September 12, 2004


In defending the thesis that knowledge is just justified true belief (a thesis I'm more sympathetic to than most), Brian argues that

The people who have Gettier intuitions are (on the whole) basically sceptics who have (perhaps) talked themselves out of their most sceptical intuitions but not this one.

His third (out of two!) argument for this is that Gettier intuitions are made much more prominent by utterances that stress the word "know." I need to preface my remarks about this by saying two things:

a. There's been a lot of work done on stress and focus.
b. I don't know any of it, so the following speculations aren't as well-informed as they might be.

(Also: Close readers may think that my examples contain an undertone of mockery of a certain political figure. I probably believe every political statement that is suggested in these examples, but I don't want to open the floor for political discussion here, and I couldn't think of better examples on the fly. The conversations are all imaginary, and you can treat the underlying experiments as imaginary as well; it shouldn't affect the philosophical point.)

Brian's idea, presumably, is that when we say things like "Henry KNOWS that he's looking at a barn" or "Does Ringo KNOW that the tour starts in New Zealand," the stress on 'know' makes certain doubts more salient--including Gettier doubts. Hence it makes us reluctant to assent to the stressed knowledge-ascription in cases of Gettierized JTBs. But we would assent to ordinary, unstressed knowledge-ascriptions in these cases, when the Gettier doubts aren't salient.

In support of this, Brian adduces examples in which the stress falls on something other than 'know' and we are willing to ascribe knowledge in Gettier cases. I don't think this provides evidence that stress is keeping us from putting aside Gettier worries, because we're sometimes willing to ascribe knowledge in similar cases even when we do stress 'know'.

The idea in part is that negative knowledge-that ascriptions usually focus on whether the ascribee has a certain (acknowledged true) belief. After the jump, there's some theses that are meant to support it, although they're not exactly argued for. (The numbering is odd because I moved that section after writing the post). But here's the key example.

Let's suppose that Morgan and Jordan are discussing a political figure called Blotus, and that Morgan thinks Blotus' policy toward Belize has failed. She might say:

(8) Blotus is going to stick to the old Belize policy. He doesn't know that his Belize policy has failed.

The most natural reading is that Blotus doesn't even think that his Belize policy has failed. If Jordan discovers that Blotus has an unjustified true belief that his Belize policy has failed, he will be unimpressed with what Morgan says.--Of course (8) is deceptive if Blotus has any sort of belief that his policy has failed, because then his lack of knowledge won't explain why he would stick to the old Belize policy.
But now let's suppose that, discourse-initially, Morgan leans hard on 'know':
(9) Blotus doesn't KNOW that his Belize policy has failed.

The most natural reading of (9), I think, is that Blotus' policy has failed but Blotus does not believe this. The heavy stress on 'know' sets up a contrast with "Blotus' policy has failed." Only if the discourse has set up another contrast--"Blotus thinks this, but he doesn't KNOW it"--will (9) suggest something other than lack of belief.

Similarly with a question--let's suppose Jordan asks

(10) Does Blotus KNOW that his Belize policy has failed?

This suggests that it's not in question at all whether Blotus' policy has failed; the question is whether Blotus thinks so. (I don't know how Howard Baker accented his famous question, but even if he said "What did the president KNOW and when did he KNOW it?" it wouldn't have exculpated Nixon if it had been shown that he had a bunch of true unjustified beliefs.)

Now, let's suppose that Blotus has a Gettierized belief that his Belize policy has failed. Here's the story. Blotus' Belize policy was formed by a group of advisors called the Nominalists--but lately Blotus has been getting most of his information from the Nominalists' enemies, the Realists. The Realists would tell him that any Nominalist policy was failing, no matter whether it was or not.
If this is the story, does (9) sound true? Not to me; and the answer to (10) should be "yes." No matter how heavily 'know' is stressed, the effect is merely to emphasize that the question at issue is not whether Blotus' policy has failed, but whether Blotus thinks his policy has failed. Stressing 'know' doesn't make Gettier-style issues any more salient here.
In fact, I think old-school skeptical worries come to the fore first. If Jordan and Morgan both presuppose that Blotus has true belief that his policy has failed, and Morgan keeps emphasizing (9), then I think Jordan will conclude that Morgan is saying that Blotus' evidence is somehow not good enough for KNOWLEDGE--that in the hurly-burly of intelligence-gathering and the vicissitudes of the future, the evidence Blotus has just isn't good enough to shut the door on the Belize policy and call it a definite failure. That's an unnatural reading, but I don't think it's nearly as unnatural as a reading on which Blotus' true belief fails to count as knowledge for Gettier-style reasons. Those wouldn't be made salient by stress on 'know' unless Jordan is a professional epistemologist.

So, in sum, I don't think Gettier intuitions can be explained away just by saying that they're skeptical doubts that are brought to the fore by stress on the word 'know'. Stressing 'know' doesn't seem to bring those doubts to the fore. I do have something of a theory of why we have Gettier intuitions--and it's not a theory that makes the Gettier problem particularly important to solve--but that'll be for another post.

The Theses
These are things I think but don't know, because I don't know how I'd check them easily. (Even if I were on the Internet where I'm typing, Google wouldn't be much help, because what I call knowledge-that ascriptions often don't contain the word "that"; I think it was John Carroll who pointed this out, in comments on Jonathan Schaffer's paper at the 2004 INPC.)

A. It's more common to say "I don't know-wh" than "I don't know-that." "I don't know what the combination is" makes perfect sense; "I don't know that the combination is 24-7-35" sounds awfully indiscreet.

B. "I don't know-that," though unusual, can be comfortably if snottily used to disclaim belief:

(1a) Jordan: Do you know that the Eskimos have over 365 words for snow?
Morgan: No, I don't know that, because Geoff Pullum has shown that that claim is nonsense.

(1b) Jordan: Do you know that the Eskimos have over 365 words for snow?

Morgan: No, I don't know that; I don't even believe it [this is perhaps philosophically infected].

It can also--so I claim--be used to signal that you're sticking your neck out when you make an assertion. In this use the suggestion is that your grounds for assertion aren't conclusive enough to allow you to claim knowledge:

(2) J: The Pirates are going to win.
M: How do you know?
J: I don't know that they're going to win, but that's my prediction.

Stress on "I" is weird, except where you're conveying that you're the only one around who doesn't believe it.

(3) J: Everyone here knows that the Eskimos have 365 words for snow.
M: I don't know it, because I know that that claim's an urban legend.

(4) J: Everyone here knows that the next meeting is Tuesday.

M: *I don't know it, but Terry does.

[If you hold the knowledge account of assertion, it's pretty obvious why (4) is wrong--since Morgan asserts that Terry knows it, Morgan should know that Terry knows it, and thus should know it herself. To explain why I, who reject the knowledge account of assertion, think (4) can't be right would take us too far afield.]

C. Third-person or past-tense negative knowledge ascriptions and knowledge questions most commonly presuppose factivity and deny belief. That is, if I say "Does she know that p?" or "I didn't know that p," that p will usually be a presupposition (possibly added to the scoreboard by this very utterance), and the salient question will be whether she believes (truly) that p, and the salient claim that I didn't believe that p.


(5) Does Terry know that the next meeting is Tuesday?

will usually sound odd if the next meeting isn't Tuesday, and

(6) ??I didn't know that Milwaukee is on Lake Ontario

sounds very odd, given that Milwaukee isn't on Lake Ontario (it's on Lake Michigan).

There's an obvious Gricean explanation for (6)--"Milwaukee isn't on Lake Ontario" entails (6) and so is more informative. Let's take a case where p's truth is unsettled. Jordan and Morgan are e-mailing about where Morgan will be next year, and whether Terry (who rents apartments in Milwaukee) has tried to sell Morgan on an apartment. Morgan writes:

(7) Terry doesn't know that I will be in Milwaukee next year.

From this Jordan would be justified in inferring that Morgan will be in Milwaukee next year. If Morgan hasn't heard yet about the pending job in Milwaukee, Jordan will be rightfully resentful. But Morgan couldn't be more informative by saying "I won't be in Milwaukee next year," because she doesn't believe that. (Perhaps "I don't know whether I'll be in Milwaukee" is the appropriate more informative locution.)

In this case, stress can make a difference; if Morgan says

(7a) Terry doesn't know that I will be in Milwaukee next year

the implication is very strong that she will be in Milwaukee next year--the implied contrast is that someone does or might know this. Similarly if "will" is stressed. Now
(7b) Terry doesn't know that I will be in Milwaukee next year

sounds to me like the most natural way of saying (7) aloud, and has the same presuppositions as (7). But if Morgan really leans on 'know':
(7c) Terry doesn't KNOW that I will be in Milwaukee next year

then the implied contrast may be that Terry thinks Morgan will be in Milwaukee next year. I can't quite tell whether the most natural reading of (7c) in isolation is one that presupposes that Morgan will be in Milwaukee next year; but in any case, my guess is that (7c) will be much rarer than (7a) and (7b). This case, however, doesn't tell against Brian's analysis, since the increased stress in (7c) really can change the reading of the sentence.

Posted by Matt Weiner at September 12, 2004 08:52 AM