Monday, June 26, 2006

Knowledge Without Belief?

Suppose that yesterday you read an email about a bridge closure. You'll need to commute on an alternate route for a month. Yet here you are today, governed by habit, driving straight toward the closed bridge. In a moment, you will remember that the bridge is closed, but you haven't yet.

Now normally, I think we'd say the following things about you: You've forgotten that the bridge is closed. And you know the bridge is closed. Consider what you'd say to a passenger, for example, the moment after you remember: "Whoops! I forgot the bridge was closed"; "Oh, that was dumb of me; I knew the bridge was closed".

But do you believe the bridge is closed, as you drive blithely toward it? I don't feel much of an ordinary-language pull one way or another on this. But most contemporary philosophers of mind regard "think" (in the simple present, not the present progressive) as a fairly straightforward ordinary-language substitute for "believe" in many contexts. Do you think the bridge is closed, in those moments of forgetfulness? This sounds strange to my ear.

So maybe we can say you know the bridge is closed, but you don't think or believe that it is? Or (to change the example), as you stand there stammering, with Larry's name momentarily escaping you, that you do know that his name is Larry, after all, though you don't think or believe that it is, right now? Hmm... that seems strange, too!

We could test ordinary folks' intuitions on this. Provide a scenario of the sort above, then ask one group of subjects whether the person "knows" the fact in question, another whether she "believes" it, and a third whether she "thinks" it. I'd predict considerably higher attribution of "knows" than "thinks".

Is there trouble here for the standard view in contemporary epistemology that (propositional) knowledge is a species of belief? Or are ordinary intuitions a poor guide? Or am I wrong about how the intuitions will fall out?

Maybe part of what's going on is that "think" is a bit more temporally narrow in its reference than "knows"?


Brad C said...

Why say that I believe or think that the bridge is closed as I drive towards it? I would say that this hinges on whether the relevant actions are intentional under the relevant descriptions.

Each of these is true of a relevant case:

(1) I driving towards the bridge.
(2) I am driving towards the bridge in order to get to work.
(3) I know* the bridge is out.
(4) I know* I cannot get to work by driving towards the bridge.

Now, this seems to be a plausible principle:

(IB) If A is doing X in order to Y, then A believes that, by doing X, A can Y (or make some headway towards that end.)

And it gives us

(2') I believe that, by driving towards the bridge, I can get to work (or make some headway towards that end.)

Roughly speaking: IB holds only when the 'in order to' entails that the person intentionally performs the action under the description "X, as means to Y".

I think the price of withholding the belief ascription is that one must also withhold the relevant intentional action ascription.

This suggests (to me) that we need to replace "habit takes over" with a more detailed phenomenological description before we can start running tests.

Eric Schwitzgebel said...

(1)-(4) all seem right to me; and if we accept IB, then we get knowledge without belief, which was what I'm shooting for right now, of course. So I'm sympathetic to your comment.

IB is appealing. But also I wonder if it's too intellectualist. So, for example, perhaps I am tilting my body in such-and-such a way in order to initiate a turn on my bicycle. Yet at the same time, I might have no knowledge that I am doing this or any ability to describe that as a procedure used in turning. It seems strange to my ear, in this case, to say that I believe that by moving my body that way I can turn the bike. (But maybe in some attenuated, implicit way I do believe it?)

Of course, IB might work in the habitual driving case even if it doesn't work in the bicycle case -- but then we need some principles governing its legitimate application.

(I agree, by the way, that before testing it would make sense to fill out a bit more of the phenomenological details of the case.)

Brad C said...
This comment has been removed by a blog administrator.
Brad C said...

A half-baked worry about the ordinary language uses of 'know' that are appealed to in order to generate this result.

My friend Phil is wracked with self-doubt, but I tell him to sumbit his article to Analysis anyway because I think his doubts are undue. He gets the acceptence letter a month later, and on hearing the good news I say "See: I knew it would get accepted!"

Possible diagnosis: we sometimes use 'know' and its cognates in order to underscore our belief that some doubts are undue or that some failures to believe are unfounded.

This seems to fit the case I describe and I worry it might fit the one with the bridge. When I say "I knew the bridge was out!" I might be expressing an attitude towards the failure of belief I was laboring under, not an assessment of my (excellent) cognitive relation to the fact that the bridge is out.

But that is not the type of use of 'know' and it cognates which philosophers have in mind when they say knowledge entails belief.

Justin Tiwald said...

Very interesting. I think Brad's diagnosis is clever, but somehow my intuitions don't match his. If I say "I knew the bridge was out" I'm reporting a cognitive state, not just expressing regret that I failed to connect the dots in some way.

I don't really know the relevant literature, but it seems like there might have been a similar sort of debate about this back in the 60's and 70's in response to Colin Radford. Radford offered the example of a test-taker who had long ago studied British history but is now convinced that he has forgotten it all. Upon taking the test, however, he discovers that he is able to accurately reproduce some of the important names and dates, such as the birth year of Queen Elizabeth. To the test-taker these answers have the phenomenological "feel" of guesses. He answers that Queen Elizabeth was born in 1603, but at the time of answering he doesn't really *believe* that Queen Elizabeth was born in 1603. The number just popped into his head or something. But much to his own surprise, he finds out that Queen Elizabeth really was born in 1603, and it seems unlikely that this was just an accident. If it wasn't an accident, then the best explanation seems to be that he really knew that Queen Elizabeth was born in 1603 all along, even as he didn't believe it.

Keith Lehrer has an interesting response to this: he thinks we can sideline it by calling it a "borderline case." That is, it looks about as much like genuine knowledge as not, just like certain objects can look as red as they look orange. But just as the existence of red-orange objects doesn't pose a problem for the concept of red, neither should these borderline cases of knowledge pose a problem for the concept of knowledge.

My own feeling about Lehrer's response, though, is that he's too quick to banish these sorts of cases to the borderlands. Most of my knowledge of second languages seems to me a lot like Eric's example of forgetting Larry's name. If someone asked me to give the Chinese for "bustle about" I would be likely to report that I don't know it. But when I'm really in the groove the Chinese phrase will pop right out of my mouth. These sorts of things happen quite often, and we usually say that they are instances of knowledge, not cases that look as much like knowledge as not.

Eric Schwitzgebel said...

I think my intuitions are closer to yours, Justin, than to Brad's in his second remarks, though the point he raises is an interesting one to consider.

The foreign language example is nice. There's some sense in which the bar is lower, I think, for "knowledge" than for "belief" -- and Radford's discussion (which I had forgotten about 'til now, though of course I knew it!) -- gets at that also. If the capacity to respond "knowingly" is there in you somewhere, then there's a tendency to say you know, despite not being able to come out with it; but it's less clear that we'd say you believe that fact or think that it is true, in those forgetful moments.

Of course -- and this may be a more general version of what Brad is pointing to above -- there may be pragmatic factors governing usage decisions here, as well as more purely semantic ones (if those can be separated).

Jonathan Jenkins Ichikawa said...

I'm not sure I see the problem here, Eric. We want to countenance my statement that "I knew the bridge was closed," because it's very natural to say that, and it seems right, and we want to respect the ordinary language usage, etc. But this is a past-tense claim: I knew that it was closed (earlier this morning). Then I forgot. I also believed it, before I forgot.

I don't think that in ordinary language it is obviously right to say that I knew it was closed during the time I was driving toward it.

I'm driving toward the bridge, and my passenger says, "you know the bridge is closed, right?" I don't think that "no, I'd forgotten!" is a terrible answer, and I can't think of any way to answer that question with an affirmative that sounds natural.

Eric Schwitzgebel said...

Thanks, Jonathan, that's helpful! If a passenger says, "You know the bridge is closed, right?", I agree that sometimes seems quite natural to say "no, I'd forgotten" (though it may also depend on how temporarily one has forgotten). But also it seems natural to say "Yeah, right, whoops -- I knew that!" My feeling is that the temporal reference of "knew" in such a claim includes the time while I was driving toward the bridge. How can we evaluate that? Well, if the forgetting wasn't very deep -- if it would have come to you independently in a moment, anyway -- then I think it's not very natural to say anything like "I used to know that, but I forgot", or "I knew that yesterday".

I think the matter might be clearer in the third-person case. Watching from a God's-eye view, as it were, will people still say of the driver that he knows the bridge is closed? I think most will. Will they say that he thinks the bridge is closed? I think most won't. Would you disagree with that?

Anonymous said...

I don't have stong intuitions about attributions of knowledge, belief, or thought in the bridge case. I don't feel compelled, in particular, to hang a lot on the after the fact self-attribution of knowledge ("I knew it"). I wonder what you think, Eric, of what looks to be a somewhat similar and potentially clearer case: tip-of-the-tongue phenomenon (TOT). In a case of TOT, say, Jones sees Smith at a conference and Smith's name is merely on the tip of Jones' tongue. Jones cannot call Smith's name to mind. What would you say, though, about Jones' knowledge, belief and/or thought re: Smith? Does Jones know, believe, and/or think that Smith's name is "Smith" during a TOT episode? I'd be inclined to answer "yes" to all three. Jones has the information and simply cannot access it (fully) consciously. Knowledge, belief, and thought attributions all strike me as ok ways of describing this info that is only partially accessible.

Eric Schwitzgebel said...

Hi Pete! For what it's worth, I'd treat the tip-of-the-tongue case like the bridge case: Ordinary intuition says you know his name is Larry (say). It's much less clear that you think his name is Larry or that you believe that it is.

I don't think ordinary intuition calls out strongly against attributing the belief either, though. Maybe this is just a case where we should regiment it for the sake of a clean theory?

On the other hand, if ordinary intuition is more liberal with knowledge than belief attributions in cases intuitively describable as "the info is buried in there somewhere, but just not accessible at this moment", then perhaps it has latched onto something important that we should attend to.

One asymmetry between "know" and "believe" in the tip-of-the-tongue case: I can say I know his name. I can't (grammatically) say I "believe" his name or "think" his name. There's a way in which this asymmetry supports my view, I think. But it also raises the question of how much differences in mere grammar are driving our intuitions.