer, since later we’re told that Godel didn’t discover the proof. But I doubt this puzzled any of the subjects all that much. They’re likely not as sensitive to the fact that if S learned that p, then p, and probably read ‘learned’ as ‘was told [perhaps falsely] that…’

    For what it’s worth, my own view is that Kripke does not appeal to what Machery et al. call the “method of cases.” The Godel and Peano cases are simply counterexamples to a generalization entailed by descriptivism. Who intuits what is irrelevant. Sure, they’re intuitive counterexamples for at least some readers, but this is a logically inessential feature. In some understandably rare cases, a counterexample is counterintuitive. I think that if the East Asians in Machery et al.’s poll were reading the question as one about semantic reference (a big if) then they got it wrong: John’s uses of ‘Godel’ refer to Godel, the proof stealer

    Reply
  41. jonathan weinberg

    Ack, I've been scooped by Jeff Raynes and Simon Rippon! Well, let me build on their comments.

    Here's how I think both Simon and I were taught to play the game: we're trying to figure out the conditions for X (e.g., what makes something an instance of knowledge; what makes it the case that one event is a cause of another event; what makes it the case that a given proper name refers to the object that it does, if it does). We do this by deploying a variety of hypothetical scenarios, which can and often are totally fictional, and see how we are naturally inclined to sort them in terms of X. When some factor F can be varied wildly from case to case without affecting the X-ness value, then we take ourselves to have good reason to think that F is not generally part of the conditions for X; such are some of the classic arguments for internalism about justification, for example. When the X-ness value turns out to be highly sensitive to some factor G, then we still might need to discern whether that is in virtue of G's covarying with yet some other factor, but we might nonetheless take ourselves to have a good candidate for part of the conditions for X; such are some of the classic arguments for, oh, to pick an example totally at random, anti-intellectualism about knowledge. 😉 We're looking for what sorts of factors are X-variant or X-invariant across changes in possible worlds.

    In the Kripke literature, then, the idea is that we come to see that we can vary the associated-description factor wildly, and yet still the referents of the names do not generally switch. Presumably if we started changing around the causal/historical specifications, we would expect to be able to change the referents that way, which is part of what's going on in several of the examples that several people on this thread. For these purposes, there's nothing special here about the actual world, or the actual referents of any of our terms, except in the following way: if you vary a case from the actual world in ways that don't change the causal/historical specifications, then the proper name in question had better have the same referent in the hypothetical case as it does in the actual world. So it's not that, for Kripke's purposes, we need to only consider worlds in which "Goedel" means the same thing that it does in our world. Rather, for Kripke's purposes, _when and only when we keep the causal-historical facts the same as in the actual world_, "Goedel" had better turn out to mean the same thing that it does in our world. So it doesn't seem like this issue of real-world continuity is really much of a worry for MMNS's (or Livingston & Sytsma's) experimental design.

    I was very glad to see Jeff pointing out something that I think has made this whole discussion a bit noisier (in the as-opposed-to-signal sense) than it needs to be; namely, that MMNS just aren't taking as their target anything nearly as big as what Jason glosses it as. They have something much more specific in mind, which they explicitly define in their paper & which Jeff has picked up on. One way in which this disconnect has led to some distortion is that Jason's gloss does make it sound like they are indeed opposed to a vast swath of linguistics, and I'm pretty sure that that just isn't their aim. Another problem that Jason's over-broad gloss has created is that it makes some responses look plausible, when in fact they are just rather beside the point. For example, if we can (arguendo) read off of MMNS' subjects' responses, in particular that they don't find the scenarios nonsensical, that they don't take "the person who discovered the incompleteness theorems" as a synonym for "Goedel"… well, so what? Nothing about drawing _that_ kind of inference was at stake here. And, as I flagged in my first comment, the fact that Kripke may have yet other, non-intuition-deploying arguments is besides the point as well.

    This point also shows that something is amiss in Jason's argument here: "At most, even if we were to accept the experimental set-up of the original paper, what follows is one of the many instances of the "method of cases" fails to be convincing (a result made somewhat problematic and mysterious by the fact that the "Peano" case presumably is convincing). This shows nothing general about employing the method of cases in support of arguments from reference. For that, one would have to take on the whole battery of Kripke's empirical arguments." Taking the larger set of empirical arguments to be the target of MMNS is, I have been suggesting, a mistaken interpretation of their paper. I also find the "well, it's just one case" move to be an odd one coming from someone who is so friendly to linguistics, though — I take it that linguists normally take themselves to be able to make very strong inferences about languages from just a handful of isolated cases, because of the expectation of systematicity. What would be surprising indeed is if MMNS' two cases and _only_ those two cases had cultural variation. This surely isn't impossible — maybe some of the material in those cases was particularly culturally laden, in a way that has nothing to do with the right theories of reference for the different speech communities — but it should hardly be our initial hypothesis, in the face of evidence of differences. I would think the default rule here is: where you find one differences in bit of linguistic phenomena, you should suspect that there are many more such differences to be found.

    I would also like to pick up again with the particular nature of MMNS' argument & how in particular it turns on their taking themselves to have evidence for inter-group differences. The problem that they are raising is that there seem to be changes in intuitions across dimensions where we wouldn't expect changes in the relevant philosophical facts. If there's such a thing as _the_ nature of reference, and that nature is revealed to us by _the_ folk intuitions, then it's a deep problem if different folk turn out to have different intuitions. But it is true that one way to respond to such an argument is to let the philosophical facts change with the changes in intuitions– namely, by going relativist (as Janice Dowell endorses above). To the extent that we really are just doing descriptive linguistics, then that's probably fine; as I noted earlier, we expect lots & lots of linguistic facts to vary from language to language. But it does follow that as a research community, we've been making just terrible overgeneralizations about 'the' nature of reference, and have rendered the literature badly Anglo-centric or at least Indo-European-centric, and we'd best start commissioning some much more serious cross-cultural work to be done. Moreover, much of MMNS more recent paper (the one linked to in the main post) shows how deeply problematic that would be for a lot of the uses that other philosophers have wanted to put the theory of reference to, so the relativist move is not one that will save "arguments from reference" from MMNS' attacks. I suspect that most philosophers working on reference would not be so sanguine about going relativist here, but that, as they say, is an empirical question.

    Reply
  42. jonathan weinberg

    (Btw, why no comments on the syntax post? I have some thoughts on the matter, but I'm reticent to put them here, since this thread is already very busy & interesting.)

    Reply
  43. Jason Stanley

    Simon,

    The question you ask is underdescribed, so one shouldn't be able to respond clearly. You write "In all other respects their world is just like ours". Does everything that seems to be a proper name works exactly like our proper names? Or do you mean, in your description of "Ledog", that they have some descriptive names, and the rest of their names work like ours? This is not my understanding of Kripke's scenario, and there is no obvious answer to your hypothetical scenario.

    Basically I'm trying to explain the following point Josh makes:

    "What is needed is that the language of the fiction simply be our language – without that assumption, the fiction is argumentative inert with respect to Thesis 3, and Kripke's evaluation of the scenario is groundless (without continuity of language, the thesis is underdescribed – there are fictional scenarios like that described by Kripke in which "Godel" (in the language of the fictional community) picks out Godel, and others in which it picks out Schmidt, and others in which it picks out Dedekind, or anything else we like)."

    Reply
  44. jonathan weinberg

    (Sorry — Jeff _Maynes_, not Raynes. My bad.)

    Reply
  45. Eric Schliesser

    I am a bit puzzled by Jason Stanley's sanguine reaction to Colin Phillips' paper. Let me grant Jason that Phillips offers a very honest methodological self-assessment of the state of a field. But I read Phillips' argument a bit differently than Jason does. I think the nub of Phillips' argument is this: whatever crisis linguists face — and Phillips is quite frank (see below) — current experimental practice won't do much better than the armchair linguist to solve it. This should come as no surprise for anybody familiar with the role of experiment on humans in psychology, which has a terrific track-record in generating surprising data, but very bad record in building durable theory. (I have been puzzled that the experimental philosophy folk never seem worried about this inconvenient fact.) One needs the right kind of background theory as a condition for experimental work to become informative. Phillips is right to note that experiments (and fancy statistical techniques to analyze them with) won't solve the "crisis" in linguistic theories (11; see also the first sentence of the paper).
    A historian of science might not be surprised by the fact that a methodological article like the one by Phillips is written during a crisis. Phillips nicely describes a field where all the trappings of 'science' exist (journals, ongoing controversy within sub-fields, empirical evidence one way or another), but where (despite best efforts) there is no established (Kuhnean) paradigm. Just take a look at the following four quotes:
    [i] "A reasonable objection might be that it is misleading to focus on disputes among competing grammatical ‘frameworks’, since these are more likely to be driven by broader philosophical considerations (not to mention aesthetics and personalities)"(6).
    [ii] "But it may also be due in part to increasing loss of consensus on what the goals of linguistic theory are" (11).
    [iii] "Similarly, many linguists are happy to talk about grammar as a ‘computational system’ of the mind, but there is relatively little concern with the question of whether the proposed computations are actually carried out by humans" (13).
    [iv] "In sum, I agree with many of the critics cited above that some fundamental questions must be addressed (or readdressed) if generative linguistics is to again seize the initiative in the study of language. The perception on the outside that mainstream linguistics is becoming irrelevant is unfortunately very real indeed" (16).

    So, I find it a bit surprising that in his posts (here and elsewhere), Jason confidently appeals to the authoritative and impeccable scientific status of linguistic theory and practice.

    Reply
  46. Kirk Ludwig

    A quick reply to Max's question about why I think the way the scenario is presented in Machery et. al. 2004 and in Kripke's discussion are different. I think Max put his finger on the right point, namely, that in the probe we are told that John is told (I agree that 'learned' will be read as 'told') that Gödel discovered the incompleteness of arithmetic, and it is clear that he was told using 'Gödel', for this is required for us to think that he uses 'Gödel' and associates with it the relevant description. But in the probe that initial use of 'Gödel' is treated as having the same referent as subsequent uses in which we are told that he didn’t really discover the incompleteness of arithmetic. So what John was told (and came to believe) and so what he reports when he says that Gödel discovered the incompleteness of arithmetic has to be about Gödel and not Schmidt, if the scenario description is coherent.

    On a separate point, I think Jason and Josh are right that in the fictional situation we are imaging we are to be supposing that language works the same way as it actually does, for we are interested in our concept of a proper name, specifically with respect to what a proper name refers to as used by a speaker given various facts about the origin of his use of the name and what descriptions he associates with the name. This doesn’t entail that we must presuppose in describing the fictional scenario that the referent of 'Gödel' as used by our ordinary man is who we who use 'Gödel' to refer to in describing the situation. It is open, at this point in the investigation, where we have are still entertaining the description theory of reference fixing, that names can function the same way in the actual and fictional scenario while having different referents. For according to the description theory, the very same reference fixing description in different circumstances may secure a different referent for the name. According to the theory, that is what it is for the name to work in the same way in the two cases. If one wanted, one could say these are different languages because in the one circumstance the name refers to one thing and in another to something else, but holding fixed reference (and so keeping the languages the same in this sense) isn’t required to evaluate the claim that names get their referents for speakers using them by way of descriptions they associate with them. We precisely want to know whether, in a scenario in which linguistic practices are the same but there are differences that on the description theory entail a shift of referent, there is in fact a shift of referent.

    Reply
  47. Josh Dever

    What Kirk (the reading of whose very nice paper on these matters before I leapt into this discussion would have clarified the formulation of some of my earlier points) says above seems right to me – what is dialectically necessary is that "Gödel" in the fictional scenario has the same meaning as "Gödel" in (actual) English, and prior to settling the question, it is open for the descriptivist to hold that that continuity of meaning combines with the alteration of non-linguistic fact to effect a shift in reference.

    But I think there's a further worry, which is what leads me to hold that the scenario is presented in a question-begging (but in a good sense) way. The notion of meaning that needs to be preserved is too theoretical to be built successfully into a thought experiment – if Kripke had overtly asked that we hold fix the meaning, but not necessarily the reference, facts regarding the word "Gödel", then of course descriptivists would on theoretical grounds have held only the associated description fixed, and would have determined that the word "Gödel" as used in the scenario referred to Schmidt. There's no sufficiently robust pre-theoretic notion of meaning to be appealed to here, so as I read it, Kripke adopts a different strategy. He just uses our language throughout, so that there's no question about what is and is not to be held semantically constant (this aspect of the strategy is, of course, even more manifest in the non-fictional Peano case), and thereby simply makes it inevitable that "Gödel" refers to Gödel (rather than Schmidt). So, again, the actual argumentative force comes not from our making the right judgment about the referent of the word, but from our accepting the scenario as coherent. (The same effect could be achieved by saying "Look, suppose we were to discover that Gödel had not, in fact, proved the incompleteness of arithmetic. Now, the very fact that you can suppose this shows that descriptivism is false." This way of putting things makes the semantic argument, in this instance, very close to the epistemic argument, but that is to be expected given the complicated interaction between the modalized example and the non-modal Thesis 3 that the example is intended to test.)

    Reply
  48. Simon Rippon

    I agree (again) with what Jonathan Weinberg said in his penultimate comment: the point of a hypothetical example is to vary just the F (in this case: the individual satisfying the description) and see whether it affects the X (in this case: the referent of the name). Whatever else is varied in a hypothetical example is only what is necessary to make varying the F possible, or what results from varying the F.
    If Kripke's scenario was described, as Josh Dever has suggested, in a way that just guarantees that 'Gödel' (in the scenario) refers to Gödel, it would be useless for performing this sort of testing.

    Jason Stanley's reply to me: "The question you ask is underdescribed…", echoing Josh's comments earlier, fails (at least as far as I can see) to play by the commonly accepted rules of using hypothetical examples in philosophy. It's a bit like someone presented with a trolley problem: "Should we divert the trolley to kill the one and save the five?" answering: "How should I know? You haven't told me whether the members of the five are planning a terrorist attack, or have made a suicide pact, or whether the one was about to invent a cure for cancer!"
    (I am understanding Jason as intending to make a philosophical point about how Kripke's example has to work, rather than as just indicating that my presentation was rough.)

    If there *is* no clear intuition in the case of the 'Ledög' version of Kripke's argument, as Jason seems to think, (or at least if there's none in the case of a more fully and precisely specified version of the scenario that better draws out the parallel with Kripke's fictionalized Gödel character), it seems to me that that would only show that Kripke's argument is weak. People would seem to have been simply confused and taken in by the way in which Kripke's own presentation uses one an actual name and then confusingly mentions a non-actual name which is a homograph of it.

    Reply
  49. jonathan weinberg

    Kirk, I don't understand why you take yourself to be in agrement with Jason and Josh, and not rather in agreement with me, Simon, and others. That which you are saying is the dialectical situation at which the case is offered — "This doesn’t entail that we must presuppose in describing the fictional scenario that the referent of 'Gödel' as used by our ordinary man is who we who use 'Gödel' to refer to in describing the situation" — seems to me to be just what Jason and Josh are denying here.

    Reply
  50. Jason Stanley

    Jonathan,

    That's because I think you may be a bit confused about the dialectic (as is Simon, but in I think a different way). As far as I can see from your previous posts, you are not distinguishing *meaning* from *reference*. The *meaning* of the name in Kripke's fictional scenario must be the same as its actual meaning for the scenario to have any argumentative weight. But this doesn't mean that the *reference* has to be the same, for this would beg the question against the descriptivist. Kripke's scenario tests the descriptivist hypothesis, by asking us, if we fix the actual meaning, whether we get a shift in reference of the sort we would expect if the meaning of "Goedel" involved its reference being fixed by a description. Kripke concludes that if the meaning of "Goedel" is the same, then its reference must be as well, contra the descriptivist hypothesis.

    More precisely, since Kripke is directing his argument not against descriptivism, but against the view that the reference of a proper name is fixed by a description: what must be the same in the fictional scenario as the actual world are the facts that fix the reference of "Goedel" (Kirk states this clearly). If Kripke's opponent is right, that means that we should think that "Goedel" would have a different reference in the fictional scenario, since what fixes the reference of "Goedel" is a definite description. But Kripke concludes that if we fix the meaning (here construed as facts that fix the reference) of "Goedel", then even if Schmidt discovered the incompleteness theorems, that doesn't affect the reference of "Goedel", as it should if what actually fixes the reference of "Goedel" is that it is associated with the description "the person who discovered the imcompleteness theorems".

    Reply

Designed with WordPress