I wanted to say a little more about philosophical methodology and naturalistic challenges to a priori investigation. A key point of dispute between, let's call them, naturalists and rationalists, seems to involve the proper role of intuitions in philosophical methodology.
Again, I'll start by describing a hypothetical disagreement, running the risk of caricaturing the arguments in question. If that's what I'm doing, it will be instructive to see how the real views are different from my presentations of them here.
Archie: Consider twin earth, which is kind of like regular earth, but where there's no H20. Instead, there's this other stuff XYZ, which is not H20, but is superficially very similar to H20. It's clear and quenches thirst, etc. Introspecting, we discover in ourselves the rational intuition that XYZ is not water. Intuitions are good evidence; therefore, XYZ is not water. Therefore, externalism about content is true.
Eddie: It's not rational to blindly treat your so-called "rational intuitions" as evidence. We can't calibrate our intuitions to check their reliability, and it's totally mysterious how it is that they could be reliable. We shouldn't just trust them. We should go administer surveys instead. If we want to know whether XYZ is water, we have to go see how the folk use the word 'water'.
In my last dialogue, I claimed there was no real disagreement, even though they clearly thought there was. This time, I think they're both agreeing to something silly. According to traditional methodology, our intuitions are supposed to be evidence? Why? I think the most attractive position is this one: Experimental Eddie is right that we shouldn't treat our intuitions as evidence, but both he and Armchair Archie are wrong that traditional methodology commits to intuitions as evidence.
When I judge that that stuff on twin earth is not water, how shall we understand the reasoning I run through? Not, I suggest, like Archie said:
(1) I have the intuition that XYZ isn't water. (introspection)
(2) So, XYZ isn't water. (1, principle about intuitions)
(I'm omitting (3) So, content externalism.) Instead, we can understand the reasoning as not involving any premise about intuitions, or based on introspection, at all:
(3) All and only H20 is water.
(4) XYZ isn't H20.
(5) So, XYZ isn't water.
This argument is valid and plausibly represents the reasoning we go through to ourselves. And no premise is based on introspection or a claim about an 'intuition'.
Someone will say: "but the only way you could know (3) is by intuition." Since I don't know what intuitions are, I won't try to evaluate that claim; the point is that (3) is something that everybody should agree we can know, and that therefore its invocation in this judgment about a thought experiment shouldn't be problematic. (Do you doubt that we know (3)? That's to doubt whether we can recognize the possibility of fake water.)
This argument, of course, is a posteriori, because premise (3) is. But there's an a priori version of it, too:
(3') All and only things that are F are water, where 'F' stands for whatever it is that it turns out makes up water.
(4') XYZ is not made up of F.
(5) So, XYZ isn't water.
Whenever possible, we should avoid putting argument in terms of 'intuitions'. I suggest that this is almost always the case.
Thursday, February 08, 2007
Why Do People Worry about Whether Intuitions Are Evidence? (by guest blogger Jonathan Ichikawa)
Posted by Eric Schwitzgebel at 8:30 AM
Labels: metaphilosophy
Subscribe to:
Post Comments (Atom)
21 comments:
Hi Jonathan,
the point of your a priori version of the argument seems to be that it is plausibly a priori that the essence of water is what makes it up. Let's grant that. But this, of course, is an ontological claim about the individuation conditions of water. How do you want to make the transition from such an ontological claim to a claim about the semantics of the term/concept "water", i.e. to content externalism?
Maybe in the following way: What makes up something is not alway obvious. It's not obvious what makes up water. If it's not obvious what makes up water, then the semantics of our term/concept "water" can't be exhausted by what is easily cognitively accessible, i.e. by an association of the obvious or surface properties of water with "water". Therefore, content externalism is true of "water".
If this is roughly the right way to infer the semantics from the metaphysics of water, then I guess my worry is that such a line of reasoning would, at least in part, be based on experience and/or semantic intuitions (e.g. the premise that it is not obvious what makes up water).
(3) All and only H20 is water.
I wonder what explains the fact that “(3) is something that everybody should agree we can know”? The argument is meant to preclude intuitions from being the source of this knowledge. I assume that the tacit explanation for this knowledge, then, is empirical inquiry; following Kripke, science tells us that (3).
How can science tell us that (3)? Let’s construe science as a purely inductive enterprise: so that repeated and careful observations of water will reveal that it is, in fact, H2O. On this model of inquiry, inquirers are given a minimal psychology—we have a generalizing ability that allows us to see what is in common with a number of samples.
This minimal psychology is necessary so that we can catalogue regularities in the world with minimal interference: it is almost exclusively the world that tells us that water is H2O—inquirers simply and mechanically correlate what a number of particulars have in common.
The worry I have is that this minimal psychology—our generalizing ability—isn’t, in fact, minimal, simple, or mechanical. Everything is similar to everything else. Any generalization requires antecedent criteria for distinguishing interesting similarities from non-interesting similarities. Moreover, many of our generalizations have a stipulative element. For example, does “motion sickness” entail that sickness is caused by motion, or would a more austere science drive us to see motion as causes of a syndrome which is, properly speaking, motion sickness (so that it would be possible to have motion sickness without the body’s being in motion)? I don’t think there is an answer to this question. Once the causes, categorical properties, and effects of motion sickness are known, there is no a priori requirement which dictates which of these is properly denoted by the term “motion sickness”. (The philosopher of science could, of course, impose one). The referent of “motion sickness,” it seems to me, is dictated in part by a series of pragmatic factors.
The point is that if our generalizations are governed, first, by our empirical, repeated observation of particulars, and, second, non-empirical selectional and pragmatic criteria then it is not clear to me that (3) is something that everyone should agree that we can know (simply) by observing a lot of stuff. The appeal to intuitions is a (clunky) way of excavating some of the non-empirical factors which are a condition for our inductive practices.
I think this point is in line with Joachim's worries.
I suspect this is in line with both Joachim's and Joshua’s note, but I’m not sure: I would think that semantic intuitions about ‘water’ are the extent to what is going on with this thought experiment – in contrast, perhaps, to the idea that we begin with a premise like (3) and follow a syllogism. For a scientist, and perhaps for a language using community that takes science as important in their daily conceptions of the world, the composition of ‘water’ would be found in the meaning of ‘water.’ We use ‘water’ to worry ourselves about the building blocks of the earth; terms like ‘water’ would reference composition. But it seems to me that this is not necessarily content externalism, just a different way to use ‘water.’ I’m afraid I would call the ocean ‘water’ even if I found out that saltiness was due to XYZ and not salt and H2O, while leaving H2O water a technical term for the scientists. This would ground a semantic intuition that is complete, without need of discovering composition. However, with enough surprises over composition, it could still be asked how usefully my concept—what I want ‘water’ to capture—maps onto anything in the world.
Jonathan,
(i) I'm not sure that you've picked an optimally illustrative example of an intuition for your discussion; it is, after all, a point of some debate still just what sort of _a priori_ component is or isn't needed for these kinds of scientific reductions. Moreover, the claim that water = H2O is not a claim that is meant to do any work in selecting between philosophical theories. (The issue with Twin Earth is not whether water = H2O, but how to characterize the content of me and my twin's mental states. I take it that all sides on the wide content/narrow content debate agree with the identity itself.)
(ii) So, what would you say about, e.g., the claim that Mr. Truetemp doesn't count as knowing the temperature where he's at? When the premise is stated baldly, just as the claim itself, then it looks terribly question-begging against the reliabilist. And if we don't want to just present it as the claim itself, it's had to see what evidence we can give for the claim. That's why it gets presented as an intuition.
(iii) Just in case you weren't already aware: Williamson wrestles interestingly with a lot of these questions in his "Philosophical 'Intuitions' and Skepticism about Judgment".
Hi everyone, thanks for the comments, and sorry for the slow response.
Joachim, I'm not sure if I'm quite seeing the question. You say:
But [the essence of water is what makes it up], of course, is an ontological claim about the individuation conditions of water. How do you want to make the transition from such an ontological claim to a claim about the semantics of the term/concept "water"...?
I think that the relevant ontology and the relevant semantics are very closely related. In particular, the semantics determines whether the term/concept picks out superficial properties or essences; so, that it picks out essences instead of superficial properties entails something about the semantics.
But I'm not sure I'm quite getting the point, so this response is probably unsatisfactory. Can you help me see it again?
Joshua:
I wonder what explains the fact that “(3) is something that everybody should agree we can know”? The argument is meant to preclude intuitions from being the source of this knowledge.
I don't take myself to have offered an explanation of that fact. I just took it for granted, considering this to be reasonable since I assumed that everyone agreed with it.
I'm not sure what I want to say if you want to hear an explanation for it. I definitely agree that we don't know (3) by induction.
There are general Kripkensteinean worries about skepticism about meaning. I don't know how to respond to them, but I'm not particularly embarrassed by this fact. Somehow, we manage to know what our concepts mean.
I guess I'm happy to leave open the possibility that something like intuitions will have some role to play in the correct epistemology of meaning. I tend to doubt it, but if someone offers a story based on intuitions, I'll listen to it and evaluate it.
Michael:
I'm not sure if I understand. Are you saying that, if confronted with surprising discoveries, you'd use 'water' for watery stuff and 'H20' for water, and deny that all and only water is H20? That sounds somewhat odd to me, but if that's what you're saying, I'll assume for the purpose of argument that it's true.
There are two ways I can interpret this counterfactual, then. I might say that some surprising empirical discoveries would cause you to change concepts (or which words pick out which concepts). Or I might say that you after all had a different concept picked out by 'water' than the rest of us did all along. Neither option sounds terrible to me.
As for this suggestion:
I would think that semantic intuitions about ‘water’ are the extent to what is going on with this thought experiment
Here's the question. What work does the invocation of intuitions do? I definitely think this is a semantic matter. If you'd be just as happy with "semantic beliefs" or "semantic facts" as with "semantic intuitions", then maybe we have no quarrel. The things that I want out of my philosophical arguments are these introspected seemings. If your intuitions are something else, then I don't know that I mind them.
Jonathan:
On your (i): I might just be insufficiently informed about the relevant debates here. That's probably likely. Maybe you can help me out. You say:
The issue with Twin Earth is not whether water = H2O, but how to characterize the content of me and my twin's mental states.
I'm having a hard time seeing the difference. I took the argument to go like this: You have a belief about water, and he has a belief about XYZ, and XYZ is never water, so you have different beliefs. Another way to see it: they have different truth-conditions.
(I do believe in narrow content, if that helps; I agree that there is something importantly in common between your belief and your twin's.)
On (ii): I agree with the criticism that there is something fishy going on with Mr. Truetemp. I just don't see how intuition is supposed to help. You say: And if we don't want to just present it as the claim itself, it's had to see what evidence we can give for the claim. That's why it gets presented as an intuition. And that last move baffles me. What good does presenting it as an intuition do? Here is, as you say, a question-begging argument:
(1) Truetemp doesn't know.
(2) So, reliabilism is false.
I suggest that no progress at all is made if we move to this argument:
(1) I have the intuition that Truetemp doesn't know.
(2) So, Truetemp doesn't know.
(3) So, reliabilism is false.
That argument, I think, is exactly as good as this one:
(1) I have the belief that Truetemp doesn't know.
(2) So, Truetemp doesn't know.
(3) So, reliabilism is false.
Nobody argues like this; why should anyone offer the intuition version?
This isn't to solve any of the challenging methodological questions in the neighborhood. My point is just that intuitions shouldn't be the issue.
(Very briefly, my thoughts about the case: there is room to talk about why we should accept the premise that Truetemp doesn't know. The reliabilist can, should, and does emphasize parallels with ordinary perception. There is tension, he suggests, in believing in visual warrant but not temperature warrant for Truetemp; one might even formulate principles and make arguments. The anti-reliabilist might very well realize that he is after all insufficiently sure about Truetemp for that premise to overthrow reliabilism. So my considered view on this case is: spoils to the victor.)
On (iii):
Yes, I know that Williamson paper; it's been a couple of years since I've read it, but I take myself to be in pretty close to full agreement with him about this issue.
Hi Jonathan,
as to your response to Jonathan Weinberg:
Compare the following argument:
(1) I have a visual perception that there is a dog.
(2) So, there is a dog.
With:
(1) I have the belief that there is a dog.
(2) So, there is a dog.
Given your own examples one ought to conclude that the second is "exactly as good" as the first - but that seems wrong, since the first is clearly better. And that's why, I guess, defenders of intuitions tend to stress the analogy between intellectual and perceptual seemings/evidence.
As to your response to my own comment:
For not "quite seeing the question", as you say, you've answered it surprisingly frankly and clear! Namely, by claiming that you want to make the transition from ontology to semantics based on the following principle:
(P) "the relevant ontology and the relevant semantics are very closely related"
You continue that part of this "close relation" for you is: "that it [the term/concept water] picks out essences instead of superficial properties entails something about the semantics". That may be so in some very general sense. But your intended conclusion was not just that the ontology of water entails "something" about the semantics of water, but that it entails content externalism. But here, I guess, a contemporary descriptivist like Jackson would surely protest. For he would surely except that water has the essence you claim, yet still deny externalism. So, something has to be wrong with your argument - at least in the sense that it has far more hidden commitments as you seem to admit. Or, to put it in another way, even if the ontology of water has "some" semantic entailments, they surely aren't substantial enough to decide the issue between semantic internalists and externalists (because they don't disagree about the ontology of water at all!).
Jonathon,
Thanks for the reply. I’m appreciating your point more. I might have been wrong using the phrase “semantic intuition,” particularly in the context of your argument; but I wonder if there is still motivation to do so: On my view, the way the question of “what is water?” is posed makes it a different question from “what do we mean by water?” It seems the fact that the semantics of ‘water’ is grounded in stuff that is never just H20 shows that ‘water’ was never in search of an anchor in a chemical compound. But as for what we mean by ‘water’:
I’m with Eddy in that I would want to know how we use the word ‘water.’ Yet, I didn’t envision any empirical experimentation or surveys; nor was I inclined to introspect as I used the word ‘water.’ This is where the word ‘intuition’ came to me I think; it is this singular “hey, that is not what I mean by ‘water’” that moves me on in search of cases of word use in support of this intuition, all the while hoping my interlocutor will have the same intuition as long as I phrase the problem right. This is not to say that we cannot change or add to our word usage: this is what it seems we have done with ‘water.’ ‘Water’ qua H20 seems to be a technical term.
As for evidence, some initial thoughts: I believe that ‘water’ does not mean H20 for common folk (even if it “is” H20). This belief operates as evidence at least for me in virtue of it simply being a belief of mine; this belief is firmer evidence for me if I not only just believe it, but take this belief to be rooted in this deep ‘intuition,’ this perceptive contact I have with my more knowledgeable unconscious mind. Since I take it to be an intuition, I argue in such a way as to appeal to the same intuition I expect other philosophers to have access to.
But perhaps my worries about your H20 example were a bit beside your more general point, and this is perhaps the first time I’ve thought much about ‘intuition.’
Joachim -- thanks for the response. One thing this thread is making me realize is that there're a lot of moves in the content internalism/externalism debate that I'm just not aware of. (I have to admit: I didn't even know that Frank Jackson was an internalist.) So I wholeheartedly agree with whomever it was who said up above that I chose a non-ideal example. I'm trying to discuss methodology, not externalism. I'd have to go look at what people like Jackson are actually saying in order to answer how my argument would engage with it.
Maybe I should have done the whole thing in knowledge-JTB terms again. I had originally meant to, then decided some variety would be nice. Now I regret it...
Michael:
I'm suspicious of this distinction:
On my view, the way the question of “what is water?” is posed makes it a different question from “what do we mean by water?”
This, I think, is pretty close to the bad disagreement between Archie and Eddie in my first post. The problem arises from this tautology: "water" means water. There can be nothing such that we mean it by "water" and it's not water.
Jonathan,
Thanks for the second reply!
“There can be nothing such that we mean it by "water" and it's not water.”
I think I agree. It is “the way” the question is posed that worries me: “What is . . . ?” The search for a chemical compound that ends with the conclusion “’water’ = H20” doesn’t seem right to me unless this is a new technical term within the language game of science. I don’t mean H20 when I say ‘water,’ although I might very well conclude that “water is H20” when the conversational context is a quest for chemical composition. Is this similar to Wittgenstein’s worry about “What is . . . ?” ?: “As the problem is put, it seems that there is something wrong with the ordinary use of the word ‘knowledge’” (“St. Augustine’s Puzzle about Time,” Blue Book). My worry is the same about what we do with ‘knowledge’. Hope that answers your suspicion at least a little bit! . . .
Jonathan,
I'm also primarily interested in methodological issues, so if it did appear otherwise, I would regret that, too. But still there's a larger issue looming here, namely that it is an open question if such a nice separation of methodological from non-methodological questions is really possible in philosophy (or anywhere). I hope so, since I'm working on methodology myself, but our discussion raises some doubts.
Since you claim that changing the example back to knowledge and JTB would help - how exactly would it help?
Joachim:
Well, it'd be something like this. Here's a bad Gettier argument:
(1) Gettier case is possible.
(2) I have the intuition that Gettier case is non-knowledge JTB.
(3) So, NKJTB is possible.
I much prefer this one:
(1) Gettier case is possible.
(2) Gettier case is an instance of NKJTB.
(3) So, NKJTB is possible.
I don't see what role 'intuitions' are supposed to play.
Briefest possible reply: Intuitions are supposed to be evidence for (2). What other evidence do we have for this premise?
I don't know. Do you have a theory of evidence in mind? What would count as evidence? (Why do intuitions count as evidence?) Do we need evidence for every premise in every argument? What counts as evidence for having the intuition? What counts as evidence for 1+1=2?
I don't mean these questions rhetorically -- I just don't know what evidence is or why it's important; if you help me clarify what you mean, I'll say what I think the evidential status of that premise is.
But it seems to be a legitimate demand to ask how we come to know (or how we come to acquire justification for) premise (2). So, my talk of "evidence" was supposed to capture something like the following rational demand: What reason do we have to accept premise (2)? Do we have any reason at all? If not, why continue to accept it instead of withholding belief? Maybe it's even fine to accept premise (2) without any "evidence" as long as it is uncontroversial. But it is not. Some people just don't judge that Gettier cases are instances of NKJTB - think about the results of experimental philosophy (Nichols/Weinberg/Stich). So, to give my worry a name: Your way of framing the issue makes premise (2) appear to be simply dogmatic.
(The intuitionist would answer my rational demand like this: Intuition is a basic source of evidence/justification, just like perception or introspection. We need such a source to explain our justification for beliefs like "1+1=2". Since it is a basic source as well as a plausible source, there's neither a threat of regress nor of dogmatism. - Surely this answer is no news to you, but what you said in your postings doesn't convince me that your own picture is superior to this more traditional one. Frankly speaking, I don't even see what your alternative picture may look like.)
Well, sure, my argument won't be a good argument to use against somebody who's describing Gettier cases as cases of knowledge. That's not the audience I have in mind. Think about what Gettier actually did -- he said, "hey, group of philosophers, K isn't JTB!" See, here is a valid argument! And then he gave my argument. And pretty much everybody agreed that he was right.
If you're talking to somebody who says of the Gettier case that it's knowledge, then obviously this argument is going to be pointlessly question-begging.
What do you do with those people? Well, there are three choices: they're confused, they're using a different concept, or I'm wrong about what my concept means. My suggestion is just that skepticism about meaning is unmotivated here. If you insist that they're right and I'm wrong, you are insisting that I mean "JTB" by "knowledge". And that would be silly.
I've suggested, but not yet stated, the answer to your question: why should we believe (2)? Because it follows from our competence with our language.
But maybe, then, there isn't such a sharp contrast between your position and the intuitionist's account. For, she could accept everything you've said and add a simple claim about how justification for (2) "follows from our competence with our language", namely via application intuitions. The idea would be that concept possession (or competence with a term) comes along with dispositions to apply or not to apply the respective concept to some given object/situation - and when applied consciously and reflectively these application dispositions sometimes manifest themselves in application intuitions. I agree that, in this picture, intuitions don't seem to be so crucial anymore and maybe that's a good thing. What really does the epistemic work is our conceptual/linguistic competence instead - and a judgment like (2) could flow from this competence directly, without any need for intermediate application intuitions. However, it is a virtue of such a view that it is ecumenical enough to integrate such intuitions: There's simply no need to deny that - sometimes, yet not always and not necessarily - application intuitions are the way how conceptual competence manifests itself in rational beings. To conclude, if something like this is or could be your view, I'm highly sympathetic to it!
Hi Joachim,
Yes, I think the view you're sketching is pretty compatible with the view I've been defending here. Of course, that view is pretty controversial, and there are a number of challenges to it on the table. That's why I give the more schematic version, instead of committing to that one. Instead of offering a story as to how it is that we reliably judge whether a concept applies to a given situation, I'm content to point out that any non-skeptic is committed to our having such abilities, at least in favorable cases. Whatever it is that explains those abilities also explains our reliability about thought experiments.
Post a Comment