Wolfgang Schwarz

Take this common bit of legal testimony: "I know p to a reasonable degree of [e.g., medical] certainty."

Here even a "reasonable degree of certainty" wouldn't entail certainty!

A fortiori, mere knowledge doesn't entail certainty. (Of this I'm reasonably certain.)

One interesting option in this neighborhood is to claim that both 'know' and 'certain' are context-sensitive, and that in every context, the standards that must be met in order to count as knowing or being certain are the same. Then one could claim that 'I know but I'm not certain' is false in every context of utterance, without denying fallibilism. It is after all perfectly appropriate to claim certainty in many contexts without being in a position to rule out that you are a BIV, etc.

It's interesting that Lewis denies that knowledge needs lots of confidence - I hadn't remembered that, so I'll need to go look at 'Elusive Knowledge' again.

It's interesting, I think, because it's a crucial premise in Williamson's anti-luminosity argument - the reliability premise is motivated by appeal to considerations about confidence, and it's hard to see how to motivate the reliability premise otherwise.

I am certain that I am lightyears behind in the knowledge debate, but doesn't it still hold that you can not know something false? If this is so, doesn't "know" have a connection to "is true"? Stress this relation and "I knew it but was not certain" is not as odd as when you stress the "holding true" relation.

Anyhow, a general remark: doesn#t this whole appraoch make you a little uneasy: What once was armchair conceptual analysis are now analysis better places in linguistic semantics?

M.

Aidan: Good point. If one allows for knowledge without belief one will have to reformulate the reliability premise in Williamson's luminosity argument. Maybe like this: if x knows p on the basis of evidence E, then there is no very similar situation where ~p but x still has evidence E.

Leo: Yes, certainty like knowledge doesn't require ruling out all alternatives. Nevertheless, I think attributions of certainty are not context-sensitive in the way attributions of knowledge are: even in epistemology class, we would say that Moore was certain that he had hands, although maybe he didn't really know it. I think the case of first person present tense is special. "I know but I'm not certain" really does sound odd in a way "I knew but wasn't certain" and "he knows but isn't certain" don't. (Though as Q notes, even the first person present tense version may be acceptable on special occasions.) I guess that is because if you're uncertain about p, you must be at least as uncertain about whether you know p; so claiming outright to know p while expressing uncertainty about p will sound odd.

M: Jason argued that knowledge with vocal stress doesn't entail truth. On your general point: I personally find analysing the folk concept of knowledge philosophically pointless anyway, so I'm happy to leave that job to linguistics. From time to time, it seems to me that somewhere near our folk concept lies a notion that can do interesting philosophical work. But it doesn't matter for that whether this notion perfectly fits the folk usage of "knowledge". Most of the time, however, I believe that we could just as well do epistemology and philosophy of mind without using the word "knowledge" at all.

There's a purely psychological sense of 'certain', and an epistemic sense, most clearly expressed in sentences of the form 'It is certain that P'. Your Moore example seems to employ the purely psychological sense of 'certain'. Obviously there are no epistemic standards that one must meet in order to be merely psychologically certain, which is why the sentence sounds fine. My suggestion was, or should have been, about the epistemic sense of 'certain'. (I also think the epistemic sense is often expressed in 'I am certain that P'. In third person attributions you have to say things like 'S achieved certainty' or 'S made herself certain' to express the epistemic sense.) Focusing on the epistemic sense of 'certain' seems in line with the original question of whether knowledge entails certainty. I take it that this question traditionally has not just been about whether knowledge requires some high degree of psychological confidence.

Incidentally, I think it quite plausible that whether it is certain that P, in the epistemic sense, does depend on the possibilities that are relevant in the context.

>Ich wusste, dass der Zug wieder Verspätung hat. Ich hätte den Vortrag zu Ende hören sollen.
>(I knew the train would be late. I should have stayed until the end of the lecture.)

I think that the German sentence you cited is one in which the verb 'wissen' (to know) is used in an improper way. What we want to say is that we have had the strong intuition that the train would be late. However, we would not say that we KNEW that the train would be late in the same sense as we sometimes say that we know that George W. Bush is president of the United States of America. It is just a rhetorical way of speaking.
Now if it is indeed nothing more than a rhetorical way of speaking, your example tells us nothing about knowlegde.

Wo,

Stress is highly context-sensitive; it may be that you are thinking of the German examples in a different context, which raises different alternatives to salience.

As far as contextualism goes: I was thinking exactly what Leo Iacono articulated. Contextualism can provide an elegant semantic explanation of the oddity of sentences such as "I know that p, but I'm not certain", without (as Lewis would say) endorsing fallibilism. Indeed, this contextualist solution is endorsed even by anti-contextualists such as Tim Williamson on p. 254 of Knowledge and Its Limits ("What seems to be at work here is a reluctance to allow the contextually set standards for knowledge and certainty to diverge.").

Furthermore, though Lewis doesn't use the term "certainty" in "Elusive Knowledge", I think he has it in mind. If a counterpossibility to p is genuinely relevant to me, I'm no longer going to be epistemically certain that p (again, as Leo Iacono points out). I think it's also going to be the case that the genuine salience of counterpossibilities undermines subjective confidence as well (one might say that such possibilities aren't genuinely salient unless they are taken on board subjectively as well as epistemically).

Wo,

It might be, as Tim Williamson pointed out in the question session, that the use of "just" in my original examples is doing some work. If you add a similar word, such as "einfach", to your German examples, you may find them easier:

(1) Ich habe einfach *gewusst*, dass du mich hintergehst. Bin ich froh, dass ich mich irrte!

Jason, I don't think that the word 'einfach' will do any work in sentences like (1). It seems to me, that in German, you can stress 'know' only in the case all the things you *knew* really happened.

Vanessa,

That's interesting. One certainly wouldn't expect this to be a language specific issue, having to do with the semantic content of "know" vs. "wissen". Rather, one would expect it to be something pragmatic in nature, and so cross-linguistic...

Ah right, I always thought of "certain" in the psychological sense. If "certain" really is ambiguous, it seems that using linguistic data about what sounds odd to figure out whether "knowing" entails "being (epistemically) certain" becomes tricky, unless there is some way to rule out the psychological sense.

At any rate, I don't think contextualism by itself offers a good explanation of why "I know that p, but I'm not certain" sounds odd, given that "he knows that p, but he isn't certain" does not sound odd; so I wouldn't think this provides a particularly good motivation for contextualism.

I agree with Vanessa that (1) sounds just as bad with "einfach" included. I don't know why this is so. One can make a strong case that "Wissen" is often used for "true belief" in German. (Gerhard Ernst argued for this in _Das Problem des Wissens_.) Maybe in English, but not in German, "knowledge" can also be used for "strong belief". (FWIW, Google has 71'400 results for "false knowledge" and only 945 for "falsches Wissen", with "knowledge" vs. "Wissen" being 710'000'000 vs. 230'000'000.)

Vanessa, I agree that the uses of "wissen" I quoted are improper in the sense that one could respond with "naja, wirklich gewusst hast du es nicht" ("well, you didn't really know"). One could also reply like that to "I just knew you cheated on me...". I just wanted to point out that people do use "wissen" like that in German, whereas they don't use it for merely strong belief.