McDowell (and Cavell) on criteria and skepticism

The other day, Currence reported on Ed Witherspoon's talk entitled "Wittgensteinian Criteria and the Problem of Other Minds." In his post, he wondered

how Witherspoon's discussion of Wittgenstein on criteria and whether he was a McDowellian or a Cavellian -- or whether it makes a substantial difference if they're essentially the same -- ties in at all to the discussion of two kinds of skepticism, except that Witherspoon thought Wittgenstein took himself to be addressing both kinds of skepticism.

Here are some thoughts of mine on that subject which I hope will be of some use. (You should probably read his post first, if you haven't already.)

Let me start with the intuitive motivation for skepticism. When the evidence points to something, the natural thing to do (the everyday thing, the default thing) is to believe it ( = take it to be true). If one finds the evidence insufficient, one suspends judgment; but that's not philosophical skepticism. The philosophical skeptic is the guy who says "sure, I don't have any problem with your evidence – it looks that way to me too, so I'm not asking for more or better evidence – but still, we might be wrong; so we better not say we know." In other words, we are fallible, in that perfectly compelling appearances can cause us to believe falsely; the skeptical conclusion is that in every case, we cannot know if this is one of those times, so we must therefore suspend judgment on the truth or falsity of (possibly "mere") appearances, no matter how compelling they may be.

In addressing this "scandal for philosophy," Kant diagnosed it as a variant of a much deeper problem. If subject and object are so radically distinct as to threaten the former's knowledge of the latter, as the Cartesian insists, it seems that this metaphysical gap threatens our ability not only to know, but even to think about the world. This "Kantian skepticism," as Conant terms it (as we will see if that darn book of his ever comes out), addresses the Cartesian confidence in the possibility of contentful yet radically false "appearances". Turning it around, we may say that if we are in close enough contact with the world to form contentful thoughts about how it might be, then the Cartesian epistemological scruples are pointless. You may of course bite the bullet and try to deny that your thoughts have content; but then why should I listen to your self-admittedly meaningless babble?

In the contemporary context, after the linguistic turn, we speak not simply of contentful thought but also of meaningful assertion. This is where "criteria" come in. The idea of appealing to criteria in replying to epistemological skepticism is to add to the original reason for belief (i.e. that the evidence points to the truth of P) the idea that to take P as true in such situations just is to use the term correctly. After all, generally speaking, that's how we learn the term in the first place: in the (apparent) presence of a yellow banana, nanny says "look at the nice yellow banana!", and we say "yehw bana," and so on (this is not to commit oneself to the "Augustinian" picture of the opening sections of Philosophical Investigations, but to the manifest facts which make that picture seem obligatory). Once we have learned the relevant terms, to withhold assent to "there's a yellow banana" in such situations is not to show virtuous epistemic scruple, as the skeptic has it, but instead to show vicious semantic incompetence. That's the thought anyway (and there is indeed something like this in Wittgenstein; the question is what exactly).

However, as both Cavell and McDowell point out, it would be wrong to suggest that "criteria" get you across the skeptical gap. When we answer the skeptical challenge, saying that we (do too) know that P because the criterion is satisfied, there are two ways in which the epistemic chain may yet break. First, the criteria for asserting P might actually apply, but fail to entail P; or the criterion indeed entails P, but we cannot know whether it actually applies here (that is, whether the "appearance" that P is veridical). We cannot have both, on pain of dogmatism. Our manifest fallibility is as undeniable a fact as any.

Where does this leave us? Putting Wittgenstein himself to one side for now, let's compare Cavell and McDowell. According to Currence, Witherspoon suggested at the talk that the two views differ mainly in a terminological difference between them concerning the term "criterion." McDowell takes the satisfaction of criteria for X to entail the truth of X, but allows that we can take criteria to be satisfied even when they are not, while Cavell goes the other way: criteria themselves can mislead us, but we can and do know when they obtain. In any case, both reject the "criterial theorist's" view that criteria take one across the skeptical gap. So they agree; but even so, this difference reflects an important difference in emphasis and strategy, in explaining which I think we can make helpful reference to the various types of skepticism, on the one hand, and that confusing talk about types of doubt (and whether LW was a "fallibilist") on the other. In any case that's what I'll try to do here.

Let's start with McDowell. One of McDowell's consistent concerns – becoming more explicit in his recent work on Kant and Hegel – has been to reject the Cartesian metaphysical opposition between subject and object, which is the source and stay of the corresponding epistemological skepticism. In Mind and World (MW), pressing the "Kantian skeptic" line (though not in those terms), he insists, following Wittgenstein, that the content of our concepts is not confined to the subjective side of the Cartesian gap, but instead "[does] not stop anywhere short of the fact" [PI §95]. In contemporary terms (MW p. 27):

[T]here is no ontological gap between the sort of thing one can mean, or generally the sort of thing one can think, and the sort of thing that can be the case. When one thinks truly, what one thinks is what is the case. [...] Of course thought can be distanced from the world by being false, but there is no distance from the world implicit in the very idea of thought.

The idea, then, is that when we turn to the more fundamental semantic consequences of radical subject-object dualism, the familiar fact of human fallibility no longer even seems to have the epistemological significance the skeptic claims it does. This can thus help us resist a perennial philosophical temptation, viz., to metaphysicalize that innocent epistemic gap and identify it with that same picture's supposed ontological gap between subject and object (thus reinforcing it, as well as the resulting skepticism). If we do this, when we turn back to epistemology, the Cartesian argument "effects a transition from sheer fallibility (which might be registered in a 'Pyrrhonian' scepticism) to a 'veil of ideas' scepticism" ["Criteria, Defeasibility, and Knowledge" [CDK] p. 386n).

Here's how that works. The skeptical argument gets its traction from the idea that deceptive appearance and veridical manifestation are phenomenologically indistinguishable, and thus that "one's experiential intake—what one embraces within the scope of one's consciousness—must be the same in both kinds of case" [CDK p. 386] That "highest common factor" [HCF] between the deceptive and veridical cases is thus essentially incapable of providing evidence for one or the other. The difference between the two cases is something extra: the actual connection to the world which makes the veridical case veridical. But ex hypothesi we cannot tell which is which; the world's contribution to the veridical case, on this view, is "blankly external" to our experience.

But as we have seen, McDowell disputes the idea that the content of our experience must be construed as an HCF in this way. When I lack an actual connection to the world, it's not that my contentful perceptual experience was deceptive, but instead that I haven't had a perceptual experience at all, only an illusion of one. (I have conflated them here, but as I read it, the connection to the somewhat different linguistic version of this thought is this. My illusion of perceptual experience can result in my having a mistaken but contentful thought if on other occasions I have indeed had (veridical) perceptual experiences in the course of learning and using the concepts which make it up. The virtue of McDowell's focus on the perceptual-experience aspect instead of the contentful-concept aspect of his picture is that the latter, like Davidson's holistic view, makes it look as if it is only global skepticism which it renders ineffective: this thought can be a contentful mistake only if not all of them are. That is of course true too; but the perceptual-experience version is more powerful – and its metaphysical import more directly anti-dualistic – in that it applies even to single cases.)

McDowell's alternative conception of perceptual experience is "disjunctive":

[A]n appearance that such-and-such is the case can be either a mere appearance or the fact that such-and-such is the case making itself perceptually manifest to someone. [CDK p. 387]

I either have a perceptual connection to the world or I do not. Let's say it seems to me that I see a cat. Either I do, and a cat has manifested itself in my experience – a veridical appearance – or there is no cat to be seen. In neither case is there an epistemological or metaphysical "intermediary" (either standing between me and the cat, on the one hand, or disguising its actual absence, on the other), in the manner of the HCF.

When a fact is perceptually manifest to one, saying or believing that it is is thus guaranteed to be true, and thus, in the skeptical context, a "criterion" of truth (though of course we may not take advantage of this fact: McDowell is careful to identify manifestation with the availability of knowledge, not its achievement, as even a veridical appearance may not lead to belief (perhaps because of skeptical scruple!)). Of course the skeptic, suspicious of this talk of epistemic "guarantees," makes the natural objection, as if McDowell were trying to sweep our fallibility under the rug in order to meet the skeptical challenge. This distinction, he says, doesn't help; the fact remains that we can't tell when we are deceived.

But McDowell's concern is not to deny epistemic fallibility, but to render it epistemologically, and thus metaphysically, uninteresting. His strategy is thus inherently anti-skeptical, affirming knowledge in the face of skeptical doubt. The skeptic can extract a concession that there remains a sense in which we cannot know when we are deceived (after all, to deny this is to deny that deception is even possible). Even where we are most sure, he says, we must leave room for doubt. But if this be doubt, it is an uninteresting – or perhaps "weak" or "imaginary" or "possible" – doubt, a mere footnote to our firm affirmation of belief. I am certain – beyond present doubt – that I am sitting before the fire at my computer; but every such belief is corrigible, in that I grant the conceptual, if not actual (or, as some pragmatists say, "serious") possibility of error. Give me more evidence, and I may come to change my view (so while I may be completely certain, I'm not "absolutely" certain – like I care about that anyway). That's all the "doubt" that human fallibility gets you – and it's not enough to underwrite a seriously skeptical position.

Currence reports that here was some discussion of this point at Witherspoon's talk:

Witherspoon did make a really odd claim, however, that everyone (Conant, Finkelstein, grad students) picked on during the discussion after the talk: Wittgenstein was a fallibilist. He said this in the context of a distinction he drew between "weak doubt" and "strong doubt". He switched between "weak doubt" and "imaginable doubt". This seemed massively confused to everyone; if being a fallibilist means no more than recognizing one has been wrong about things before, then every reasonable person is a fallibilist, and the claim is uninteresting. If, however, being a fallibilist means something substantial -- and I think it does, something along the lines of making intelligible assertions like 'I am justified in believing x, but I could be wrong' -- then Wittgenstein is about the last person I'd want to say was a fallibilist.

Conant made this point through a humorous example in which someone doubts that the room is not safe, saying, "This room is not safe!"; we ask, "Why isn't the room safe?", and they respond, "I don't know, it's just not!"; we wouldn't say they've offered a doubt at all. "Chicken Little worries" do not fall under the genus "doubt", and it is a sham to call them "weak" or "imaginary" or "possible" doubts.

My first reaction to this was that I don't see why anyone (anyone in that room, anyway) should be surprised if Wittgenstein thinks it important to stress something which "every reasonable person" believes, and which would thus constitute an "uninteresting" claim. On some views, that's all he ever does. But let's let that go.

Conant's Chicken Little example looks weird to me. (I assume that there is a typo here, and that it should be "doubts that the room is safe," not "not safe".) If someone says "This room is not safe!", he is indeed doubting that the room is safe; but he's also claiming that it isn't, which is not a skeptical thing to do. Chicken Little acted on his belief, insisting that the King be notified that the sky was falling. We have no difficulty attributing belief to him (and thus actual doubt about our safety). In fact his problem is not skepticism but gullibility, and our proper response to him is itself skeptical (i.e. garden-variety rather than philosophical): his evidence – a conk on the bean (or "I don't know, it just is") – is insufficient for such a remarkable claim. So Conant is right that this shouldn't count as "doubt" in the relevant sense; but that's only because it's a belief and thus not relevant to the issue of skeptical doubt. In particular, that's not what I take to be the point of talk of "weak" or "imaginary" or "possible" doubt.

The more relevant case, it seems, is this. I say, or assume, that the room is safe, but our friend demurs. Unlike Chicken Little, though, he is perfectly happy to remain here with us. Practically speaking, he says, the evidence is sufficient to warrant staying; yet he prefers (he says) to suspend judgment on the truth of "the room is safe," for familiar skeptical reasons: if we were deceived, and poison gas were about to seep from the ventilation, killing us all, things would look exactly as they do now. There's no evidence that this will happen, so there's no reason to leave. But we cannot claim knowledge that the room is safe.

This is what Peirce calls "paper doubt": philosophically motivated (e.g. Cartesian) demurral, conspicuously not backed up by action. You say you are in doubt; but you not only show no intention to leave the room (as you would if you were actually in doubt about your safety), but you're not even trying to allay your alleged doubt through inquiry. In claiming to doubt, you are simply registering your fallibility and drawing what seems to you to be the proper philosophical conclusion. But purported doubt (or belief), which has no connection with inquiry and deliberation is not doubt (belief) at all. The problem is not that the "doubter" cannot support his purported doubt, but that given his actions there's no reason, beyond his mere assertion, to attribute it to him at all. So the Chicken Little case is not germane. No-one denies that C. L. actually believes the sky is falling (and thus doubts that we are safe); it's the best way to explain his actions, including his urgent desire to see the King.

"Weak doubt" is not a good word for the bare concession of fallibility, but there's nothing wrong with "possible" as opposed to "actual" doubt (again, pragmatists oppose "theoretical" to "serious" possibility of error; it is when the former aspires to the latter condition that they (we) expose it as "paper doubt"). As for what Wittgenstein thought, that's a thorny issue. His reflections on these and related matters in On Certainty are inconclusive at best. I do agree, though, that simply to state that he was a "fallibilist" (not that that simple view is Witherspoon's) is highly misleading. But I won't get into it here.

I've already gone on for ages, so let me defer extended discussion of Cavell's position. I'll just finish the comparison with McDowell re: skepticism and criteria. As we saw, Cavell too concedes the failure of criteria to bridge the skeptical gap. But his philosophical strategy is very different from McDowell (even while sharing a great deal, in Wittgenstein and out). I quote from the back cover of Richard Eldridge's Cavell volume in the "Contemporary Philosophy in Focus" series:

At the core of [Cavell's] thought is the view that skepticism is not a theoretical position to be refuted by philosophical theory [i.e. "constructive philosophy"] but a reflection of the fundamental limits of human knowledge of the self, of others, and of the external world that must be accepted. Developing the resources of ordinary language philosophy and the discourse of thinkers as diverse as Wittgenstein, Heidegger, Thoreau, and Emerson, Cavell has explored the ineliminability of skepticism in philosophy, literature, drama, and the movies.

It's a good collection (check the link for an irresistible used price!), but unfortunately there's no article devoted to Cavell's views on skepticism in particular. Anyway, where McDowell as well as Cavell grants human fallibility (where the "criterial theorists" had no room for it, as the skeptic shows), the former shrugs it off as uninteresting, on the way to reaffirming our metaphysical and epistemological connections to the world, while the latter instead allows it to bring in its train the skeptical point about fundamental limitations on our knowledge, famously granting "the truth in skepticism," i.e. that our relation to the world "may not be one of knowing – or at least what we think of as knowing." So the skeptic wins; but then the victory turns to ashes from a Cartesian point of view, as Cavell proceeds to reinterpret its significance profoundly. So while he and McDowell may not easily be seen to agree on doctrine, or even on Wittgenstein interpretation, their views may yet be seen as helpfully complementary.