Swain, Alexander, and Weinberg

Stacey Swain, Joshua Alexander, and Jonathan Weinberg (Indiana University), “The Instability of Philosophical Intuitions,” with commentary by Adam Feltz (Florida State University). The paper, commentary, and authors' reply can be found here.

Comments

[Neil]

I'll get the ball rolling on this one...

First, a word in defence of a contextualist response. The contextualist does not merely predict that intuitions will be subject to an order effect: she predicts the nature of that effect. Coinflip lowers the standards for knowledge attribution; Chemist case raised it. Indeed, that’s your hypothesis too. So we should see kind of limited endorsement of contextualism as arising from the results. The objections the authors raise do not seem insurmountable: as to what time frame is relevant, well that’s at least in part an empirical question: the contextualist can agree that more experimental work is needed, but take what there is as provisional confirmation of their view. The objection that the reactions to the Charles case cover the entire Lickert scale seems to me misguided: the contextualist should be explaining the mean, not the range (of course we’ll get outliers, and of course we'll get performance errors).

Here’s an analogy. We know that taste is subject to context effects: eat some chocolate and your mouthful of coffee will taste bitter. Does that show that wine tasters are charlatans? No (at least, not by itself): presumably they know how to produce the relevant baseline context for wine tasting. They might drink water between sips of different wines, and make sure their palate isn’t spoiled by recent consumption of chocolate (or whatever it may be). Perhaps epistemologists need to be discovering the rules that allow the epistemic palate to return to baseline.

Second, I think we should see in the experimental literature a challenge for both experimenters and traditional epistemologists to say more about what knowledge is . My worry isn’t Adam’s; that the folk concept might differ from philosophers’. My worry concerns the philosophical concept of knowledge. Accounts like JTB are adequate just in case they give us a set of necessary and sufficient conditions for knowledge, but they don’t necessarily illuminate what knowledge is. I guess I’m after something like a real definition. The reason why I think both sides need to say something illuminating about knowledge is this: each side needs to give us an argument as to why intuitions are or are not going to be reliable guides to knowledge. Suppose that ‘knowledge’ is essentially response-dependent: than intuitions are going to bear upon it in a much more direct way than if knowledge is more like the subject matter of science. If knowledge is more like 'goodness' than it like, say, 'causation' then intuitions are prima facie relevant to epistemology.

[Adam Feltz]

I would like to say once again that I am overall in agreement with Stacey, Joshua, and Jonathan—both with the main argument of their paper and with their replies to my commentary. However, I would like to push on a couple of points if for nothing else it might help exorcise my own demons.

In response to my suggestion the folk may either have different concepts of knowledge, an incomplete understanding of the concept of knowledge, or that they are just making performance errors, the authors reply that each of these “apply just as well to philosophers who rely on intuitions as evidence.” It seems to me that this move is a bit too quick. It seems plausible to me that it is possible that each of these worries apply to philosophers' intuitions. Nobody would deny that (I don't think—but maybe they would), and it does not seem that we need any experimental data to establish the mere possibility that such worries apply to philosophers' intuitions. But it is another step to say that the worries actually do apply to philosophers' intuitions, and that is what I take Stacey, Joshua, and Jonathan's reply to be. It does not appear to me that the data from the folk establish this link.

This leads into my second point. It would seem odd to me that the mere possibility of philosophers suffering from the defects noted above would be sufficient to call into question the standard practice of analysis. Rather, what would call into question the standard practice is if we have some data that philosophers' intuitions are as unstable as the lay folks' intuitions. Since we don't have that kind of data, then it seems that even as things are currently, we don't have much reason to suspect that philosophers' intuitions are as radically unstable as Stacey, Joshua, and Jonathan suggest. And since there is something to be said for tradition, it will take a strong argument to establish the claim that we should shelve something so central to the current philosophical endeavor as using intuitions as evidence. Granted, all this will require some empirical investigation, and I'm all for that. But it seems to me that the burden is on the critics of the tradition to provide compelling evidence that those who are in the tradition need to do some empirical leg-work. (If such evidence is not forthcoming, then maybe it's Miller time!)

[David Leon Gil]

Let's say that philosophers claim that women have fewer teeth than men. Some enterprising experimentalist goes out and counts the teeth of college students, and finds out that women and men have the same number of teeth. So the experimentalist comes to the philosophers and says: "I've found out that your speculation about teeth is wrong." But, "Aha," the philosophers say, "You haven't found out that male and female *philosophers* have the same number of teeth." This would be quite a silly objection.

It just seems to me that, as with anatomy, you can't get anywhere particularly interesting by presuming that people are different. If you do presume differences of this sort, not only do undergrads' intuitions not speak to philosophers' intuitions; but philosophers' don't speak to "star" philosophers'. And so on.

It's possible that philosophers are different. But the only reason that Adam suggests seems to be that they use intuition *a lot*. I simply don't understand why, though, this suggests that philosophers are different. Certainly using intuitions doesn't seem to have brought philosophers, as a class, to particularly stable positions; truetemp is, after all, a fairly contested case. Perhaps if philosophers' intuitions were very different from the folks' we'd have good reason to think philosophers different. But the results here don't seem to diverge in that way. Or perhaps if philosophers already believed that their intuitions were less liable to such fluctuations, this might establish somewhat of a presumption in their favor. But I don't think that philosophers did already--they didn't even think that this problem existed. And taking this view now seems just a post-hoc and sadly a priori defense.

[Stacey, Josh, & Jonathan]

Thank you, Neil, for getting things rolling! Regarding your first point -- that an empirically informed version of contextualism might well be fully consistent with our findings -- we absolutely agree. Our target is not contextualists in all their possible forms, but rather any contextualist who would use her contextualism to try to help the intuition-driven philosopher avoid our arguments. As we say in our reply to Adam Feltz's commentary (which unfortunately was mislinked when you were writing your comment), "We do not set out here to refute contextualism; we merely intend to point out the difficulties that arise from relying on intuitions that even a contextualist will have to grapple with."

We are a little unsure about your second point, though. We are definitely sympathetic to your claim that standard 'iff' analyses of knowledge are not entirely satisfying. But we have trouble seeing how there's any implication from the kind of thing knowledge is, to what is the best method for studying it. Numbers are, we figure, about as response-independent as anything is, yet armchair methods are clearly a fine way to study them; phonemes are, we figure, about as response-dependent as anything is, yet they must be studied with rigorous work in the lab and the field.

[Ben Nelson]

Interesting and helpful paper. I certainly agree that these are empirical questions, and I applaud these forays into the sociology of philosophy. I'm also sympathetic to the conclusion. From the getgo, I think it's safe to say that intuition when taken alone is not evidence.

But what is meant by "unstable"? Stability is presumably across the lifespan and with experience. But this was not a longitudinal study. Rather, it shows an interesting correlation between variables on naive attempts to address a relatively arcane and undecided subject on the basis of limited information. That's not philosophical. In order to show something about stability which affects the philosopher, we'd have to know how they answer the question in 5 months, a year, etc.

(For what it's worth, I think the JTB account is nonsense, since truth may be deflated. But an appeal to popular opinion would not be very satisfying in an argument to that effect.)

To reiterate, no good philosopher relies upon intuition alone. Rather, they (as well as reasonable non-"professional" philosophers) use the combination of intuition and doubt within a relevant argument for the purposes of a cooperative conversation. The output of whole process MUST be a kind of philosophical evidence. Else, we're all just wasting our time.

If this kind of 'cooperative evidence' is not entirely stable, perhaps that simply shows that it is fallible. Or perhaps the imagination and goal-mindedness of the philosopher can make them pigheaded; in which case it shows that even the philosopher is not really all that willing to be reasonable.

[Stacey, Josh, & Jonathan]

Adam is concerned about our response to his suggestion that perhaps the folk are employing different concepts of knowledge, are in possession of only partial understanding of the concept of knowledge, or are engaged in making performance errors. In particular, Adam expresses a concern with the claim that “each of these ‘apply just as well to philosophers who rely on intuitions as evidence.’” It is important to note that this wasn’t our reply. Our reply was that “if these worries apply to our subjects, then they apply just as well to philosophers who rely on intuitions as evidence.” It is one thing to claim that 'if P, then Q' and quite another to claim that 'Q'. We certainly agree with Adam that the claim that “the worries actually do apply to philosophers’ intuitions” is not supported by our data. We never claimed otherwise. Rather, our point was this. Adam presented each worry as a potential problem for folk intuitions. The worries were potential problems because they were speculative and unsupported by any extant empirical research. We meant only to reply that as potential problems each applies equally to philosophers’ intuitions.

In the second part of his comment, Adam expresses doubt about the length to which speculating about potential problems to which a methodology might be subject can go towards challenging established methodology. His appeal to a principle of conservativism in defense of intuitionist philosophy raises interesting questions about the application of the principle: when it is legitimate to employ such a principle, how much weight the principle will bear, what burdens are shifted legitimately and which are taken on by employing the principle, etc. Each of these questions deserves serious consideration. For present purposes, however, it is more important to note that we agree that in the current state of play, our results don’t directly challenge a philosopher's reliance on philosophers’ intuitions as evidence. Our challenge is, rather, indirect, depending on an inference from a demonstrated instability in the intuitions of the folk to a likely parallel instability of the intuitions of philosophers. For reasons along the lines suggested above by David, we find this inference to be in good order so long as no one has any good, empirically-defensible evidence that philosophers work significantly differently from the folk in this regard.

[Adam Feltz]

That sounds right. Thanks ya'll.

I guess I have a residual worry (that might be slightly orthogonal to your paper given your claim that the studies don't directly challenge philosophers' intuitions—so I'll try to make it short). If the pressure being put on the tradition is dependent on an inference from the instability of folk intuitions to a parallel that philosophers are *likely* to have unstable intuitions, then that won't get much traction with those in the tradition. It seems then that we should just test philosophers' intuitions directly and see if they are unstable. That way we can get a nice result (one way or the other) without depending on the inference that some aren't going to buy.

[Opinionated Layman]

What is reported is that, if you ask somebody whether a particular scenario qualifies as knowledge ("Does John literally know x?"), then you can influence the answer they are going to give by presenting them with different scenarios first and asking the same question. If you show them a case where knowledge is clearly not present, then they will subsequently be more likely to say yes to ambiguous cases. On the other hand, if you show them a case where knowledge is clearly present, they'll subsequently be likely to say no to the ambiguous cases.

So what's going on? The authors suggest that we can infer from this study that the use of intuitions as evidence in favour of philosophical positions is unacceptable. After all, they say, here we can see that intuitions, about such matters as whether somebody knows something or not, depend on such irrelevant factors as the order in which questions are asked. Surely such shaky foundations can't be used to build a proof with the certainty needed by a philosopher.

But wait a moment; that seems a little hasty. Really what has been found is that the way in which individuals use the word "know" depends on the order in which questions are asked. Tell them that you're about to ask them a yes/no question about whether somebody "knows" something, and give them an example of the way in which you expect the word "knows" to be used in advance, and you will influence the answer given. So if you prepare the subject by first asking him, in a case where the answer is clearly "yes", whether an individual knows something, then the subject "sets high standards" for the usage of the word "know", and will subsequently only say "yes" if what is presented next is as good an example as the example one.

So, the first serious objection to the paper's central claim is that the results can be explained by supposing:

i. The intuitions of the experimental subjects did not depend on the order in which the questions were asked.
ii. The subject's guess about how the experimenter intended the word "know" to be understood depended on the order in which the experimenter presented the questions.

For example:

1. X knows Y in sense 1 of the word "know" and sense 2 of the word "know".
So does X really literally know Y?

2. X knows Y in sense 1 of the word "know" but not sense 2 of the word "know".
So does X really literally know Y?

Well, if I'm an experimental subject, then it's my job to figure out what this experimenter wants me to do and then do that. After seeing case 1 and answering "Yes", now I see case 2, and I think the fact that sense 2 of "know" is not applicable must be important somehow. Probably the experimenter wants to see how I respond to the absence of sense 2 but the presence of sense 1. Well, the last case was "Yes", and this case is clearly deliberately different so I suppose it must be a "No". So my guess is that the experimenter wants me to use "know" when both sense 1 of the word "know" and sense 2 of the word "know" are applicable, but not when sense 2 is absent.

If this is what's going on in the mind of the subject when answering the survey, then it would not follow that intuitions are unreliable, only that the language which the subject uses to describe those intuitions can be influenced in advance by giving them examples of the way in which certain words (like "know") are expected to be used. The intuitions, about whether sense 1 applies or whether sense 2 applies, can, therefore, be completely stable, without contradicting the results of the study.

[Ari Krupnick]

I have a question to ask the authors of this paper. But I first want to apologize for the fact that I have not carefully read either this paper or any of the other papers that the authors cite. I may very well be asking the authors a question that they or someone they cite has already answered. But I’ll ask anyway, because I am very curious to hear the authors response if they would like to share it.

In their introduction, the authors write, “A growing body of empirical literature challenges philosophers’ reliance on intuitions as evidence based on the fact that intuitions vary according to factors such as cultural and educational background, and socio-economic status.”

I am curious as to how “the fact that intuitions vary according to factors such as cultural and educational background, and socio-economic status” challenges philosophers’ reliance on intuitions. Here is why I am curious.

There does not seem to me to be any philosophical argument that is shown to be unsound by the fact that intuitions vary according to cultural and educational background, etc. For no philosophical argument has a premise that says that most people have a certain intuition, or that intuitions do not vary in that way. No philosophical argument has a premise that says that a particular person's intuitions are a “good guide” to the truth. No philosophical argument has a premise that says that a particular group's intuitions are a good guide to the truth. In short, no philosophical argument has a premise that is shown to be false by the fact that intuitions vary across cultural and educational background, etc.

For instance, Lehrer’s argument against Reliablism does not have such a premise. That argument does not have a premise that says that most people have the intuition that Truetemp does not know. Nor does it have a premise that says that the fact that Lehrer (or some group of people) has the intuition that Truetemp does not know indicates that Truetemp does not know. His argument is simply this:

1.If Reliabilism is true, then Truetemp knows that the temperature is x.
2.Truetemp does not know that the temperature is x.
3. Therefore, Reliablism is false.

The premises of this argument do not say anything about what intuitions people have, or about how widely certain intuitions are shared, or about whether Lehrer’s or anybody else’s intuitions are a good guide to the truth.

Now suppose that Lehrer’s argument was something more like this:

1. If Reliabilism is true, then Truetemp knows that the temperature is x.
2. It is intuitively clear to me (Lehrer) that Truetemp does not know that the temperature is x.
3. Therefore, Truetemp does not know that the temperature is x.
4.Therefore, Reliablism is false.

If that is his argument, then his argument does seem to depend on some claim to the effect that his own intutions are a good guide to the truth. For some claim like that seems to be required to get from premise 2 to premise 3. But that is not his argument.

So if there is no philosophical argument that is shown to be unsound by “the fact that intuitions vary according to factors such as cultural and educational background, and socio-economic status”, then in what sense does this fact challenge philosopher’s reliance on intuition? That is, if that fact does not show that philosophers say false things about intuitions when they state their arguments, then how does it show that philosophers rely on intuitions in any problematic or troubling way? Surely it is not problmeatic that philosophers rely on intuitions in the sense that what leads a philosopher to take on thing for granted rather than another is that the first is intuitively true to him whereas the second is intuitively false to him. Or there is some philosophical argument that is shown to be unsound by the fact that intutions vary across cultures etc.? What am I missing?

[David White]

This is a very interesting paper. Here are a few thoughts I have about the subject:

(1) I quite like the idea that if one wants to make appeals to so-called "intuitions" about examples that one ought to be able to back it up with experimental data. But I wonder just how often it has been the case that appeals to intuitions have been faulty because of a lack of such data. In some cases (e.g.; the sheriff who must decide whether or not to punish the innocent man to quell the angry mob) the examples are well enough known that we have, in effect, a large pool of responses to appeal to. Every time we present a famous thought experiment to a large undergraduate class we get feedback on just what kind of intuitions people have on these cases. In fact, this informally replicates what the Truetemp experiment measured.

(2) Sometimes the purpose of the thought experiment is to show that questions are not as clear-cut as we might have supposed. Bernard Williams famous "Jim and the Indians" case is one example. Here the point is made not by showing that our intuitions do not match our theories, but by showing that intuitively the case is a hard one when theoretically it should be a no-brainer. In such a case all one needs to make the point is to show that there might not be a clear and easy intuition about the case, a point that is easier to establish anecdotally than statistically.

(3) About the issue of philosophers' intuitions vs. non-philosophers' intuitions, I am surprised to see the question suggest that philosophers' intuitions might be *superior*. If anything, the worry typically seems to me to be that they will be *inferior*. This is because as philosophers, we have thought too much about the issues and may well be in the grip of a theory when trying to consider a new thought experiment and, as a result, our "intuitions" are so tainted that they are not intuitions at all. Non-philosophers, by contrast, are better positioned to give a more true "gut reaction" that we want when testing intuitions. So while I agree with those who think there might be a relevant question of just whose intuitions we are testing, I would suggest that we are quite often better off only asking non-philosophers.

(4) When doing intuition experiments the *distribution* of responses might be more important than the *average* of them. So, for example, if the mean response of 100 people to a case is 3.00, we might want to know if that is because 50 people scored 1 and 50 scored 5, because each of the five answers was equally picked, or because all 100 scored 3. All three distributions give the same average, but the wild differences in the distributions might well be important information we need to account for.

[Neil]

David,

How can someome have thought too much about an issue?? Should we reject physicists claim about the nature of reality because they've thought too much about it? If you're a fan of reflective equilibrium, then you should think that philosopher's intuitions are better than non-philosophers. They've been altered by considerations of consistency with other intuitions.

[David White]

Neil,

"How can someome have thought too much about an issue?? Should we reject physicists claim about the nature of reality because they've thought too much about it?"

My claim is not that someone can have thought too much about an issue to have a valid belief about it. My claim is that one can have thought too much about an issue to have a valid *intuition* about it. Standardly an "intuition" is supposed to be your first thought on a subject or situation, not your last one. So, for example, most people have the intuition about the Monty Hall problem ( http://math.ucsd.edu/~crypto/Monty/montybg.html ) that it does not matter if you switch after seeing the first open door. But it is easily demonstrated that your odds are twice as good if you switch. So after having had the puzzle explained to me, I would not say that my "intuition" about it has changed, but my "considered judgement" about what the right answer is has changed. The lesson we learn from the Monty Hall problem, then, is not that mathematicians who have thought a lot about probability are unreliable sources of information about probabilities. It is that even educated people should beware of their intuitions about probabilities because they can be very wrong sometimes, even when they generally agree.

"If you're a fan of reflective equilibrium, then you should think that philosopher's intuitions are better than non-philosophers. They've been altered by considerations of consistency with other intuitions."

I'm not a fan, but there is a reason that most RE fans I do know prefer to talk about "considered judgements" rather than about "intuitions". It is precisely because they are more interested in what views one has about particular cases *after* careful consideration, including comparing the case to other ones and factoring in theoretical commitments, than they are in that first "gut reaction" to the case. But intuitions - *real* intuitions in the "gut reaction" sense - are important tests of the plausibility of theoretical views we might have. How can I continue to be a consequentialist, one might wonder, if I my intuitions are that the sheriff is wrong to punish the innocent man and the doctor should not kill one person to save five and Jim should not kill the Indian? Maybe, if I am a RE fan, I might find that with enough consideration I change my views on these cases, but then they are my considered judgements, not my intuitions. The fact that I had the intuitions that I had is still important to the question of what, ultimately, is the right theory here even though my judgement has changed. And because they are important, questions about whose intuitions matter and the context in which these intuitions are produced also matter.

[Neil]

Okay. But I think you're insisting on a distinction where other people are looser with their talk (which isn't to say that your way of talking isn't preferable). Notice two things: first, on your definition philosophers don't have intuitions with regard to standard thought experiments (I don't recall what my *first thought with regard to the Gettier problem was, so I have no intuition to consult). Second, this isn't the way Swain, Alexander and Weinberg are using the words, since they *do* think that philosophers have intuitions.

Let's rephrase the debate in your terms: why the **** would anyone think that folk intuitions are better than philosophers' considered judgments on philosophical topics?

[Yentz Mahogany]

Neil,

Possibly because contemporary philosophers favor obscure and complicated falsities to obvious and simple truths.

[Neil]

Yentz,

That may seem simple and obvious to you, but I have obscure and complicated reasons for doubting it.

[David White]

Neil,

"Notice two things: first, on your definition philosophers don't have intuitions with regard to standard thought experiments (I don't recall what my *first thought with regard to the Gettier problem was, so I have no intuition to consult)."

To say you don't remember what your intuition was is not to say that you did not have one. But philosophers, just like anyone else, will have a first "gut reaction" to a thought experiment they have never heard before. Sometimes we even remember them. (For me: don't shoot the Indian, don't punish the innocent, don't kill the healthy person to save five, don't unplug the violinist, Smith does not have knowledge, don't switch doors, and choose one envelope rather than two.) When I read an article that introduces a new thought experiment, I typically take note of what my first thought on the case is. Whether or not it is the one the author thinks I will have is often quite important to how sympathetically I read the rest of the article. I suspect others are like me in this way.

"Second, this isn't the way Swain, Alexander and Weinberg are using the words, since they *do* think that philosophers have intuitions."

Well, as I just said, I think philosophers do have intuitions, but I also think SAW are using "intuition" in the "gut reaction" sense rather than the "considered judgement" sense. Otherwise, their experiment would have been to present people with five cases, ask them to consider them all together and in light of any theories that they might have about what counts as knowledge and to report back in a week. The whole point, as I understand it, of discussing the question of what the person was considering immediately prior to considering the Truetemp case is to see what influences the judgement people make at precisely the moment that they first consider a case.

"Let's rephrase the debate in your terms: why the **** would anyone think that folk intuitions are better than philosophers' considered judgments on philosophical topics?"

Again, the point is not that they are "better". Mathematicians who have thought a lot about Monty Hall have better judgements on what one should do than ordinary folk do when giving their first reaction. The point is that if you think that "gut reactions" to cases do matter at all, then you better ask non-philosophers. And when you do so, as SAW point out, you better be careful of the circumstances in which you do it. If you don't think that "gut reactions" matter, then I don't see what interest SAW's paper would hold. You could agree that one's initial opinion on a case might be shaped by what you had just been thinking about before considering it, but that the process of considering and reconsidering the case against a lot of other cases and theoretical ideas gives us plenty of time to mitigate the effect. So it would not really matter at the end of the discussion what we had thought about the case at the start, and thus what we had been thinking about the moment before that would matter even less.

[Neil]

David,

First, I used the present tense: on your definition I don't have intuitions; I had intuitions but now I don't (except with regard to new cases). Neither do you.

Second, I agree that Swain et al. are talking about gut reactions. My point was that philosophers in standard analytic epistemology are not committed (methodologically) to thinking that these kinds of responses matter. It's perfectly coherent to think that
considered judgments matter, not gut reactions. The question, to repeat, is what grounds are there for thinking that gut reactions are *better* than considered judgments? I know philosophers often talk this way, but it seems to me confused.

[David White]

Neil,

"First, I used the present tense: on your definition I don't have intuitions; I had intuitions but now I don't (except with regard to new cases). Neither do you."

When you previously wrote "I don't recall what my *first thought with regard to the Gettier problem was, so I have no intuition to consult" I took you to mean that if you did recall what your first thought had been then you would have an intuition to consult. This might be devolving into a merely semantic point, so I'll just move on to your second point.

"Second, I agree that Swain et al. are talking about gut reactions. My point was that philosophers in standard analytic epistemology are not committed (methodologically) to thinking that these kinds of responses matter."

I agree.

"It's perfectly coherent to think that considered judgments matter, not gut reactions."

I agree. Most (if not all) RE folks would not only say that it is coherent, they would say that it's true.

"The question, to repeat, is what grounds are there for thinking that gut reactions are *better* than considered judgments? I know philosophers often talk this way, but it seems to me confused."

I agree that it is confused. I don't think that they are better and don't know why one would think that, so I cannot help you here. But what I *can* say is that I do think they matter, even when one has come to change their mind as a result of philosophical reflection. (Of my intuitions I listed above, my considered judgement is different from my intuition in the cases of the violinist, Monty Hall, and the two envelopes.) One can think that gut reactions matter even if one does not think that they are better opinions about a case. They can matter for many other reasons. And it is because they matter that I find SAW's paper so interesting.

But maybe someone else here can tell you (and me) why one might think that folk intuitions are better than philosophers' considered judgements. It does seem to be a rather odd claim (especially if it comes from a philosopher!).

[Yentz Mahogany]

Neil,

I'm serious.

For example, take the "weather it", as in "It's raining". Classical grammarians are stuck on the idea that it is a dummy word which could not be filled by a referent. Chomsky, wanting to defend the idea that it was capable of being filled by content, but wary of the idea that it could not be generalized (presumably since we're attempting to explain the semantic class of "weather it", here) appealed to the notion that it can control PRO, which shows that there's an implicit anaphorical "it" in a sentence like "It's rainy and PRO windy". In other words, he produced a systematic explanation which depends upon a concept (PRO) which is part of a greater, obscure system.

Meanwhile, it is obvious that the sentence, "It's raining", has an "it" which can (and likely often is) used to talk about clouds. And the fact that it can act as an exophor in this instance decisively demonstrates that it is not a dummy word. Any further discussion, about generalization etc., is at best beyond the scope of the question.

How do I know this? Intuition, partially. In another part because I know that I'm competent in the language. In another part, because I have cooperatively attempted to understand these doubters, and have come up with the conclusion that they are producing deeply wrongheaded (or at least unnecessarily convoluted) explanations.

Mostly, though, I know it because I have nothing to lose, and so have no incentive to forget my intuitions. Academics, however -- especially young ones -- have a lot to lose by disagreeing. So they forget.

Incidentally, the Gettier cases are solved by the understanding that knowledge must always be expressed as a proposition about some clear and stable referent: indicated in language by use of the definite article, satisfied only when object can be placed in Cartesian coordinates. The use of an indefinite article to express a belief is to automatically demonstrate that one's beliefs are provisional and sub-optimally justified, and so, less plausible as candidates for knowledge (are probabilistic at best). In other words, knowledge itself is not true justified belief plus some extra criterion; rather, a proposition of knowledge concerning a clear and stable referent is (ostensibly, for non-deflationists etc.) a true justified belief.

[Rof]

As I understand intuitions, if one adopts a policy of never accepting intuitions as evidence, then one could never claim to know anything at all. We accept the validity of syllogisms and the axioms of arithmetic because they agree with our intuitions, and there is no mathematical proof, for example, that a straight line connecting two points is the shortest path. It is merely our intuition that the straight line is the shortest path, and it subsequently becomes an axiom (not a theorem) in our axiomatic formalism.

In addition, whenever one remembers something, the memory is manifested in the mind as an intuition. Consequently, if one rejects intuitions as evidence one must reject the reliability of one's own memory, and one will then be in a very poor situation if one wants to be in possession of knowledge.

In their paper, the authors mention that we do not know which of our intuitions we can rely on. What we can do, however, is know which of our intuitions rely on which other intuitions. We can therefore identify the most relied-on intuitions, those intuitions on which other intuitions rely the most. And these most relied-upon intuitions are familiar to us. Among them are the elementary relations of logic, the distinction between one and two, and so on. I must rely on my firm intuition that one and two are different numbers if I wish to understand such elementary things as what situation I am in. In addition, my intuition of objects around me in space is relied on by me when I want to understand what the present state of affairs in my vicinity is.

Perhaps the current thinking in philosophy is that one does not need to rely on intuitions to be certain of the truth of mathematical statements, because we have the symbolic proofs to assure us. However, our trust in the symbolic proof is conditional. We must inspect the rules of the system to assure ourselves, through comparison with our intuitions, that the inferences will all be valid, before we accept the conclusions. Symbolic proof is not an oracle whose statements we blindly trust for no reason.

[Oscar]

Ari Krupnick raised an important question. In what way does the fact that people have different intuitions undermine any philosophical argument? I am very interested in hearing a response.

[Antti Kauppinen]

I think it's a very bad idea to confuse 'intuitions', as philosophers usually talk about them, with gut reactions. I agree with Neil that gut reactions are really not fit to play any justificatory role in philosophical theories. In debates about people's concepts, in particular, we have excellent reasons to believe that gut reactions as such are irrelevant, as discussions of Wittgenstein's rule-following arguments have shown (centrally, you have to be able to make mistakes by your own lights). I have rehearsed these points elsewhere, so I don't want to go deeper here. But I do think it's worth pointing out that there is a sense of 'intuition' that is relevant: intuition as pre-theoretical judgment. That is, the judgment of someone who doesn't already hold philosophical views on the issue, indeed isn't even aware of such or else manages to leave them out of consideration. To make an intuitive judgment on a Gettier case, I must at least bracket the epistemological views I hold. This may be difficult or psychologically impossible, which is why philosophers' views on how things appear pre-theoretically can be unreliable.

In any case, I think the beginning of wisdom in thinking about intuitions is bearing in mind that 'pre-theoretical' doesn't equal 'gut reaction'. In my paper on this topic, I make a distinction between robust intuitions and surface intuitions that pretty much corresponds to this one. As I see it, the Swain et al. results support the skeptical point about surveys - since all we get are easily manipulable surface intuitions, survey results have little evidential value. Fortunately, living with non-philosophers gives us better access to the folk's pre-theoretical judgments than surveys ever did.

[tnadelhoffer]

Antti,

You said:

"Fortunately, living with non-philosophers gives us better access to the folk's pre-theoretical judgments than surveys ever did."

This is a point similar to the one you made in your aforementioned paper. Unfortunately, I think it is based on a highly suspect premise. After all, your suggestion is that we do not need to run any polls to guage what people's pre-theoretical intuitions are given that we already live amongst and interact with non-philosophers on a daily basis. On this view, our interactions with the non-academic masses gives us all the insight we need. But this is only true if some of these social interactions involve discussions concerning Gettier cases, True-temp cases, determinism, the relationship between intentional action and moral responsibility, etc. Now perhaps mine is an isolated case, but I don't find myself talking to my non-philosopher friends about these things very often. And on the rare occasion when I do, I find myself doing little more than trying to see where their intuitions lie. But if that is all I am doing, wouldn't I be better off testing their intuitions in a more empirically rigorous way? Moreover, if we followed your advice, we could, at best, only gain access to people's intuitions a few people at a time--which is, by my lights, inferior to getting access to hundreds of people's intuitions at once. In any event, I suspect that one of your worries is that the kind of data collection we do is not up to par--methodologically speaking--involving as it does polling people rather than conversing with them about the target subject. Is this correct?

In the event that some of you may be interested, you can find an older version of Antti's very interesting paper at:

http://experimentalphilosophy.typepad.com/experimental_philosophy/2005/12/index.html

[Antti Kauppinen]

Thomas, I don't want to hijack this discussion - I know we'll be returning to these themes time and again - but here's a couple of things. First, what we want from the True-Temp case and other outlandish scenarios is an insight into some otherwise hard-to-notice aspect of the everyday concept. What goes on when a layperson or any concept-user reaches a determinate answer about such a case? Well, she extrapolates from a past pattern of uses considered semantically appropriate by her peers. This is, obviously, a theoretical description, and very hand-wavy - see e.g. Pettit's Rules, Reasons, and Norms for a detailed story (http://www.oup.co.uk/pdf/0-19-925187-8.pdf - scroll down a bit for a nice summary chapter). If I can latch onto this pattern of ordinary use - as I hopefully have done, or I've been talking past people all my life - why couldn't I project it just as well, and with equal authority as to whether *our* concept of knowledge, say, would apply to the test case? (Obviously, there are disagreements, various explanations of disagreements, and a long story altogether to be told.) If this is the general picture, the essential thing is grasping the pattern in everyday uses of the concept, and for this, I don't need to talk about Lehrer with the guys I play football with.

Second, whatever your view of the relative priority of language and thought is, surely we deploy our concepts in communicating with each other. This creates a pressure for uniformity: if you associate different truth-conditions with knowledge claims than I do, it's at least hard if not impossible for us to keep track on what we're committing to when making them, and thus at least difficult to come to agree or disagree about things. (It's not necessarily impossible; if I notice you're using 'knowledge' for 'subjective certainty', I can translate your idiolect onto mine and get your score right.) But of course it's not just you and me who are under pressure to share the same concept, but also our other interaction partners, and their interaction partners, and so on. The web spreads out to cover the entire linguistic community. This is, in outline, how I get a hold of the folk's concept by interacting with only a few people. To be sure, the matches aren't perfect, and there may be gaps and skips in the web. In this respect, the Weinberg et al. studies on high-SES and low-SES subjects are intriguing. Their implication is that people in these groups have different, if partially overlapping concepts of knowledge. If this is the case, when the CEO asks his driver if he knows whether the bank is open, the driver's response will be guided by a different concept and may thus give the CEO the wrong idea for semantic reasons alone. (Whether contextualism can help here is an open question.) This is a line of inquiry that seems potentially fruitful to me.

Thanks for plugging the paper, Thomas - I was too bashful to do that myself. I'll upload a very slightly revised version at the same address.

[Stacey, Josh & Jonathan]

Ari summarizes Lehrer's argument against reliabilism as follows:
[1] If reliabilism is true, the Truetemp knows that x.
[2] Truetemp doesn't know that x.
[C] Therefore, reliabilism is not true.
Ari then argues that since [1]-[2] contain no reference to intuitions, empirical
research into the instability of intuitions does no work to show that the argument is unsound.

But, Lehrer is relying on intuitions as evidence that [2] is true. That is, Lehrer expects the reader to accept that Truetemp doesn't know that x, and the
basis for this acceptance is its supposed intuitive appeal (to the reader, not just to Lehrer). Therefore, if we can cast suspicion on using intuitions as evidence, we can undermine the reason for accepting the truth of [2] and, therefore, undermine the soundness of the argument. More generally, we can cast suspicion on arguments like the one above that rely on intuitions as evidence. (We suspect this way of intuitions being relevant to the status of a premise, without being premises themselves, may show a subtle error in some of Williamson's recent writing on this issue.)