Jon Kvanvig

Jonathan Kvanvig (University of Missouri-Columbia), “Coherentism and Justified Inconsistent Beliefs,” with commentary by Michael Bishop (Northern Illinois University).  Both the paper and the commentary can be found here.

Comments

[Christian Lee]

This wasn't brought up in the papers. I'm wondering though what exactly the objection to Coherentism is?

If Coherentism is true, then there can be justified inconsistent beliefs? But, there cannot be justified inconsistent beliefs. Etc.

But, if that's the objection, there aren't inconsistent beliefs in the Lottery Paradox or the Preface Paradox.

For example, Ticket 1 will lose, Ticket 2 will lose, Ticket n will lose and some ticket will win.

Say I believe all of those. There isn't any obvious inconsistency there. They cannot all be true, but justification over set of beliefs should not require joint compossibility. As Jon points out, inconsistency is not the same as necessary falsehood.

The worry then is that there is a notion of consistency that is being attributed to Coherentist. They need not accept it. They can accept that beliefs need to be consistent in a different way.

For example, consider the following set of consistent beliefs: Ticket 1 will lose, Ticket 2 will lose, Ticket n will lose and some ticket will win.

Coherentist can accept that these beliefs are consistent and each of them justified. Michael's point seems apt. These beliefs might lack "maturity" but that's fine.

[Christian Lee]

A "small" change:

If Coherentism is true, then there "cannot" be justified inconsistent beliefs? But, there can be justified inconsistent beliefs. Etc

That's the objection to Coherentism, right?

[Mike Bishop]

Christian,

Your revised account of the objection is correct.

As for your original comment: As I read him, Jon means to defend a fairly standard version of coherentism, where consistency is a necessary (though not sufficient) condition for coherence. The notion of consistency at issue is the boring, standard one: A set of statements is consistent iff there is some interpretation under which all statements in the set are true. The lottery paradox is supposed to involve one in an inconsistent set of beliefs because, as you note, those beliefs "cannot all be true" (i.e., every ticket will lose and some ticket to win).

Your comment suggests that perhaps you're working with some other notion of consistency. If so, then I don't think you have in mind the sort of coherence theory Jon means to defend.

[Christian Lee]

Thanks for the response Mike.

I suspect you are exactly right. If I'm not mistaken, then that notion of consistency Jon is interested in will entail that if Coherentism is true, then there cannot be justified necessarily false beliefs.

To push the point, that means many philosophical claims, mathematical claims, the standard lists, will not even be possibly justified. If that is right, then it seems to me a Coherentist will rightly respond, that that is not justification in any sense they are interested in.

If that is right, then I'm very surprised that Coherentists have defended the view that justification requires coherence in the sense Jon is interested in.

[jon kvanvig]

Christian, sorry for the delay; for some reason, I was expecting comments to appear next week, not this one. Anyway, thanks for responding.

Mike is right about the boring notion of consistency assumed by nearly all coherentists. The reason for adopting the boring notion is given by Lehrer: to allow inconsistencies is to pass up the possibility of achieving the epistemic goal of believing the truth and avoiding error. Not a very persuasive argument, I think; Dick Foley talks a bit in his book "Working without a Net" about the unrealistically high standard in play here. In any case, I expect something like Lehrer's argument is behind standard coherentism.

In any case, however, there are good reasons to use the boring notion in the lottery and preface cases, at least if you agree with the usual claim that such beliefs are not candidates for knowledge even when true. You might rationally believe that your ticket will lose, but you can't know that. If you agree with that claim, then I think coherentism needs something like the solution I propose to explain why lottery and preface cases don't refute the view.

[Christian Lee]

Dear Jon,

"If you agree with that claim, then I think coherentism needs something like the solution I propose to explain why lottery and preface cases don't refute the view."

Suppose that I agree that lottery and preface beliefs are not candidates for knowledge, but that I also deny that logical consistency is required for some beliefs to be justified. I argue that a sophisticated confirmation relation must hold between my beliefs, but not consistency. Given that kind of coherentist view, do you think your solution is still required?

[jon kvanvig]

Christian, I can't quite tell whether you think your sophisticated confirmation relation ends up endorsing the contents in question, but I think that's what you intend. If not, then I'll start over, but let's assume that things go this way. Then you can say that on the basis of this relation holding, the beliefs are all justified (even though inconsistent in the boring sense). They are still not candidates for knowledge, however, so some further explanation is needed as to why not. One possibility is the gettier condition, but as I argued in the paper, this seems to be the wrong place to look. So if the claims in question are not candidates for knowledge, something more is needed along the lines of what I propose in terms of a distinction between ordinary justification (even if it involves a sophisticated confirmation relation) and epistemic justification.

[Christian Lee]

Jon, the way I understood your paper was that you wanted to distinguish a number of different kinds of objections to coherentism. One objection is that it makes justified inconsistent beliefs impossible. I suggest that the right response to this objection, on behalf of the coherentist, is to require a different notion of consistency than logical consistency. They need to do this because of preface and lottery cases and the possibility of justified necessarily false beliefs. You draw the moral that they can accept logical consistency as a requirement, but need to distinguish regular justification from epistemic justification, such that the (RJ) does noit require logical consistency but (EJ) does.

There is a further question about how a coherentist is to analyze knowledge and you can argue that (EJ) is necessary for successful account. That might be right. But it seems to me that the objection has now shifted because coherentist can adequately respond to the problem of justified inconsistent beliefs, but they need a story about knowledge, a different point. I read you as accepting this, roughly. So, I think I agree that coherentist need to explain why lottery belief do not count as knowledge, and part of this explanation may end up being that these beliefs are not epistemically justifed. What isn't clear to me is that the problem of inconsistent justified beliefs raises this issue, rather than a general problem for any theory of knowledge.

I think that the lottery problem is particularly problematic for any version of coherentism. We don't know our ticket will lose, but this belief is support to degree .999999 (or whatever) by all of our beliefs. So, support isn't sufficient for knowledge. Any other condition that can be added to "support" will turn coherentism into foundherentism.

[jon kvanvig]

Well, I don't know what you have in mind in terms of some other notion of consistency other than logical consistency, but the uniform history of the theory is that coherence is something above and beyond logical consistency.

I think this started because your account of the paradoxes is a bit impoverished. You're right that simply believing of each ticket that it will lose is not inconsistent with the claim that some ticket will win. But then you say that all the beliefs can't be true. Yes, they can: as in when each of those beliefs are held and one increases the total number of tickets. What can't all be true is when the additional belief is added that tickets 1...n are all the tickets there are, together with the beliefs that some ticket will win and that each ticket will lose. In that case, one can deduce a contradiction, and it is impossible that all of the beliefs are true.

[Christian Lee]

Yeah. I think other premises need to be added to [ticket 1 will lose, ticket 2 will, etc...and some ticket will win], in order to derive the contradiction. That was kinda my point, that the inconsistency is not of the p and not-p variety, but something else.

You say that the "uniform history of the theory is that coherence is something above and beyond logical consistency." Sure. I totally agree. My worry is that when coherentists attempt to explain why one's belief that ticket 1 will lose, when the lottery has many tickets, does not count as knowledge, that the coherentist will need to appeal to some condition which will not seem coherentist at all. Of course that remains to be seen.

The idea is that the probability of this belief can be very high, confirmed by one's other beliefs to a very high degree, but this belief will need to meet some other condition to be knowledge, says intuition. This is a different issue from the one you aim to address so don't let me sidetrack you. This is just a worry I have.

You wrote "I don't know what you have in mind in terms of some other notion of consistency other than logical consistency." I don't have a favorite view. But this one will do the trick: A set a premises is consistent iff that set does not contain p and not-p. Both the preface and lottery cases are consistent in this sense.

Perhaps what is driving my worry is the following idea: one might be coherentist about justification and a non-coherentist about knowledge. I don't know the literature on coherentism, but I'd be curious if such a view would avoid your worry altogether. If so, then again, perhaps that is another way to respond to the objection without appealing to two senses of justification.

[jon kvanvig]

Christian, here's how one arrives at stronger requirements of consistency from the one you mention. Start with your account, and then note that it leaves open the possibility of justifiably believe p, p implies q, and not-q. Perhaps such is possible, but it is very hard to see how. Or imagine a person believe p&q, and not-q (but not believing q). A theory of justification has to answer a bit more to the logical connections than your account of consistency involves, and the question is how much a theory has to answer to logic. To the extent that coherentists are trying to offer a theory of justification in service of an account of knowledge, and they surely have been historically, there is pressure to adopt a substantive closure principle to the effect that one's evidence can't confirm p if it deductively implies the opposite. Or, put in a related way, a body of evidence can't confirm both p and ~p, and one way for a body of evidence to confirm a claim is to deductively imply it. The argument for the latter claim is this: you can come to justifiably believe a claim by deducing it from evidence you already possess.

Now it may be that some of this is mistaken, but if it is, there are two things to note. First, whatever coherence involves, it will be hard to argue that it doesn't involve logical connections in the way described. Second, even non-coherentists endorse the claim that a body of evidence can't confirm both p and ~p. So even if there is some way to work things out so that a weaker notion of consistency is involved in coherentism, it can't be done just by deciding to weaken the requirement in the face of the problem of justified inconsistent beliefs.

[jon kvanvig]

Response to Bishop

I want to express my thanks to Mike for the time he spent on a project of mine when he has significant projects of his own to which he could have used the time. I have learned from his remarks, and will be able to improve the paper considerably in light of them.
I will be quite brief here, however, and as a result will not remark on everything worthy of discussion in what he writes. Instead, I’ll focus on what I think is a miscommunication that occurred, which I believe is primarily a result of my not explaining one point sufficiently. The point is this. Attributions of justification either to actions or mental states are ambiguous. In the case of the justification of actions, the ambiguity is between type/token predications; regarding mental states, the ambiguity is between state and content. Start with the case of actions. When I say, “Joe’s driving 55 is justified in these circumstances,” what I say does not semantically imply that Joe is driving 55. He might be driving 35, and I say what I do to complain about his slowness. In other cases, of course, Joe is driving 55 (or more), and there is no need to resort to an action-type in order to specify that to which justification attaches. The point, however, is that some such distinction is required to acknowledge the point that we can speak of the justification of action whether or not the action is being performed.
The same is true of the justification of mental states. I can correctly say that believing that Bush is President is justified for Joe even when Joe does not believe that Bush is President. My utterance might be motivated by a dissatisfaction about the way in which Joe refuses to acknowledge the obvious. In other cases, I might be remarking on the epistemic status of one of Joe’s beliefs. In order to account for this ambiguity, we must treat some attributions of justification to belief as ascriptions of the property of justification to a propositional content rather than to a token belief. In other cases, since the token belief is present, we are free to treat the attribution as attaching to the mental state itself.
This ambiguity plays out in Mike’s comments when he interprets my remarks about the justified attitude of closure to further inquiry in terms of requiring some actual, token attitude to be in place. This interpretation leads him to attribute to me a position about the “maturity” of inquiry, a position which I resolutely resist. I did write that a person who knows has a justified attitude of closure to further inquiry, and that when you know, you have the experience of closure. But regarding justification, I did not say that any such token attitude needed to be present for epistemic justification to obtain. I only said that the attitude needed to be justified, and here the ambiguity I just pointed out is crucial. When I say the attitude is justified by your evidence, that doesn’t imply any token attitude. Such justification can obtain in the same way that Joe’s belief is justified when I am complaining about his capacity to deny the obvious. The point is that even if knowledge does require such a token attitude, all that I argued for and all that I needed to argue for was that the attitude type is licensed by the quality of your evidence when you have epistemic justification. Attitude types are best thought of in terms of the content of a possible token attitude, and thus my position doesn’t require any actual attitude.
I have devoted quite a bit of attention to this point in prior writings, but failed to include the point here, so the misunderstanding is certainly my fault. The result led Mike to introduce the concept of maturity of inquiry, and led him to formulate interesting objections to the requirement of such maturity, all of which I think are exactly correct. Since the position I’m arguing for doesn’t impose any such requirement on epistemic justification, however, these objections are good ones to those who would wish to revise my view in light of other inadequacies with it to include a requirement of maturity of inquiry.
Even given this correction, Mike’s worry about the skeptical direction of my defense of coherentism will remain, for it is well-known that the heart of, e.g., Unger’s argument for skepticism appeals to the attitude of closure to further inquiry. I believe this worry is exactly as it should be. Unger’s argument is a powerful argument for skepticism, and though I believe it is mistaken (and have posted on Certain Doubts about what I think the mistake is), it is not mistaken in virtue of claiming that knowledge involves a justified attitude of closure to further inquiry. Unger thinks that the attitude in question is a dismissive one concerning the possibility of counterevidence, and I believe that overstates the content of the attitude. I believe the content is something more like this: any counterevidence that might be acquired would be misleading evidence. This content can still be used by Unger in an attempt to deny the attitude of closure to further inquiry to closemindedness, so the substitution of a slightly weaker content doesn’t settle the issue of skepticism. What stops this attitude from having the skeptical implications Unger claims is a scope ambiguity. For Unger’s argument to work, he needs it to be the case that if you gain further evidence, you are justified in dismissing it as misleading. What I say doesn’t imply that strong a claim. Instead, all it implies is that you are justified in the present in believing that if you gain further evidence it will be misleading evidence. The scope ambiguity concerns whether the justification operator governs the entire conditional or only its consequent. The view I’ve defended here has it governing the entire conditional; Unger’s position is that the operator can be moved to the consequent. I won’t try to settle that dispute here, but simply note that the question of whether my view has skeptical consequences is just this question of whether Unger is entitled to the narrow scope reading from the wide scope reading that I argue for here. The only thing I will say in my defense here is that this result that calls to mind the threat of skepticism is a pleasing one: any adequate theory of the kind of justification required for knowledge ought to make us have to do some serious philosophical work to avoid collapsing into Ungerian skepticism.

[Christian Lee]

Jon, I think I'm beginning to see where you are going. I'm also beginning to think this issue is really, really tricky.

Above you noted that my account of consistency is consistent with each belief in a belief set of the following form to be justified: p, p implies q, not-q. The suggestion I made is consistent with this. I agree it is hard to see how this is possible.

This point is exactly the kind of point that I think is important. On the one hand, if we weaken consistency too much, then justification will be too cheap. An instance of the above set will be possibly justified. On the other hand, if we strengthen the consistency requirement too much, then justification will be too expensive. Preface and lottery beliefs will not count as sets of beliefs that can be justified. Necessarily false beliefs will possibly be justified. Either way, there will be a problem. I think this is a problem and not just a problem for coherentists.

My point is just that we "already" face this problem, regardless of what view we adopt, coherentist or otherwise, and so the weakening of consistency, in some way, is independently motivated. Here I have in mind weakening consistency so that beliefs sets we intuitively think can be justified (I-justified) will correct count as justified. In your paper I recall that you think preface type beliefs can be justified (I-justified). That still leaves open the question for you as to whether beliefs of the form p and not-p can also be justified. You haven't taken a stand one way or another as to whether simple inconsistencies can be justified, and if not, what the explanation for this fact is.

So, when you say above that "So even if there is some way to work things out so that a weaker notion of consistency is involved in coherentism, it can't be done just by deciding to weaken the requirement in the face of the problem of justified inconsistent beliefs" I agree. But, again, I think there are other reasons, independent of the problem of justified inconsistent beliefs, to weaken the relevant notion of consistency.

You write, "To the extent that coherentists are trying to offer a theory of justification in service of an account of knowledge, and they surely have been historically, there is pressure to adopt a substantive closure principle..."

Here is where I think things get tricky. I do not see why coherentist should offer a theory of justification to service an account of knowledge. Instead they might simply consider intuitive cases of justified belief, bracket off the issue as to whether these beliefs are also instances of knowledge, then attempt to give a coherence theory of justification. Perhaps Lycan's proposal is correctly seen to be an endeavor to do this. Perhaps the trying to work out a theory of consistency that classifies cases correctly is another proposal to do just this. So, it's not clear that coherentist should be (even if they have been) developing a theory of justification so that that view is consistent with and helps to explain her theory of knowledge. She might tackle these two projects separately.

Your view of E-justification is analytically connected to knowledge. And this notion is essential to your cohernetist's reponse to the problem of inconsistent justified beliefs. This suggests to me that you see her project as one that tries to explain justification so that it is consistentent with a view of knowledge. Perhaps this is the right way to go, I'm not sure.

[Christian Lee]

Correction: Necessarily false beliefs will "not" possibly be justified.

[jon kvanvig]

Christian, I think we agree on nearly everything here. We agree that the consistency issue is a problem for everybody, and that a strong closure principle is going to land any view in serious trouble. So weakening the notion of consistency is going to be required of any adequate theory. The reason for my paper is that coherentists seems especially threatened here, since they have universally insisted on a very strong consistency requirement. They have universally claimed that coherence goes beyond logical consistency, requiring more than simple compossibility. Here their view is threatened by the paradoxes, and once we grant that compossibility can't be a requirement on justification, what the coherentists have said makes it look like the analogue of the incompleteness results for Hilbert-style formalism. Given coherentist motivations, however, I think there is a way out, though of course that doesn't preclude the possibility of other ways out as well. My concern above was that other ways out involving some weakened notion of consistency have to be motivated by coherentist ideas in some way to count as a providing a means of salvation for the view.

[Christian Lee]

Cool! Sounds like we're pretty much on the same page. I'm eager to see if any coherentist will pop up in the future that will take to task its deliverance. Thanks for the conversation. Best, Christian