Below is from a response to his “Contextualist Solutions to Skepticism”, that I sent to Stephen Schiffer.
1. CONFOUNDING, BAMBOOZLEMENT ARE COMMON (EVEN FOR QUITE COMPETENT USERS OF A TERM)
I understand that you don’t want to be just pushing a problem that’s an instance of the very general “paradox of analysis”: This is to be a special problem for indexicals. With a non-indexical, it’s sufficient for a speaker, when asked what he meant by saying “P”, to respond that he meant that P. But where indexicals are involved, that’s not good enough, b/c there is no proposition that P; “P” expresses, depending on context, one of any number of different propositions. And the problem, I take it, is that, if the speaker’s in no position to tell you which of those prop’s he intended to assert, then he didn’t intend to assert any of them (as opposed to any other), and then he can’t have meant any of them (as opposed to any other).
I, on the other hand, am not all that worried by the facts that you adduce against Contextualism, like that “No ordinary person who utters ‘I know that p would dream of telling you that what he meant and was implicitly stating was that he knew that p relative to such-and-such standard” (p. 16), because I think the “confounding” that you complain that my theory involves speakers in is actually very common and pretty harmless.
An example: In a class discussion of a lottery case, one of my students insisted that the character in the example (who had purchased one of 20 million tickets, where only one ticket was a winner) knew he would lose the lottery when he bought the ticket. Other students attacked this position. In the course of defending himself, the first student insisted that the character was very confident that he would lose. I asked the student whether he meant to be backing away from his earlier claim that the character knew to the weaker claim that he was very confident. This is quite close to an exact quote of the student’s response: “No, I’m not backing down. It’s *not* a weaker claim. ‘Knowing’ and ‘being very confident’ mean exactly the same thing!” This occured on Monday, and so was after our latest e-mail exchange, so I thought to ask the student: “So, you’re telling me that by saying that Henry knew, you just meant that he was completely confident?” He responded with a “Yes.” I then trotted out a series of examples in which characters are very confident of things that they nonetheless don’t know to be so. (Many, but not all, of the examples involved false but very confidently held beliefs.) In the light of these examples, the student admitted that he was wrong about “So-and-so knows such-and-such” meaning the same thing as “So-and-so is very confident that such-and-such.” What’s more, he not only admitted that the sentences mean different things, but admitted that, unbeknownst to himself, *he* had all along meant something different by ‘knows’ than he meant by ‘is very confident.’ Here, I’m very inclined to agree with the student’s later self about his earlier self: He *had* all along meant something different by ‘knows’ than he had meant by ‘is very confident’, even though, if you had asked him (before he was exposed to my convincing counter-examples), he’d have told you — as he told me — that he meant the very same thing by the two. Part of the reason why it’s so plausible to say that, unbeknownst to himself, this student had all along meant different things by the two is that all along he was inclined to describe certain cases (like those involving confidently held false beliefs) as cases of confident belief, but not as cases of knowledge. He had all along been sensitive, in his use of ‘knows’, to the fact that knowledge must be of a truth, even though he wouldn’t have come up with the truth truth-condition on ‘knows’ if you had asked him for an analysis of “S knows that P”.
Is this student an extreme example of a confused speaker? Hardly. He’s actually quite bright, and seems to be a very competent user of ‘knows’. I suspect that if you got out into the world of the *real* Stellas and Als — no more college students at selective universities — you’d have to admit that this student is a *very* mild case. These others will express much more radical mistakes about when they mean by one sentence the exact same thing as they mean by another. And these will be mistakes. They mean different things by the sentences, even if they insist they mean the same thing by them.
Did my student know what he was saying? Well, yes and no. He in some sense knew not to apply ‘knows’ to all cases in which the subject was very confident, although he’d have denied knowing this. He had the ability to apply ‘knows’ correctly. But, to mimic a passage from your paper, he thought that by “Henry knows that…” he meant the very same proposition as that expressed by “Henry’s very confident that…”, when, in fact, he meant some quite different proposition. Happens all the time. In some important sense, we don’t always know what we mean, though we mean it nonetheless.
So, in response to your wrap-up of section IV —
But that error theory has no plausibility: speakers would know what they were saying if knowledge sentences were indexical in the way the Contextualist requires. (p. 18)
— I’d say, regarding the material following the colon, “Well, yes and no.” If Contextualism were correct, they’d apply ‘knows’ differently in different contexts: Saying subjects know in relatively easy contexts when they’d have described those same subjects as not knowing if they (the speakers) had been in tougher contexts. They’d be sensitive to this variation in content in how they applied ‘knows’. But that’s fine: This expectation of Contextualism is pretty well met. What ordinary speakers are happy to call knowledge in easier contexts, they will deny is knowledge in tougher contexts. But I’d deny the claim that if Contextualism were correct, then ordinary speakers would, if asked, say that they mean something different by ‘knows’ in different contexts. I don’t see why their failure to do that would be any more damning than is my student’s failure to realize (before the examples) that he meant two very different things by ‘knows’ and ‘is very confident’.
And I’d deny the claim that if Contextualism were correct, then if speakers were asked what they meant by saying that someone knew, they’d respond by saying that they had meant that the person knew according to such-and-such a standard. But more on that below.
2. “IT’S RAINING” VS. “…KNOWS THAT…”
One who implicitly says that it’s raining in London in uttering, ‘It’s raining’ knows full well what she’s saying; if asked, she’ll tell you that what she meant and was implicitly stating was that it was raining in London. (p. 16)
I have my doubts. Sometimes, maybe. But usually not. You’re in New York, talking on the phone with your non-philosophical friend, who’s in London. She looks out of her hotel window and says, “Oh, it raining.” You say, “What do you mean by ‘It’s raining’?” The only response I’d advise you to expect is something along the lines of, “All I mean is that it’s raining”, or, even more likely, “What do you mean what do I mean?” That is, you’d get responses analogues to the responses you’d likely get if you asked your ordinary guy what exactly he had said when he uttered “I know that Placido Domingo is scheduled to sing at the Met this season.”
You will get a more hospitable response if you ask your friend a *directed* question — a question that comes complete with several options to choose among — like: “What do you mean? Do you mean that it’s raining in London, or that it’s raining in New York, or that it’s raining in Madrid, or what?” But you’ll also likely get a more hospitable response if you ask your opera fan a directed question, like, “What do you mean? Do you mean that you know that absolutely for certain, or that you know it well enough, or pretty well, or what?” Most will accept such a question as a containing a legitimate of options as to what they meant. Very few, I suspect, would reject the question by saying something like, “‘Know’ doesn’t mean all those different things. It just means know. And all I meant was that I know he’s scheduled to sing at the Met this season.” The fact that speakers would accept a directed question about ‘knows’, and would not reject it as above, I think — given your insistence that speakers know what they mean — helps to make plausible the thought that they do mean different things — they do express different propositions — by sentences containing ‘knows’.
Anyway, I think that, as I’ve argued in the above two paragraphs, the difference in what ordinary folks would say if asked what they meant by “It’s raining” or “I know” is not so pronounced as you let on. I don’t claim there’s no difference at all. You’d probably be a *bit* more likely to get a hospitable response to a non-directed question in the case of “It’s raining” than in the case of “I know”, and perhaps a very few folks actually would reject a directed question regarding “I know”, while nobody — or almost nobody — would reject a directed question regarding “It’s raining.” Still, these differences will be quite small, and not nearly pronounced enough to support the argument you want to make. Or so it seems to me.
3. THE ERROR THEORY INVOLVED IN YOUR UNHAPPY-FACE SOLUTION: OUT OF THE FRYING PAN, INTO THE FIRE
The Contextualist’s problem, according to you, is that, according to him, “Those uttering knowledge sentences are both referring, unbeknown to themselves, to different knowledge relations and confounding the knowledge relations to which they’re unknowingly referring” (p. 17). That is they express different propositions, depending upon context, by their utterances of sentences containing ‘knows’, and they themselves don’t realize they’re doing this — which can lead them into mistakes.
But which proposition, according to *your* (unhappy-face) solution, are they expressing? A straight-forward invariantist can simply claim that they’re expressing the proposition that S knows that P, that being the only proposition expressible by the sentence. But this invariantist can’t account for the (at least seeming) skeptical paradox. On your view, the concept of knowledge contains a deep-seated incoherence (p. 21). And, given especially what you write in fn. 14, but also given some of the material in the text around that note, my guess as to the answer to the question with which I began this paragraph is this: They’re expressing no proposition at all.
If that guess is right — and the rest of the comments of this section are conditional on it being right — then you seem to me to be involved in, if anything, a *more* radical (and more problemmatic) error theory than is the Contextualist. According to the Contextualist, they don’t realize they’re expressing different propositions by the same sentence. They think they mean the same thing, but they really mean two different things. But according to you (given my guess), they think they’re expressing a proposition, but in reality they aren’t. They think they’re meaning something, while in fact they mean nothing at all!!
To be honest, *I’m* not all that bothered by such a view. Other things being equal, I’d prefer to avoid a theory according to which folks are failing to express any proposition at all when they think they’re expressing something true or false. But othe things are often not equal, and I wouldn’t be all *that* reluctant to accept such a view under sufficient pressure. (In my dissertation, I gave such a treatment — not unlike your own — of the sceptical paradox a good run for its money before finally rejecting it. I considered it at length b/c I thought it was a real contender.) But my openness to such views is partly explained by my not thinking that ordinary folks are such great experts regarding what they themselves mean (or fail to mean).
But *you* should be very worried, I’d think. You reject Contextualism b/c it involves speakers in thinking they’re expressing the same proposition by a sentence in two contexts, while in fact they’re expressing different propositions. But you end up with a view according to which they think they’re expressing a proposition, when in fact they’re not expressing any proposition at all. Don’t they know they’re failing to express any proposition? How can they be so ignorant of what they themselves mean (or fail to mean)? Actually it’s worse: It’s not that they’re wrong about which particular proposition they mean; rather, their very belief that they’re expressing *any proposition at all* turns out to be false!
4. MILD SKEPTICISM AND VAGUENESS
It’s clear that knowledge sentences are subject to this sort of vagueness-related variability. In certain conversational contexts you count as knowing that your spouse is faithful; in others you can’t really be said to know. The reason this sort of variability is of no use to the Contextualist is that speakers are perfectly aware of when it’s going on. If you claim to know where your car is, and someone challenges you, “But how can you be sure it wasn’t towed or stolen?”, you’ll merely get impatient at the questioner’s obtuseness. (pp. 17-18)
The first two sentences above make you a Contextualist in my book, assuming that by saying that in easy contexts “you count as knowing” you mean that you can be *truthfully* said to know, and that you also mean that in the other, tougher contexts, you can be truthfully said to not know. Then, you’re committed to speakers expressing different knowldge-relations by ‘knows’ in the two contexts. But that’s OK by you, b/c you think speakers are perfectly well aware of what’s going on here. By contrast, according to you, they’re not similarly aware of such a shift in meaning when the tougher standards get to be ultra-tough, or TOUGH.
As with the contrast you allege and I addressed in section 2 of these comments, I don’t find such a contrast.
The obtuse questioner you consider is presenting a skeptical hypothesis to challenge a claim to knowledge. He’s using the same basic argument as is the BIV-skeptic. Only his hypothesis is milder: It’s that the car was towed or stolen, not that one’s a bodiless brain-in-a-vat or some other wild creature of philosophy.
When I present the skeptical argument I call AI — whose basic form is:
1. I don’t know that not-H
2. If I don’t know that not-H, then I don’t know that O
So, C. I don’t know that O
I often start off with such a mild version — where something like MY CAR IS IN THE MAIN STREET PARKING LOT is the O, and MY CAR’S BEEN STOLEN AND IS IN A CHOP SHOP is the H. And, of course, I also deal with more radical H’s, like the BIV hypothesis, where the targeted O is the seemingly more secure I HAVE HANDS, or the like. I don’t find the contrast you allege. The impatience at obtuseness of which you speak is also present — in fact, is *more* intense — in the case of the radical argument. In both cases, ordinary speakers are aware that *something* is going wrong. But, in both cases, they disagree widely when pressed as to what exactly is going wrong. Some go the (roughly) Moorean route of claiming that they really do know that not-H. Some grope towards a Contextualist solution. I find no marked difference. Again, it’s not that there’s no difference at all. For some reason, the Moorean route seems a *bit* more popular where it’s the radical version of the argument that’s being discussed. But it’s not *at all* as if ordinary speaker’s know precisely what’s going on in the mild version. Since we both appear to go Contextualist on mild versions of AI, we’ll agree that “knowing what’s going on” with mild versions of AI amounts to going Contextualist on the paradox the argument presents. Unfortunately, the contrast you allege just isn’t there. Ordinary speaker seem no more inclined to go Contextualist on mild versions than they are on the radical versions of AI. In my experience, both the Moorean and the Contextualist route are a little *less* popular wrt mild versions than they are wrt radical versions. The approach that becomes more popular when it’s a mild version of AI under discussion is the skeptical approach — far more seem willing to admit that the skeptic is just right — though they’ll also typically add something to the effect that he’s being a pain in the butt! At any rate, I’m quite certain that you won’t find any dramatic increase in appeal of Contextualism in cases of mild versions of AI. I think in fact that there’s a slight decrease. Speakers no more know what’s going on with the mild version than they do with the radical versions.