Justin Fisher

Justin Fisher (University of Arizona), Pragmatic Conceptual Analysis,” with commentary by Frank Jackson (Australian National University).  The paper and commentary can be found here.

Comments

[Justin Fisher]

I’d like to thank Frank for the interesting comments on my paper. I’m happy to hear that he shares so much of the spirit of my project. Here I’d like to respond to his concerns regarding the details. I see two general concerns in Frank’s comments. Let’s take them one at a time.

First, there is the concern that, since Pragmatic Conceptual Analysis concentrates on the pragmatic usefulness of concepts, it will fail to say appropriate things regarding concepts that aren’t especially useful. I discuss this issue briefly on page 5. There, I acknowledge that some concepts -- e.g., the concept of phlogiston -- may turn out not to be especially useful. Still, PCA helps to capture whatever limited usefulness such concepts have, and hence it offers a good charitable interpretation of their meanings. However, I also acknowledge that it may sometimes be reasonable for us to abandon such concepts, and to adopt more useful concepts -- e.g. the concepts employed by modern chemists -- in their place. So, I think I do allow for the possibility that some of our concepts are quite useless, and I think that PCA does handle such concepts in a reasonable way.

Second, Frank is concerned that I might be over-stepping my role as a philosopher in legislating how various empirical researchers, including field linguists, ought to think about the useful concepts that various people have. Frank imagines a version of my Huron example where field linguists do careful fieldwork to confirm that the Huron take it to be essential to the referent of their scurvy-concept that it is an ailment caused by enemy spell-casters. Frank thinks that, given these findings, the field linguists should (or at least could) conclude that the Huron suffer a particular sort of massive error: namely the Huron mistakenly apply their concept to lots of people, when in fact none of these people actually fall under their concept because none actually have an ailment caused by enemy spell-casters. Frank counts it against my view that I don’t think the field linguists should attribute this particular sort of error to the Huron.

I would, however, attribute a closely related error to the Huron: namely the Huron mistakenly believe that it is essential to scurvy that it is caused by enemy spell-casters. I think the Huron *correctly* apply their scurvy-concept to lots of cases of scurvy, as is evidenced by their fairly consistent success at curing this illness. The Huron’s main error is just that they have some false beliefs about the nature and origins of scurvy. (Incidentally, many European sailors made similar errors, taking scurvy essentially to be a contagious disease spreading through a ship’s crew, rather than a vitamin deficiency that displayed its symptoms more or less quickly in different sailors.)

So, which of these errors ought we to attribute to the Huron? Or, putting things more positively, what sort of charity ought we to show the Huron? Frank thinks we should be charitable to their semantic intuitions -- if the Huron say their concept means something essentially caused by enemy spell-casters, then Frank says we should believe them, regardless of how many mistaken applications that leads us to attribute to them. I think we should instead be charitable regarding the things the Huron regularly do successfully with their concept -- if they regularly use their concept to diagnose and treat cases of scurvy, then I say that what they have is a concept *of*scurvy*, regardless of how mistaken their rudimentary theory surrounding this concept might be.

In my bootstrapping argument, I point out two benefits of this pragmatically oriented brand of charity. First, my pragmatic analysis of the Huron’s concept captures the practical ideal that the Huron are approximating in their usage, and hence my analysis would do well to help guide their future applications of this concept. Second, my analysis captures what underlies the regular successes of their usage of this concept, and hence helps us to explain why success often comes about in cases where the Huron apply their concept in accordance with my analysis, and why failure often comes about when they fail to apply it in accord with my analysis. Jackson’s intuition-driven analysis yields neither of these benefits.

I think that, as a matter of empirical fact, our attributions of concept-meaning (i.e., our attempts at conceptual analysis) often deliver the application-guiding and explanatory benefits just described, and that it is their capacity to do this work that makes such attributions worthwhile. Insofar as we want our attributions of concept-meaning to continue to do work like this, we have reason to favor my proposed brand of charity over Frank’s. And insofar as field linguists want their attributions of concept-meaning to do this work too, they’ll need to be charitable in the way I suggest too.

[uriah]

Very cool paper, Justin. As you know, this is not my cup of tea - but I can appreciate the way everything hangs together, especially knowing your independent views on teleosemantics etc.

Just to stir trouble in your cup, here are some half-baked critical thoughts.

First, the first phase of the argument only works if Intuitive Conceptual Analysis can't solve the problem raised by Jackson otherwise than how Jackson solves it. I think it can, by saying that incompatibilism is not a fully platitudinous claim about free action, only a “highly” platitudinous one. (An issue arises here regarding the notion of degree of platitude. But if the issue can be resolved satisfactorily, we I think it can, then this should give the Intuitive Conceptual Analyst a non-Jacksonian way out.)

Second, the second phase of the argument has a fishy air of circularity about it. To some extent, you concede that by saying that it makes PCA “self-ratifying.”

Third, you talk about our concept of concept-meaning. On first glance at least, it's not clear that we the folk have a concept of concept-meaning. It's a technical notion, so not one that can be gotten at by conceptual analysis.

[Justin Fisher]

Thanks, Uriah, for stirring trouble in my teapot. Let me see if I can calm things down.

First: I think you're right that there's room for some fans of Intuitive Conceptual Analysis to part ways with Jackson's incompatibilism. For many people do have some compatibilist intuitions, and some versions of Intuitive Conceptual Analysis might allow these to trump Jackson's incompatibilist intuitions. However, I think my more general point still holds. Intuitive Conceptual Analysis can't guarantee that whichever intuitions end up counting as top dog will be ones that track what's useful. So even if you think the top dog intuitions regarding free action actually do happen to get things right, there's no guarantee that *all* top dog intuitions will get so lucky. Even if some top dog intuitions do do fairly well, when push comes to shove, we'll do better to revise our concept-usage in light of strong empirical evidence about what actually would be useful, rather than on the basis of intuitions about how our concepts ought to be used.

Second: I am proud that my notion of pragmatic meaning is self-ratifying in the sense that it says that the meaning of 'meaning' is pragmatic meaning. If a theory of meaning didn't ratify itself in this sense, then, by its own lights, it should be rejected -- with theories like these, who needs enemies? I agree though, that it would be fishy to accept a theory of meaning *simply* because it is self-ratifying. In addition, you'd need a premise saying that that theory is normatively authoritative -- i.e., that we generally ought to accept whatever it ratifies. I get this premise from Stage 1 of my bootstrapping argument, which proceeds independently of all the stuff about self-ratification, and hence is innocent (I hope) of any charges of circularity.

Third: I agree that ordinary folk don't run around saying the word "concept-meaning", but I still think that there is a folk concept of concept-meaning. I think this concept is pretty closely tied to our folk concept of *word*-meaning. However, the folk do recognize that words don't come in a perfect one-to-one correspondence with concepts. E.g., we recognize that the word 'parent' is used sometimes to express a concept of something like a biological parent and sometimes to express a concept of something like an official guardian. We can ask questions about the correct application conditions for these two concepts. E.g., you might ask, "When we're thinking about parents as guardians, does the long-term lesbian partner of a kid's biological mother count as a parent or not?" When we ask questions like this, I take it that we are tacitly employing our folk concept of concept-meaning.

[Richard]

Justin, I have a clarificatory question: do you mean for your pragmatic interpretation to override not just general beliefs about what's essential to the concept, but also the reflective judgments they would make in response to particular cases?

Suppose we ask the Huron to consider the epistemic possibility that the cases to which they apply their scurvy-concept ("scurvy-H") are actually caused by vitamin deficiencies rather than enemy spellcasters. We can highlight how they're still tracking the "useful" features of the situation, and all that. Most likely they would respond by recognizing that spellcasting isn't essential to their concept after all. (I'd expect them to simply want a concept of "that which fills the scurvy role".) But suppose they didn't. Suppose that the Huron unanimously insist that, on the hypothesis that the given scenario is actual, the cases are not instances of scurvy-H. They insist that if it turns out they're mistaken and the world actually has no magic, then it also turns out the world has no scurvy-H.

Are you advocating such a strong kind of "linguistic paternalism" that you would override their claims even in this case?

[Justin Fisher]

Richard asks whether I'm so much of a 'linguistic paternalist' as to be willing to override even the empirically informed reflective judgments of people regarding what falls under their concepts.

My answer is, yes, I'm willing to be 'linguistically paternalistic', but I feel fairly optimistic that this wouldn't have to happen very often.

Yes, insofar as 'reflective judgment' sometimes enshrines intuitive superstition[1] over empirical evidence regarding the useful work that a concept has been doing, then I think such 'reflective judgments' really should be overriden. However I'm optimistic that reflection is a cure for superstition, and hence that, if the judgments are empirically well-informed and reflective enough (including perhaps reflection on arguments like those in my paper), then these judgments won't need to be overriden.

[1]It might be a bit strong to use the label 'superstition' for all firmly held but contra-useful folk intuitions. But hopefully this phrasing still gives a good sense of the general shape of the line I'd like to take. Feel free to substitute more politically correct language if you like.

[Justin Fisher]

One more thought on this topic:

What Richard calls 'paternalism' is actually the norm in science. Many people have 'reflective judgments' that they don't have AIDS, or that they don't harbor resentment for their first spouse, or that unstructured interviews help them to make better hiring decisions. In many cases like these, good science rightly dismisses people's reflective judgments. It should be quite surprising if people had a special sort of infallibility about the meanings of their concepts, when their reflective judgments about virtually all other aspects of themselves are clearly fallible.

So, I think we should antecedently *expect* a theory of concept-meaning to be 'paternalistic' in Richard's sense, and the burden should be on 'non-paternalistic' theories to justify their extreme confidence in people's 'reflective judgments' about concept-meaning.

(Perhaps the best attempt to defend such a 'non-paternalistic' theory is by Dave Chalmers, but I think Dave is extremely confident only that people would reach the right conclusions about meaning, and about everything else too, after *extremely*idealized* rational reflection on the basis of *exhaustive* empirical evidence. It's open to the defenders of other theories of meaning, including mine, to say that extremely idealized rational reflection would lead people to recognize the truth of these other theories, and hence that Chalmers isn't really disagreeing with them.)

[Terrance Tomkow]

It probably dates me but, I was surprised to find no footnote in your paper to the man who wrote the book on identifying the concepts of foreign tribes: Quine in "Word and Object".

My questions are: Do you see your enterprise as continuous with the one Quine described as Radical Translation? Do you think that your proposed method is relevant to the Indeterminacy of Translation: i.e. Quine's argument that, ultimately, there may be no fact of the matter about which concepts the natives have?

[Justin Fisher]

Terrance notes that my project is similar to Quine's work on radical translation, especially in that we both tackle the question of how a field linguist might attribute meaning to the thoughts and utterances of natives who speak another language. I see three important differences between my work and Quine's.

First, Quine works within a behaviorist paradigm that ties the would-be translator's hands behind his back: "All the objective data he has to go on are the forces that he sees impinging on the native's surfaces and the observable behavior, vocal and otherwise, of the native" (W&O, pg 28). In contrast, I imagine the would-be translator as having access to much more data, including information that bears on the shape of the internal cognitive processes that the native subjects are going through, and information about what has underlain many past successful instances of these cognitive processes.

Second, Quine does not have clear criteria for what would count as successful translation. In contrast, in the second stage of my bootstrapping argument, I attend carefully to what work we might reasonably demand that successful attributions of concept-meaning be capable of doing: namely helping to guide future application of these concepts, and helping to explain many cases of behavioral success (and failure) in terms of correct application (and incorrect application) of concepts.

Third, as a consequence of these first two differences, Quine's conclusion is much more pessimistic than mine. Quine thinks a translator with his hands tied behind his back, and with murky criteria for success, would be bound to fail. In contrast, I think that a translator with much more information available, and with fairly well-defined criteria for success, might likely succeed.

This is not to say that we will always be able to attribute perfectly determinate application conditions to every concept. We often will be missing relevant evidence. And even when we have all the relevant evidence, it may be that existing patterns of successful usage would be equally well (and equally poorly) sustained by various different proposed explications. But I think we should expect a limited dose of indeterminacy out of any information-tracking biological system, so I don't think there's anything wrong with the limited amount of indeterminacy that my approach attributes to our concepts.

[Terrance Tomkow]

Well I can see how you might say this on the basis of chapter 2 of  Word and  Object, but Quine went on [JofP 67,1970], to give a quite different argument for indeterminacy that doesn't rest on behaviorism or pessimism but on the collapse of positivism. It seems to me to bear directly on your program so allow me to briefly reprise it.

Verificationism might be expressed as the doctrine that if you and I have contradictory opinions about anything then either there is some empirical evidence that would resolve the issue between us-- proving one of us wrong-- or the disagreement between us must be purely verbal (because meaning is exhausted by empirical content).

Now, for reasons we need not rehearse here, we don't believe in verificationism any more. But in rejecting it we are correspondingly committed to saying it *is* possible that you and I have contradictory  theories (a genuine contradiction not a matter of equivocation)  and yet there be no *possible empirical evidence* that would confirm/ falsify your theory that would not likewise confirm/falsify mine.  Our theories may be empirically equivalent yet logically incompatible because empirical content *doesn't* exhaust meaning.

There is, moreover, nothing to say that such theories may not be as comprehensive and  empricially successful as one might hope: they may both be continue to be confirmed forever more.

Now this seems a direct consequence of the rejection of verificationism. But it is a disturbing result nevertheless. If our theories contradict one another then it seems that at least one them must be false even though empirical methods might *forevermore* be unable to tell us who is right about the facts.  This upshot led some post-Quineans to become anti-realists ("so much the worse for 'the facts!') and drove poor Richard Rorty to comparative literature.

Unsettling as this might be, Quine argued that it has even more disturbing upshots when we think about the enterprise of translation; the enterprise, that is, of determining what theories and concepts other people hold.

Thus, suppose you and I hold contrary but empirically equivalent theories of the world and assume we (somehow) know this about each other. Though our disputes are empirically irresoluble we may at least agree that we disagree about the facts. But now we come across a tribe of natives who appear to be operating a theory which is every bit as empirically adequate as yours or mine (and suppose it is).  Do I impute to them my (I think) true theory of the world or your (I think) false one? Let my  theory be as powerful and  as comprehensive as you like-- let it be a Physicist's Grand Theory of Everything-- still, Quine argued, it won't settle the the matter of what theory the natives *really* hold. The indeterminacy of translation he concluded was "separate and additional" to the underdetermination of theory by evidence. There is nothing for the alternative translations to be right or wrong about.

Quine took this to show that talk of "meanings" and "concepts" ultimately has no place in serious science (in which he included, of course, philosophy).

The relevance of this to your account seems to me pretty direct: surely theories that are empirically equivalent are, in the end, going to be pragmatically equivalent by all your measures (and so much the worse if they aren't).  And if that is so, then it seems to me that you are eventually going to have to confront Quine's deep doubts.

[Justin Fisher]

Terrance rightly notes that there are multiple strands in Quine's arguments involving radical translation. I can't hope to fully address all these interesting issues here. Instead, I will focus on just a few of the links between my work and Quinean considerations.

First, I want to note that my view is not verificationist. For example, I think there is a fact of the matter about whether there are intelligent beings so small that they will be forever undetectable to us. How does this fit with Pragmatic Conceptual Anlaysis? Well, PCA says that our past successful uses of a concept of intelligence lay down criteria regarding what something would need to be like to count as being intelligent. These criteria will capture what's explanatorily relevant to the past success of our practice of sometimes attributing intelligence and sometimes not. Presumably, this has something to do with the presence and absence of various information-processing abilities and nothing to do with size. Hence, it might be that there is something very small (even too small ever to detect) which has the relevant information-processing capabilities, and hence counts as intelligent. Or maybe there are no such things. We'll never know for sure.

Second, PCA draws heavily upon empirical hypotheses, including (as we just saw) hypotheses about what is explanatorily relevant to the success of our past use of our concepts. Insofar as there are deep Quinean worries about the underdetermination of empirical theories by empirical data, there will be deep Quinean worries about our capacity to get the empirical hypotheses that we'd need to employ PCA.

Fortunately, third, the really important issue here is not whether the empirical data are *logically*compatible* with multiple theories, but instead whether the data make it *epistemically*rational* to believe various theories. The more epistemic principles there are that allow non-demonstrative (ampliative) reasoning, the more tools I will have to whittle away a lot of indeterminacy. (This is related to Larry Laudan's critique of Quinean underdetermination.)

Fourth, and finally, questions about whether we can ever find a successful account of concept-meaning (or of translation) will depend upon what we set as the goals of this account. I honestly don't know what Quine thinks a successful account of concept-meaning would need to do, but he clearly has the bar set extremely high. In my paper, I suggest that we should seek an account of concept-meaning that will do certain application-guiding and explanatory work, and I argue that if we find such an account, we'll have good reason to call it an account of concept-meaning. As noted above, Quine's general worries about underdetermination of theory by data could threaten this project, but won't if there are enough ampliative epistemic principles. But his "separate and additional" concerns about meaning probably don't apply to my project, because his concerns arise regarding the peculiar and probably unrealistic goals he has in mind, not the much more managable goal that I set for myself.