Skip to main content.
February 9th, 2007

Knowledge as the Most General FMSO

Carl Ginet is running an excellent seminar here at Cornell on Timothy Williamson’s Knowledge and Its Limits. Here is a point that David Liebesman and I were pushing a couple of weeks ago against Williamson’s idea that knowledge is the most general factive mental state.

Imagine we discovered a community that had a word “schnow”, which is like our “know”. Their view about schnowledge is quite like our view about knowledge. For instance, they unhesitatingly say that Gettier cases are not cases of schnowledge. But they hold that there are fewer defeaters for schnowledge than we think there are for knowledge. For instance, in case like Harman’s example of the person who happens to not see the misleading newspapers saying the dictator is still alive, they will say that the person schnows that the dictator is dead. In general, it turns out, they don’t think that unobtained misleading evidence defeats schnowledge. They are, however, future externalists in the following sense. The fact that someone will obtain misleading evidence may defeat current schnowledge, though it doesn’t defeat current justification. I’m going to assume (perhaps wrongly!) that their view on schnowledge is strictly weaker than our view on knowledge, since we allow never unobtained misleading evidence to defeat knowledge, but strictly stronger than our view (and theirs) on justified true belief.

Now here are two questions for Williamson.

First, is schnowing that p a mental state? I can’t see anything in the arguments for knowledge being a mental state that would count against schnowledge being a mental state. Note, in particular, that it isn’t (easily) factorisable.

Second, is schowledge weaker than knowledge? That is, do they denote a weaker relation by ‘schnows’ than we denote by ‘knows’? I can see going either way here. On the one hand, they do use ‘schnows’ in a slightly different way to how we use ‘knows’. On the other, when it comes to normative terms, we are generally quite generous about allowing that people with different usage nevertheless have the same meaning. When Osama says, for instance, “Killing Christians is good”, he is falsely saying something using our common concept of goodness, not truly saying something using a different concept of goodness. Perhaps the people in question are just misusing ‘schnows’, or perhaps we are misusing ‘knows’.

But I think there is a problem for Williamson on either answer he gives to this question. If schnowledge is weaker than knowledge, then knowledge is not the most general factive mental state, because schnowledge is more general. If schnowledge is the same as knowledge, then it turns out our term ‘knows’ is not plastic. Small deviations, even large deviations, don’t produce a difference in denotation. But in Vagueness, his view was that vagueness in language is grounded in semantic plasticity. And it would be intolerable to say that ‘knows’ is not vague. So I don’t see a way to hold on both to the view that knowledge is the most general FMS, and the view that vagueness is a product of semantic plasticity.

Posted by Brian Weatherson in Uncategorized

12 Comments »

This entry was posted on Friday, February 9th, 2007 at 7:54 pm and is filed under Uncategorized. You can follow any responses to this entry through the comments RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

12 Responses to “Knowledge as the Most General FMSO”

  1. Baron says:

    Hi Brian,

    I actually made a somewhat similar argument in a paper called “Accidentally Factive Mental States” (published in PPR in 2005). The idea is that S gnows that p when S is in a Gettier-type case with respect to p. Obviously, this state is factive but not cognitive. So, ‘knows’ cannot be the most general factive mental state operator, where this would mean that every other mental state operator appears in sentences that entail sentences with ‘knows’. Moreover, being a factive mental state operator does not (by itself) explain why we make ‘knows’ assertions.

    The main difficulty is defending the claim that gnowledge is a genuine mental state (or, at least, that it has as much of a claim to being a genuine mental state as knowledge does). For example, Williamson thinks that knowledge plays an ineliminable role in action explanations. My claim is that an equally good case can be made for the claim that gnowledge does as well.

  2. joesalerno says:

    Baron,

    I haven’t yet seen your paper, but defending the claim that ‘gnowledge’ is a mental state operator is the difficulty. To be such a thing, it must attribute a propositional attitude (minimally, attribute understanding that p) and be unanalyzable. ‘gnowledge’ clearly attributes an attitude, since among other things it attributes belief. But it doesn’t seem to be unanalyzable. You stipulated that “S gnows that p when S is in a Gettier-type case with respect to p.” Isn’t that to say that S has a justified true belief based on a false lemma? But then we have the analysis of ‘gnows’.

    And so, gnows is not a genuine mental state because, strictly speaking, only a proper part of gnowledge is mental.

  3. Baron says:

    Joe,

    Actually, it isn’t all that difficult. Here are a couple of reasons why.

    First, suppose that gnowledge really could be analyzed as justified true belief based on a false lemma. We would then have an analysis of what makes for a Gettier-type case. That, in turn, would allow us to have an analysis of knowledge: it is justified true belief that is not based on a false lemma. But, if that really is a satisfying analysis of knowledge, Williamson’s epistemology is mistaken. At the very least, the motivation he cites for it in the introduction of KNOWLEDGE AND ITS LIMITS—viz., that we are very unlikely to ever have a satisfying analysis of knowledge—would be lost.

    Second, and more important, gnowledge cannot be satisfactorily analyzed as justified true belief based on a false lemma. As Richard Feldman showed in “An Alleged Defect in Gettier Counter-Examples” (Australasian Journal of Philosophy, 1974), it is easy to construct Gettier-type cases where the accidentally true belief is not based on a false lemma. Moreover, the barn facade case from Alvin Goldman’s “Discrimination and Perceptual Knowledge” involves an accidentally true belief where no false lemma plays a role.

    As far as I know, there isn’t a satisfactory account of what a Gettier-type case might be. We often think of them as being cases of justified but accidentally true belief, but we do not have a good account of what it is to be accidentally true in the relevant sense. (Again, if we did, we could analyze knowledge as nonaccidentally true belief. Unger proposed something like this, but, as far as I can tell, no one has taken this to be very satisfying.)

    As I say in my paper, we have come to understand the description, “Gettier-type case,” largely through pointing to paradigm cases—just as Williamson says we have come to understand “knowledge.” (Something similar could be said about “gnowledge,” if we chose to begin using that term.)

    Williamson’s case for treating knowledge as a mental state rests largely on our inability to provide a satisfactory analysis of it. He also argues that, because we cannot analyze knowledge, knowledge plays an ineliminable role in action explanation. My argument is that we are just as incapable of analyzing gnowledge, and thus we have just as much reason to think gnowledge plays an ineliminable role in action explanation. So, whatever reason we have for thinking that knowledge is a mental state separate from belief, we have just as much reason to say that gnowledge is a mental state separate from belief.

  4. Matt Weiner says:

    we have come to understand the description, “Gettier-type case,” largely through pointing to paradigm cases—just as Williamson says we have come to understand “knowledge.”

    This seems doubtful to me. Don’t we understand Gettier-type cases as cases of justified true belief that are not knowledge — or perhaps as cases of internalistically justified true belief that are not knowledge, or of prima facie JTBs that are not prima facie knowledge (to accommodate those like Brian who think that they are knowledge)? I doubt that there is anyone who’s come to learn what Gettier-type cases are by having the barn case explained to them, and having the newspaper cases explained to them, and having the original Gettier case explained to them, without having it explained that these are all supposed to be JTBs that don’t amount to knowledge.

    This matters because in note 15 of your paper you acknowledge that we could give an account of gnowledge as “S gnows that p iff S has a justified
    true belief that p that is not knowledge.” Yet, as you point out, we could also give an account of knowledge as “S knows that p iff S has a justified
    true belief that p that is not gnowledge.” You argue that there is no reason to take one account as more basic than the other. But if we really learn what knowledge is through paradigm examples, and only learn what gnowledge is through an explanation that involves gnowledge, then there is reason to take the account of ‘gnows’ as more basic; and thus reason to think that gnowledge, and not knowledge, decomposes into a mental state and a non-mental state.

    It may be that you can argue that our way of learning what ‘gnowledge’ and ‘knowledge’ are is too contingent to support any deep claims about which is a mental state and which not. It may also be that I haven’t placed enough weight on the last sentence of footnote 15, which says that in any case the above analysis of ‘gnows’ is too uninformative to prove that gnowledge isn’t a mental state.

  5. Baron says:

    I think it actually is likely that we have acquired the term ‘Gettier-type case’ through pointing at paradigm cases. Imagine, prior to Gettier’s paper, that someone had tried to introduce a new term covering instances of justified true beliefs that are nevertheless not knowledge. I suspect philosophers would have had a hard time understanding exactly what, if anything, the term was supposed to be used for. Moreover, what has driven our understanding of the term has been the increasing number of cases that seem to have some interesting common feature. Yes, we often describe them as being instances of justified true belief that is not knowledge. But we also describe instances of knowledge as being instances of justified true belief that are not Gettier cases. In both cases, we clarify our grasp of the paradigm cases by saying what they are not.

    In any case, I agree with what you say in your final paragraph. Suppose that no one actually has acquired the term ‘Gettier-type case’ through looking at paradigm cases. In the paper, my point is merely that this could happen. We could introduce a term, ‘gnowledge’, by pointing at Gettier’s original cases, Lehrer’s cases, the barn facade case, etc. That should give us reason enough to reject Williamson’s account. It would, after all, be very strange to think that we could get any sort of insight into what knowledge is by pointing to the fact that ‘knows’ happens to be the most general factive mental state operator. To put the point another way, what makes knowledge interesting is not just that it is a factive mental state; what matters is the way in which it is factive.

    (Side note: we should not understand Gettier-type cases as being internalistically justified true beliefs that are not knowledge if in doing so we think that externalistically justified true beliefs do not face the Gettier problem. The barn facade case (and others, like the Gettier-type cases raised for Plantinga’s proper function view) show that the Gettier problem cuts across the internal/external divide.)

  6. Matt Weiner says:

    Imagine, prior to Gettier’s paper, that someone had tried to introduce a new term covering instances of justified true beliefs that are nevertheless not knowledge. I suspect philosophers would have had a hard time understanding exactly what, if anything, the term was supposed to be used for.

    I suspect that they would have had trouble understanding what it would’ve been used for because it would’ve seemed like an empty term. People might have reacted similarly to someone talking about “necessary a posteriori” before exposure to Kripke’s examples, but that doesn’t show that “necessary a posteriori” isn’t analyzable. I take the rest of your points, though.

  7. joesalerno says:

    Baron,

    Thanks for your replies. You say,

    “Suppose that gnowledge really could be analyzed as justified true belief based on a false lemma. We would then have an analysis of what makes for a Gettier-type case. That, in turn, would allow us to have an analysis of knowledge: it is justified true belief that is not based on a false lemma.”

    But that’s not right. It doesn’t follow from having an analysis of Gettier cases, that we have an analysis of knowledge. That’s because the problems with analyzing knowledge are not limited solely to Gettier examples. Barn cases for instance, WHICH ARE NOT “GETTIER” CASES, pose a problem for analyzing knowledge. The No-False Lemmas approach doesn’t help us here.

    So by “Gettier case” or “gnowledge” you must have something else clearly in mind, like “any justified true belief that is not knowledge”. But this is analyzable into justification, truth, belief, knowledge and negation. Importantly, at least one component of gnowledge is then not mental—-viz., truth. So I’m not seeing the counterexample to Williamson’s claim: ‘s Os that p’ entails ‘s knows that p’, where ‘O’ is an FMSO.

  8. joesalerno says:

    Brian,

    How about this? ‘Schnowledge’ is not a FMSO, because ‘Schnowledge’ is synonymous with a complex expression whose meaning is composed of the meanings of its parts—-viz., the expression ‘knowledge or something that is the result of adding an unobtained misleading defeater to what would otherwise be knowledge’. ‘Schnows’ is factorisable.

  9. Kenny Easwaran says:

    But that factorization goes the other direction too. Knowledge is just “schnowledge with no unobtained misleading defeater.” In fact, this analysis is more clearly a correct analysis than the analysis in the other direction, because it’s just a simple conjunction, rather than a disjunction involving a counterfactual.

  10. Baron says:

    Joe,

    In my paper, I’m careful to talk about Gettier-*type* cases, rather than merely Gettier cases. I think there is a good argument for treating the barn facade cases as Gettier-type cases, even though they are somewhat different from Gettier’s original cases.

    But, in any case, it is unnecessary to insist on that point. All that matters is that we can introduce a term, ‘gnowledge’, by pointing at paradigm cases: Gettier’s original cases, Lehrer’s cases, Chisholm’s sheep in the meadow, the barn facade case, etc. Introduced in this way, there is no reason to think that the purported mental state picked out by ‘gnowledge’ factors into justification, truth, knowledge, belief, and negation. At the very least, there is no more reason to think that it factors than there is to think that knowledge factors into justification, truth, gnowledge, belief, and negation. You may not like the case for gnowledge, but for exactly the same reason you should not like Williamson’s case with respect to knowledge, either.

    The same goes for Brian’s ‘schnowledge’. I take it Williamson would not accept the principle that a term that is synonymous with a semantically composite expression is itself semantically composite. If he did, his opponents could scotch his project simply by introducing a semantically composite synonym for knowledge.

  11. christian says:

    Hi Brian,

    One thing that troubles me about Williamson’s claim that knowing that p is the most general FMSO is that it seems to me that being aware that p is the most general FMSO.

    One cannot know that p without being aware that p and knowing that p is, arguably, a way of being aware that p in the same way that seeing that p is a way of knowing that p. Awareness seems more general anyway to me, but weaker still, we should say Williamson has “shown” only that either awareness that p or knowing that p is the most general FMSO.

  12. joesalerno says:

    Hey Kenny,

    Thanks for this. I don’t recall Williamson worrying about the degree of complexity of an analysis. If there is an analysis (of the relevant kind) of knowledge, then that will be bad enough for his project. As for your analysis of knowledge in terms of Schnowledge, it must be that Williamson is more worried about non-circular analyses. Schnowledge was introduced by Brian in terms of knowledge, and then you analyzed knowledge in terms of it. So it is useless for establishing that schnowledge is conceptually prior to knowledge.

Leave a Reply

You must be logged in to post a comment.