Gettier Cases and Internalism

Brian Weatherson had a couple of posts critical of Williamson last week at his blog. While I’m sympathetic to what I think Brian’s general point is, I’m not sure I agree with the claims in this post. The dispute is a question more of what’s accepted than what’s right, so I’d like to canvas your opinions on the question.

Brian notes that Williamson argues that Gettier refuted the justified true belief analysis of knowledge, and that he also claims that “a flat-out belief is fully justified if and only if it constitutes knowledge.” But if the second is true, justified true belief is knowledge (and vice versa). Brian continues:

From the article it looks like the way to resolve the apparent contradiction is that Williamson thinks that the Gettier cases only work if we interpret ‘justified’ as ‘justified by the best version of internalist epistemology’. He doesn’t think that there are counterexamples to an externalist version of JTB. I don’t think this is particularly plausible. The intuitions supporting Gettier cases don’t turn on whether we’re internalists or externalists about justification.

I ask: Do you agree with the last sentence? I don’t, because of these historical points:

(1) Gettier’s own cases were, I take it, directed at internalist versions of JTB–is this generally accepted?
(2) Gettier cases have been used to support externalist versions of JTB; for instance, as I read Goldman, he uses the Ginet fake barn cases to support reliabilist JTB (the theory of knowledge he puts forth in “Discrimination and Perceptual Knowledge” is a JTB theory on the theory of justification he puts forth in “What is Justified Belief?”–if I’m remembering correctly.)

Some externalist versions of JTB are vulnerable to Gettier counterexamples, but it’s not clear to me that every externalist version of JTB is vulnerable. But I’m curious if others share my sense that Gettier cases are particularly suited to internalist theories.


Comments

Gettier Cases and Internalism — 14 Comments

  1. Sorry, Matt: I’m going to agree with Brian that “the intuitions supporting Gettier cases donâ��t turn on whether weâ��re internalists or externalists about justification.” Moreover, I think Williamson agrees with this too.

    As Linda Zagzebski and Trenton Merricks, apparently working independently, have both shown, warrant (i.e., that which must be added to true belief to get knowledge) must itself entail truth, since otherwise Gettier cases can still arise. So all that is needed to get Gettier examples going is the point that justification doesn’t entail truth — something that many externalists accept as well as internalists.

    I’m pretty sure that Williamson accepts this. Roughly, I think Williamson’s view is that the extent to which a belief is justified is directly proportional to the conditional probability of the belief given one’s evidence. Given his principle E = K, one’s evidence consists of all and only those propositions that one knows; so the propositions of which one’s evidence consists are all true.

    I think that by a “fully justified” belief, Williamson means a maximally justified belief — that is, a belief that has conditional probability 1 given one’s evidence. So such a belief must be a logical consequence of one’s evidence, and so must also be true. No Gettier cases will arise here, then, since “full justification” entails truth.

    On the other hand, it is possible that a belief that is justified to any less-than-maximal extent could be false. So there could still be a case where a belief is justified to that extent, and true, but it is an “accident” that it is both justified to that extent and true. But that would be a Gettier case for Williamson. So even on Williamson’s conception of justification, JTB only coincides with knowledge if by justification we mean “full justification”.

    Of course, Williamson will deny that when he identifies knowledge with fully justified belief, he is giving an analysis of knowledge: any such analysis would be completely circular, since the notion of knowledge figures rather crucially in his analysis of justification!

  2. Ralph, I agree with your assessment here that the crucial issue is whether the condition besides true belief is sufficient for truth on its own. (And I think the first person to point this out was Scott Sturgeon, in 93 in Analysis, I think). I have a minor quibble with the following, though:

    I think that by a â??fully justifiedâ?? belief, Williamson means a maximally justified belief â?? that is, a belief that has conditional probability 1 given oneâ??s evidence. So such a belief must be a logical consequence of oneâ??s evidence, and so must also be true. No Gettier cases will arise here, then, since â??full justificationâ?? entails truth.

    I don’t think it follows that p is a logical consequence of e just because the conditional probability of e on p is 1. At least, it doesn’t on the standard accounts of probability that leave it possible that things happen that have a zero probability of happening.

  3. Matt, as for (1) I think youâ��re right. If I remember the paper correctly, Gettier directs his criticism at something like “various attempts made in recent years” and refers to Ayer’s The Problem of Knowledge and Chisholm’s Perceiving, which are both internalist theories. He also mentions Plato, I think, but it’s fairly controversial whether Plato was engaged in the same kind of task as epistemologists like Ayer and Chisholm, i.e., whether he was in fact trying to analyze propositional knowledge. (I think Shope discusses this in his book on the Gettier debate.)

    As for (2), I think Goldman can be said to defend a (non-traditional) JTB analysis in “What is Justified Belief”–after all he starts out by saying that the “aim of the paper is to sketch a theory of justified belief.” In the same paper (in the Liaisons volume), however, he says that he, in “Discrimination and Perceptual Knowledge,” “denied that justification is necessary for knowing,” but also that he by “justification” there meant “‘Cartesian’ accounts of justification.”

    Personally, I’m at least inclined to say that internalist accounts are particularly vulnerable to Gettier cases, but I’m not sure whether the intuitions supporting Gettier cases turn on whether we’re internalists or externalists. But doesn’t it suffice to say what you seem to say, namely that while most internalists accounts (maybe even all presented so far) are vulnerable, merely some externalist versions of JTB are?

  4. “As Linda Zagzebski and Trenton Merricks, apparently working independently, have both shown, warrant (i.e., that which must be added to true belief to get knowledge) must itself entail truth, since otherwise Gettier cases can still arise.”

    I’m not sure about the “shown” part. If I remember correctly, a principle that has to be true for the argument that warrant entails truth to go trough is that if a false belief is warranted and possibly true, then it is possible for it to be warranted and true (Merricks explicitly appeals to something like that). This, of course, is not a logical truth, and I don’t see why we have to expect it to be true–as the variety of Gettier cases show, warrant behaves in strange ways.

  5. Lot of good points! I’ll say the important thing first, and then some other things.

    The important thing: Even given that Williamson does think that counterexamples can be found to any JTB analysis of knowledge where justification is non-factive, he still doesn’t contradict himself. Brian saw a contradiction between using Gettier cases as counterexamples to the JTB analysis and claiming that justification is knowledge. But if we read ‘justification’ in the JTB analysis as ‘internalist justification’, or even ‘non-factive internalist or externalist justification’, there’s no contradiction if we read ‘justification’ in ‘justification is knowledge’ as ‘Williamson’s notion of justification’.

    Here I like the point Jon made in the post above about reasonable vs. permissible belief; when Williamson counterexamples JTB analyses he’s reading justification as reasonable belief, when he says justification is knowledge he’s reading justification as permissible belief.

    Now the other stuff: Ralph is right that Williamson probably does agree that any JTB account where justification is externalist but non-factive will be vulnerable to Gettier-style counterexamples. But this doesn’t mean that it’s apt to say that the intuitions about Gettier cases are independent of the internalism/externalism debate. Actually–I just thought of this, and perhaps should have put it in the main post–our intuitions won’t be affected, because our intuitions are of the form “Henry doesn’t know there’s a barn there,” and don’t refer to internalism or externalism at all. What might be affected is whether the case that we intuitively judge not to be knowledge is a counterexample to the JTB theory. On Goldman’s theory of justification, we still intuitively judge Henry to lack knowledge, but Henry also lacks justification so it’s not a counterexample to JTB.

    That is to say–internalism/externalism affects whether particular Gettier cases are counterexamples to JTB. The arguments against non-factive accounts of warrant aim to show that every JTB account with non-factive justification is counterexampled by some Gettier case. And, like Juan, I’m a little skeptical about this. I don’t know the Merricks/Zagzebski arguments, but Edward Craig has a similar argument based on the idea that a true belief could always be accidentally warranted if warrant is non-factive. Here it’s not clear to me that the accidentality of warrant always destroys knowledge–we may have to examine cases.

  6. I was unable to get the Scott Sturgeon article that Jon mentioned without actually getting out of my chair, and I should probably read it before making this comment, but I’m going to be foolhardy.

    Ralph describes (without endorsing) the thesis that warrant (i.e., that which must be added to true belief to get knowledge) must itself entail truth, attributing it to Zagzebski and Merricks. Jon says, Ralph, I agree with your assessment here that the crucial issue is whether the condition besides true belief is sufficient for truth on its own.
    It seems pretty trivial that warrant (as defined by Ralph) does not entail truth. The proposition that p is warranted must be equivalent to the proposition p â�� p is known (that’s supposed to be a material conditional; I hope that’s what it looks like).
    That conditional doesn’t entail p.

  7. Jaime, I think you are right. What people saying things like “warrant entails truth” mean, I gather, is that the proposition that a belief is warranted plus other propositions regarding warrant, that they take as something like analytic, entail that the belief is true. As I said above, it is not clear to me that some of the principles involved are true, let alone something like analytic.

  8. I’m unconvinced by Matt’s attempt to get TW out of the contradiction. If what we need to say is something like “by ‘justified’ he didn’t mean ‘justified’ he meant ‘non-factively justified’ or something similar” then I think we’re agreeing that he contradicted himself, and finding something similar he could have said that wasn’t contradictory.

    I think what’s really interesting about the JTB theory is the idea that we can usefully think of knowledge as having a normative component, an alethic component and a psychological component. Whether you call the normative component justification, warrant, reasonableness, permissibility or just the normative component of knowledge doesn’t really affect what look to me like the philosophically interesting questions. Maybe I’m just being idiosyncratic. But I think we should distinguish the JTB theory from the theories of justification that its proponents held. (Comparison – it isn’t a deep objection to Hempel’s DN theory of explanation that we find some explanations whose component laws aren’t really laws on Hempel’s theory of lawhood. The DN theory fails iff it can’t work with the *right* theory of laws, if it can’t work with *Hempel’s* theory of laws. Of course the DN theory of explanation is false, so maybe that’s not the best analogy…)

    Mike Huemer has a paper making the point Jamie raises. I think the Merricks/Zagzebski line works provided you are some kind of realist about warrant. That is, provided you think warrant is some non-disjunctive, independently valuable property, the M/Z argument looks good. More precisely, you need warrant to be closed under known entailment (or at least known and competently executed entailment) but I think that follows from the claims about value.

    Finally, I think there are some methodological issues here that should be of general interest. TW says that Gettier cases can simply be *seen* to be instances of non-knowledge, but then turns to theory rather than intuition or insight to say whether they are cases of justified belief. (And proceeds to mock to some extent those who think otherwise. The mocking is mostly deserved though – at least in my case.) I go the other way around, I think we can simply *see* that they are justified beliefs, and then let my theory decide whether they count as knowledge or not. There doesn’t appear to be a methodological difference here, unless there is some way to argue that insight about knowledge is privileged over insight about justification. Maybe that’s what it is to be a knowledge first guy is to think that.

  9. OK, Jamie, let me put the point a bit less sloppily: Any analysis of knowledge according to which the state of knowing p is equivalent to the conjunction of believing p, p’s being true, and some other element X will be vulnerable to Gettier counterexamples if X does not itself entail that p is true.

    The use of the term “entail” here doesn’t mean logical or even analytic entailment (as Juan suggests). It only means that it is metaphysically impossible for X to hold without p’s being true.

    Thank you so much for pointing out that Scott Sturgeon was the first to make this point. Scott is a friend of mine, so I’m shocked that I didn’t know that!

    I’m willing to say that Sturgeon, Zagzebski and Merricks have “shown” this. Certainly they haven’t given an apodictic proof, but I do think that they have made the point extremely plausible. The examples that they cite make it highly plausible that if the factor X that a theory appeals to doesn’t entail truth (and isn’t gruesomely gerrymandered in a certain weird way), then it could be an accident that it is simultaneously the case that X holds and p is true, and that intuitively will be a Gettier case.

    BTW, it’s not so unusual for theories to invoke a factor X that in fact entails truth. E.g. in Nozick’s definition of knowledge, the conjunction of conditions (2) and (3) seem to entail that p is true — i.e. condition (1) is strictly redundant. Goldman’s “local reliability” (as opposed to global reliability) entails truth as well.

  10. In response to Matt’s most recent comment, I think it should be noted that the barn facade case generally is taken to be a counterexample Goldman’s reliabilism. Henry is in fact justified, in a reliabilist way, when he believes that a barn is in the field–after all, his belief is the product of a reliable belief-producing process. (To avoid this result, Goldman would have to make the implausible claim that Henry is not using visual perception per se but rather something like visual perception in an environment with lots of barn facades. This would not only be ad hoc but would also make the generality problem much more difficult to resolve.)

    For what’s worth, Williamson’s view is also vulnerable to the Gettier problem, as I argue in a paper called, “Accidentally Factive Mental States,” which can be found at: http://www.niu.edu/phil/~reed/Accidentally%20Factive%20Mental%20States.pdf. Let’s say that someone is in a mental state called “gnowledge” iff she is in a Gettier case. In the barn facade case, then, Henry would gnow (but not know) that there is a barn in the field. Gnowledge is a factive state–one can have gnowledge only when the relevant proposition is true. So, the fact that knowledge is factive cannot be what differentiates it from a state that is not knowledge. (The main objection to this will be that gnowledge is not a genuine mental state. My argument is that there is just as much reason to think that gnowledge is a mental state as to think that knowledge is.)

  11. Ralph said:

    I�m willing to say that Sturgeon, Zagzebski and Merricks have �shown� this. Certainly they haven�t given an apodictic proof, but I do think that they have made the point extremely plausible. The examples that they cite make it highly plausible that if the factor X that a theory appeals to doesn�t entail truth (and isn�t gruesomely gerrymandered in a certain weird way), then it could be an accident that it is simultaneously the case that X holds and p is true, and that intuitively will be a Gettier case.

    But what is the reason for thinking that warrant isn’t gerrymandered? After all, one might think that one of the morals of the literature on the Gettier problem is that there is no clear definitive catalogue of ways in which one’s true and justified beliefs can fall short of knowledge.

    That said, notice that the way in which warrant would have to be gerryamandered isn’t that weird once you think about it: if a false belief is warranted, then if it becomes accidentaly true it ceases to be warranted–it sounds right to me. It certainly doesn’t sound worse than its negation, and that is what is needed for the argument to go trough.

  12. Baron, I think I’m in agreement about Williamson’s vulnerability to that case–I don’t think the analysis (?) of knowledge as the most factive mental state succeeds. I guess that, if Williamson were to accept that gnowledge is a factive mental state, he’d conclude that this was a counterexample to the FMS view and would still think that full justification was knowledge.

    Brian, I’m to some extent in agreement with this:

    whatâ??s really interesting about the JTB theory is the idea that we can usefully think of knowledge as having a normative component, an alethic component and a psychological component.

    But I think it can make a huge difference whether we take the normative component to be determined by internalistic factors. I understand what it means to say that a belief violates a norm because the believer did something wrong in arriving at it, or because it isn’t supported by the believer’s evidence (which latter might not be internalist on all views); I don’t really understand the sense in which a belief that isn’t knowledge is impermissible. So a theory on which the normative component is Williamson’s norm won’t yield an interesting JTB theory in your sense unless we have an account of why beliefs should satisfy the knowledge norm that doesn’t itself rely on the importance of knowledge. (Perhaps this is connected to the point in the last paragraph of Ralph’s first post, or to what you say in your original post about order of analysis.)

    Anyhoo, isn’t the alethic component just another norm on belief? Beliefs should be true, and beliefs should be justified.

  13. Thinking it over, I think this is what I should have said in response to Brian:

    If the interesting thing about JTB theories is the idea that knowledge has a normative component, an alethic component, and a psychological component, then Williamson’s theory is importantly unlike JTB theories–as are all other theories on which justification is factive. On those theories, you can’t disentangle the normative component from the alethic one. So it makes sense to say, with Williamson, “Gettier cases counterexample any theory that decomposes knowledge into non-factive normative, alethic, and psychological components,” and also “The normative component of knowledge comprises an alethic component.”

    It makes sense, but I think it’s wrong. I do think that the interesting norms on belief aren’t factive (except insofar as truth is a norm on belief), and that the JTB theory Brian describes comes closer to making knowledge important than Williamson’s theory does. (I would say “does a better job of explaining knowledge’s importance,” but I don’t think knowledge is important in itself.)

  14. Matt, for interesting discussion on this, see Zagzebski’s 1994 article “The Inescapability of Gettier Problems.” She argues here that all accounts (including externalist) of knowledge that allow for a justified true belief to be true because of luck or chance are vulnerable to Gettier counterexamples. (The content of this article, slightly amended, comprises a chapter in her book “Virtues of the Mind”) She claims that theories are immune to Gettier attacks only if justification entails truth… but this isn’t a desirable route to go (she thinks) because, if justification entails truth, then the the notion of justiciation is vacuous given that justified beliefs would just be true beliefs. She thinks her virtue-based account is capable of preserving non-vacuous justification without opening the door to Gettier attacks… but, for what it’s worth, I am convinced that Gettier-style counterexamples can nonetheless be presented against her account. –Adam

Leave a Reply

Your email address will not be published. Required fields are marked *