Knowing That P Without Believing That P

(with Blake Myers-Schulz) now available in draft here.

Abstract:

The standard view in contemporary epistemology is that knowledge entails belief. Proponents of this claim rarely offer a positive argument in support of it. Rather, they tend to treat the view as obvious, and if anything, support the view by arguing that there are no convincing counterexamples. We find this strategy to be problematic. In particular, we do not think the standard view is obvious, and moreover, we think there are cases in which a subject can know some proposition P without (or at least without determinately) believing that P. In accordance with this, we present four plausible examples of knowledge without belief, and we provide empirical evidence which suggests that our intuitions about these scenarios are by no means atypical.

Comments welcome, as always! (Either on this post or to my email address.) This research was previously summarized in this post. The current version presents the issues and results in more detail and includes some new controls to address objections raised in the comments to the earlier post.

[Cross-posted at The Splintered Mind.]

Comments are now closed.  Continuing comments are welcome at the Splintered Mind crosspost.  There was also considerable discussion at Certain Doubts.

Comments

[Clayton]

Interesting paper. I had a couple of quick comments.

(i) I have a worry about the prompts. "thinks" vs. "believes in the philosopher's sense". I might worry that "think" is a bit like "recall" and it is more keyed into what the thinker has before the mind and affirms rather than what plays the functional role of belief. I had thought that "belief" just meant something like "would treat as true for the purposes of getting things right". In response to the modified Radford case, I wonder what the subjects would say if you asked them to comment on an explicit denial. Would they agree that she does not believe what you claim she knows?

(ii) I won't deny that epistemologists sometimes say things are obviously true that aren't obvious or true, but I do think there are arguments for the K-->B view worth considering. I'd look at Stanley's arguments in his paper on knowledge and certainty. Consider, "She knows that Reykjavik is not in Sweden. Indeed, she believes that it is not." Seems strange, to me. Why? On the view that there is no entailment there, the belief-report adds additional information not contained in the knowledge-ascription. Why does it seem like a redundant conjunction akin to, "She knows that Reykjavik is not in Sweden. Not only that, Reykjavik isn't in Sweden"?

(iii) There's an interesting paper of Marc Moffett's on the relation between knowledge and belief you might look at. If I recall, it offers some evidence that belief and knowledge relate you to different things, propositions in the one case and facts in the other. (See his, "Knowing facts and believing propositions: A solution to the Problem of Doxastic Shift" in Phil Studies).

[Eric Schwitzgebel]

Thanks for the comment, Clayton.

(i.) You might be right about the intuitive difference between "thinks" and "believes". However, on three of the four examples, the results look the same either way, so we aren't *too* worried about that. And: We didn't want to put any explicit attributions or denials of belief or knowledge in the prompts, so as not to tilt the field too much one way or another. My guess is that if the unconfident examine had said "I don't really believe she was born in 1603, but I suppose that's my best guess", that would have depressed both the know and believe attributions. Whether it would have reduced knowledge attributions below 50% I don't know. We, or someone else, could try a variant of that sort if it seemed theoretically important enough.

(ii.) Thanks for the tip on the Stanley paper. I agree that statements of the sort you make seem somewhat unnatural; but to my (possibly corrupted) ear they don't sound wholly unnatural or like *nothing* is added by the belief clause.

(iii.) Thanks for the tip on Moffett too!

[John Gee]

This is an incredibly interesting subject, so thank you for bringing it to my attention. I wouldn't comment except that the paper reminded me of something I read recently in Hume's Enquiry Concerning Human Understanding (Section VIII, Of Liberty and Necessity). He suggests that everybody believes in determinism, because everybody acts as if it were true (assuming the same causes will produce the same results), even if they avow a contradictory belief.

On that basis, I would suggest that there are four elements at work here: in addition to capacity and tendency, we can also distinguish between acting as if P and saying that P. If you put those in a 2x2 matrix, you get four possible definitions of knowledge and belief. So, there is one arrangement in which Juliet believes that student athletes are equal but does not know it (if belief=tendency to say that P, and knowledge=tendency to act as if P). I'm not saying those are the correct definitions, but it's not immediately clear to me that capacity and tendency to act as if P should be the two preferred definitions (which it seems they are in the paper.)

I apologize for making your paper fit my terms. In any case, I know that's not the main thrust of the paper, and I think it succeeds admirably in its stated goal of muddying the waters.

[jonathan weinberg]

Regarding the Reykjavik case: granting that the linguistic data is as claimed, there still seems to be a further issue that the following sort of claim can sound ok, in the right supporting circumstances:

She knows that Reyjavik isn't in Sweden, but she doesn't believe it.

E.g., she's on a quiz show, and is having a brain freeze on a relevant question leading her not to trust her sense of the matter, even though she has a well-attested mastery of European capitals. Might be more palatable as "Dang it, she _knows_ that Reykjavik isn't in Sweden, but she just can't bring herself to believe it right now."
(See Jason's examples (18) and (19) in his paper, and his discussion of 1p/3p asymmetries in this area more generally.)

If we postulate that there's a _default_ inference from Ksp to Bsp, perhaps even a strong one, but that it can be defeated in the right sort of circumstances, then it seems that we can accommodate both the sentence in Clayton's comment and ones like this one (and like in the circumstances of Eric & Blake's targets).

[DenubisX]

It's been my understanding that Knowledge destroys belief. (Rather that belief is faith without factual backing, which is one construction of knowledge).

Clearly I'm not of the same mind as the current literature (which is mostly fine, as this isn't my research area.) Do you have any links to discussion showing the ontological change in the nature of belief, or is my understanding of the knowledge/belief dichotomy just purely incorrect?

[Marcus Arvan]

Here are my two cents: I suspect that your respondents (non-philosophers) are working with an incorrect folk-psychological view according to which beliefs are *occurrent* mental states. For example, in your test-taker example, the student who answers the test-question correctly would not *say* "I believe such-and-such is the answer." She is, after all, unsure. It is perhaps for this reason that a majority of your respondents said the test-taker does not believe the answer. But is this a philosophically adequate view of belief? Maybe not. Last time I checked (which was a while back), many philosophers take beliefs to simply be *dispositions* of various sorts. If this is right, then if the test-taker is disposed to answer the question correctly -- which she is, by hypothesis -- then the respondents to your studies who deny that she believes the answer are simply operating with an incorrect philosophical conception of the nature of belief. In that case, the studies don't suggest what you take them to suggest: that knowledge really is possible without belief. They show at most that many respondents *believe* this to be possible (on the basis of an incorrect conception of belief).

[Blake Myers-Schulz]

DenubisX, thanks for comment. You may want to check out Merrill Ring's (1977) article "Knowledge: the cessation of belief." I'd have to double check. But if I recall correctly, Ring argues for a view that is somewhat similar to the position that you propose.

[Eric Schwitzgebel]

Thanks for all the comments, folks!

John: I think you're probably missing a "not" in a crucial place. (Either that, or I'm not getting something important in what you're saying.) Interpolating that "not", I am inclined to agree with your model -- though in the mess of human cognition I'm inclined to think the saying / acting as if distinction will get blurry and there will be few really clean cases of one without the other.

Jonathan: Thanks, that's helpful, and I'm inclined to agree.

Marcus: That might be made to work for three of our four cases, but I don't think that will work for the implicit racist case. That case is not referenced to a particular occurrent moment, so it seems it has to be interpreted dispositionally. Also, I suppose I'm inclined independently to doubt your starting hypothesis that ordinary English speakers normally understand "believe" occurrently. So, for example, I think few would implicitly or explicitly accept that if I go into the next room and fall into a dreamless sleep you can no longer talk about what I believe. (Is that a fair example?)

[Eric Schwitzgebel]

Correction: Where I say "implicit racist case" above, I mean the case of the professor prejudiced against student athletes. (Originally this was a racism case, but we changed it so as not to invoke that distracting issue.)

[John Gee]

Eric: I assume you think the “not” should go in my counterexample, like so: “Juliet believes that student athletes are not equal but does not know it.” That is, she does not know her own belief. I was trying to say that Juliet believes that student athletes are equal (because she will say so if asked) but she does not know that student athletes are equal (because she doesn’t act as if that were the case). It would probably be better to reverse “know” and “believe," but my point was primarily that I think Juliet’s example works better on an acting/saying distinction than a capacity/tendency distinction. The Margolis quotation, at least superficially and out of context, includes both: knowledge implies a “capacity to provide information,” while belief implies a tendency to “perform appropriately.”

I think you’re exactly right that this would make things blurry, but I’m not sure that you can avoid “the mess of human cognition” with the capacity/tendency account either. What are the criteria for having a capacity? Is there a ceteris paribus mental state in which a person will act on their capacities (building on Jonathan’s account of interference with the default)? Such would clearly be the case in the test-taking and missing-the-turn examples: remove the test or the music, the shakiness of belief goes away, and knowledge is unambiguous. But what’s the interference in Juliet’s case: a culture of academic snobbery, exposure to bad examples of jocks, an ex-spouse? Those sorts of cognitive interferences are far more stable, and I think they present difficulties. I’m not sure, then, that Juliet actually does have the capacity to act on the information given to her. Given how flagrantly she ignores the information, she might have the same capacity to treat her students equally if she didn’t “know” they were equal. But the popular intuition appears to be that she does meet some other knowledge criterion.

[Eric Schwitzgebel]

John: Yes, thanks for that clarification. I agree that my (and Blake's?) view does not avoid the messiness of human cognition, and the distinctions will get blurry. I see that as a feature not a bug, though; I wasn't sure whether you would feel the same way about the view you put forward.

I agree that we haven't developed very well in what sense Juliet has the knowledge-relevant capacity. The capacity to provide information seems like a reasonable candidate; but I'm not ready to sign up for it quite yet. The capacity to deploy the information in action-guiding (and speech-guiding) ways seems appealingly more general, but also has too much of the whiff of the storehouse metaphor for my tastes; and it remains too vague. (I develop some of my thoughts against the storehouse metaphor in McGeer & Schwitzgebel 2006 and in an unpublished manuscript on container metaphors on my academic homepage.)

[Ichthus77]

We can know without knowing we know, because there are different levels of knowing. “Knowing that we know”–that sort of knowing does involve belief. Intuitive knowing does not involve belief, until it does, but then it is ‘reasoned’ knowing. Spiders know webs, but they don’t know they know webs. Birds know nests, but they don’t know they know nests. We know moral truth intuitively, but many of us do not know we know it.

[Eric Schwitzgebel]

Thanks, Ichthus! Broadly speaking, I'm inclined to agree with that.