Fantl/McGrath and Optionalism

F/M argue for the Equivalence Thesis, the claim that you are justified in believing something iff you have knowledge-level justification for it. I’ve argued against this thesis here, so I was especially interested in how they argue for it in Chapter 5. They argue for it relying on the claim that you are justified in believing something iff you are required to believe it. They note, in footnote 6 on p. 131, that the argument could be weakened slightly by relying instead on the claim that you are justified in believing p iff of the three attitudes of belief, withholding, and disbelief, belief is the one you should have. The difference here isn’t important for the issue I want to raise, so I’ll ignore it.

I wonder why epistemologists think that we are always, usually, or even sometimes, under an obligation to believe. I think it is obvious that certain attitudes are forbidden (given the relevant state of information one is in), but given the obvious point that the denial of a forbidding is a permission, I wonder why we we’d want something more. I suspect the answer is that we think of belief on the model of action, and we are clearly under moral requirements to do certain actions at certain times, so perhaps we should expect something similar in the domain of belief.

But here I balk. First, there’s a bevy of worries about too much analogizing between belief and action. Most important here is that, with respect to action, there is nothing like the Quine/Duhem issue to be found. With respect to justified or rational belief, the introduction of new information that requires accommodation in some fashion always (nearly always?) leaves open a variety of ways in which to accommodate the new information in a way that preserves justification. Nothing similar seems to be true about what is required of us in terms of action: change the relevant circumstances, including what one is aware of, and there will still, at least typically, be some specification action type required of one. Or, to put the point differently, though Buridan’s Ass situations are possible in the case of rational or moral action, they are rare in comparison with the implication of the Quine/Duhem thesis that every change in state of information leaves open a variety of ways of accommodating the new information.

So, call Restrictivism the view that, given that one takes an attitude toward a proposition in a given total epistemic context, there is exactly one attitude that one may take toward that proposition in that context. And call Optionalism the view that there is always more than one attitude one may take toward a given proposition in a given context. Perhaps both positions are false, but given the Quine/Duhem point, it seems to me that Optionalism has to be closer to the truth. Yes?

One counterargument below the fold.

One may think, however, that this is too cavalier in how to handle the Q/D point. So suppose one has a total set of cognitive attitudes, all of which are justified, and one learns some new information that forces a change in total noetic structure. The Q/D point allows options for change here, but not without making additional adjustments. To put the point precisely, we had an original state of information S, which changes to S’ when the new information is learned. S’ lacks the coherence needed for the total set to remain justified, and so adjustment is necessary. The Q/D point allows various options here, but the important point to note is that taking any of the options requires another change in total noetic structure. To simplify, suppose there are exactly two options to restore justification to the total set of attitudes. Taking the first option will replace S’ with some other total state of information S”, and taking the second option will replace it with S”’. And that leaves open endorsing Restrictivism relative to each of these options, compatible with acknowledging the truth of the Q/D thesis.

The central point needed here for rescuing Restrictivism is questionable. The central point we might call “the 3 stages” reply: it insists that there is an initial state of information S, the middle state of information S’, and the resolution state (either S” or S”’), and that Restrictivism holds relative to each such state.

The 3 stages reply is correct, but irrelevant. The Q/D thesis is about how to accommodate new evidence, and one’s total evidence is typically not one’s entire system of information. So while the 3 stages point is clearly correct about one’s total states of information, it isn’t correct about one’s total evidence. Limited to that subset of one’s total state of information, the Q/D point is that one’s total evidence can initially be E, and when it changes to E’, one has options with respect to the changes in the remainder of one’s total noetic structure. And this interpretation is a better one of the real Q/D point that the one that insures that the 3 stages reply works.

So, the bottom line, I think, is that Restrictivists must reject the Q/D thesis.


Fantl/McGrath and Optionalism — 20 Comments

  1. Hear, hear!

    (I’d add this. It seems to me that two subjects might have the same evidence and differ wrt what they believe (one believes, the other suspends) and this reveals that one subject is willing to take risks the other does not. I don’t think this follows: either one subject is too risk averse to permissibly suspend judgment or the other subject isn’t justified in her judgment. That’s not an argument, but it’s an intuition that I think lots of people have. Not believing p isn’t like not feeding your pets. The world might not like to be misrepresented, but it doesn’t demand representation.)

  2. Here’s a possible counterexample to F&M’s thesis that you are justified in believing something iff you are required to believe it:

    (P) I don’t believe sentence P.

    Because one’s epistemic access to one’s own beliefs is pretty good, at least good enough to recognize whether one believes P, it would seem that one has justification to believe P. But how can one be required to believe P if it’s obvious to one that believing P falsifies it? Is not suspension the correct stance to take concerning P despite the robust quality of one’s evidence for its truth?

  3. Clayton, love the pet metaphor! So cool.

    Paul, nice example, and I’m a great fan of conditional fallacy problems here (together with my fantastic co-author Chris Menzel). But I think F/M may have a response here. Why not just say that you are required to believe, but if you actually do so, you’ll have done something wrong. Why is that different than having a claim you are justified in believing which is such that if you were to believe it, your believing would be unjustified?

  4. Paul, that response was a bit quick, so here’s an elaboration. Suppose F/M said, “OK, we embrace optionalism. So being justified only means that you are permitted to believe.” Your problem still surfaces. So, they say in response: “Right. Your present situation, absent belief, has the feature that believing the claim is permissible. But if you add the belief, you are in a new situation, one in which believing isn’t permissible.” That’s a bit surprising, but the stuff of which the conditional fallacy is made.

    So, I don’t think it’s the obligatory reading of justification that is the source of the problem.

  5. Thanks for all these great posts related to our book! On this one, though, I think I have to disagree. I guess I think there are cases (lots of cases) in which one really ought to believe if one takes an attitude at all. Consider a few weeks ago, when you weren’t posting as regularly as you have been recently. Before checking CD, say, on the morning of Feb 1, I might reasonably withhold judgment about whether Jon had posted that morning. Suppose I knew it was Feb 1, having checked my calendar in the morning. Then I check CD and I see a post which says “Feb 1 10am by Jon Kvanvig.” Suppose, finally, I think about whether Jon has posted that morning. In this situation, I think disbelieving would be unreasonable given my evidence, and so would withholding judgment. If I’m to have any of the three attitudes of belief, disbelief or withholding judgment, it ought to be belief. I don’t think it’s reasonable for me to keep withholding judgment on whether Jon has posted but start doubting whether, say, there is another person named ‘Jon Kvanvig’ who has posted or maybe whether there is an error on the website. If the Q/D point implies it would reasonable to make those other adjustments, I’d be very tempted to reject it.

    One other thing here, just for clarification (and this addresses Silva’s comment). We simply stipulate our intended meaning for ‘justified’ (the obliging meaning) and go from there. We’re not suggesting there aren’t perfectly good permissive notions of justification. Nor are we asserting a uniqueness claim in general. Maybe there are cases in which one is permitted to believe but not required to believe (nor required to believe if one takes an attitude at all).

    We focus on the obliging notion because we think there is a good argument that the obliging notion applies iff one has knowledge-level justification (and that this is not as obviously true for the permissive notion). I’m betting Jon is not going to like that argument, at least for the part of it which attempts to show that the obliging notion applies if one is knowledge-level justified.

  6. Matt,

    If I can follow up, suppose that you’re right that of the three attitudes you might take in C, believing is the only reasonable attitude to take. Why should we think that it follows that the attitude you epistemically ought to take is belief? I can see that this would follow if you thought that “It is rational for S to believe” entails “It is permissible for S to believe”, but I can’t see how it would follow otherwise. Does your argument against optionalism depend upon the assumption that the rational is the mark of the permissible? I thought that you rejected that assumption. I know we’ve had a number of exchanges in the past that touched on these issues, but I thought that your view was that it didn’t follow from the fact that, say, A-ing is the only reasonable option the agent had that A-ing was permissible. (Here I have in mind cases in which someone believes rationally but mistakenly that they must A where the mistake isn’t due to factual ignorance or mistake, but normative ignorance or mistake. If, say, someone ought to believe that she must A and does indeed believe that she must A, it seems A-ing is the only reasonable option for the agent, but it’s consistent with this that the agent’s obligation is to refrain from A-ing.)

  7. Matt, here’s another issue, besides what Clayton raises. I think the restrictivism route confuses what the experential info you receive is evidence for, with what you can make of it. I agree it is evidence for nothing other than that I posted, but that’s not the end of the story. You also know that the timestamp can be edited, so you might reasonably decide that I wrote this days beforehand, which I often do, and that I tend to arrange posts so that they are spread out. If you know that as well, then you have sufficient background info to make it permissible to conclude that I didn’t post on that day. My motto: it’s all about the background info! (If you like “O brother where art thou?”, you’ll hear this in the voice of Big Dan say, “It’s all about the money, boys” as he hammers them with a fallen branch.

    Not that I’d do the branch thing.

  8. Jon, if I had *that* background information, suspending judgment on the matter might have been reasonable, but with my naive information….

    I think there are lots of examples, though. My wife tells me she’ll be home at 6. I pick up the kids and come in the door at 4, and there she is. Is it reasonable for me when I first step in the door and see her clearly to withhold judgment that it’s my wife I see? I get an email and it’s listed as from “Peter Markie,” is it reasonable for me to withhold judgment on the identity of the sender? I guess I don’t think anything that survives from the background provides much of any meaningful counterevidence in these cases. By the way, I’m not trying to propose general principles here — anytime anybody comes home, etc. I’m thinking of boring actual past cases involving me.

  9. Hi Clayton. I think that there are many cases in which it’s permissible to believe but not permissible to withhold judgment or disbelieve. I think when I see my wife in the house at 4, having expecting her not to be home till 5, it isn’t epistemically permissible to believe it’s not my wife nor to suspend judgment on the matter. I shouldn’t suspend judgment about whether it’s my wife in this case; nor disbelieve; what I should think — if I take an attitude at all — is that it’s my wife. Even if reasonableness of belief in general doesn’t imply permissibility of belief, the sorts of cases I have in mind are ones in which it’s both reasonable & permissible to believe the relevant proposition and neither reasonable nor permissible to disbelieve or suspend judgment on it.

  10. It’s the only one of the three attitudes that is permissible. Since it’s the only one that is permissible, if you’re going to have one of three it should be that one.

  11. Yes, well, I can see that if somebody gives you the premise that it’s the only permissible option you’ll get your conclusion that it’s mandatory. How did you establish the premise? I still don’t see why refraining from believing isn’t permissible.

  12. Clayton, just to be clear: I am not arguing that in these cases one ought to believe tout court, only that if one takes one of the three attitudes (belief, suspension, disbelief), it ought to be belief.

  13. Some people seem to think one can take no attitude at all, even in cases like these. I do think it’s hard to see how that works in these cases, but it certainly seems possible in other cases.

  14. Matt, the idea isn’t quite that you have to *already* have the background info in question, but rather that there are lots of ways of accommodating new experience. The most obvious, and natural, is just to take the experience and form beliefs that those experiences are (direct) evidence for. But there is also the possibility of more elaborate responses, and if a person takes that route, it is hard to see why one would have to accuse them of irrationality. It is an interesting question why some folk will go the elaborate route and others won’t, but the same can be said of testing of scientific theories. When the experiment goes badly, the most obvious thing to conclude is that the theory is mistaken. But other elaborations are possible to make sense of the new information, some of which are irrational, but many of which are permissible rationally.

    So, to take one of your other examples. An email shows up in your inbox, and is said to be from P.M. One way to make sense of this experience, the most obvious one, is to form the belief that P.M. emailed you. But there are other ways of making sense of this experience, requiring a more elaborate doxastic adjustment. Perhaps one suddenly becomes quite concerned about hacked email accounts. Perhaps such possibilities become especially vivid, leading one to conclude that one needs to investigate to make sure that such a thing hasn’t happened in this case. These attitudes and opinions and concerns need not have been present prior to receiving the email, and one may wonder why this particular experience has triggered all this. They wouldn’t trigger such things in you or me, I grant. But when someone takes the more elaborate route, finding it more attractive from the point of view of getting to the truth now and avoiding error now, it is hard to see why one should insist that they are irrational. As I see it, the central point is the perspectival one, and once concerns and attitudes of the sort I describe have occurred, one can’t ignore the differences between that perspective and ours. Rationality, as I see it, is part of the sense-making cognitive enterprise, and there are normal ways to do this and more exotic ways.

    Not that this will convince you, I expect. 🙂 But I wanted to give a picture of how the Q/D point regarding the testing of scientific theories has a fairly straightforward analogue in other areas in which confirmation and justification are involved.

  15. Hi Matt,

    Sorry, I don’t want to drag this out forever, but I thought that the 4th option you mentioned wasn’t an option in the relevant cases. If it is and the 4th option is both available and permissible, then doesn’t the argument against optionalism fail? If it’s either not available or not permissible, isn’t the 4th option a distraction?

  16. Jon, thanks for this explanation. If I got concerned about the possibility of body-snatcher duplicates when I saw my wife at home early, I imagine she’d think I was being irrational! I’d say the same about the email case (the *actual* cases in my life of receiving emails from PM). Still, it’s interesting how you’re thinking of the role of (manufactured? illusory?) stakes here.

    Clayton, suspension of judgment is a doxastic attitude. The idea would be that one could permissibly fail to form any of the three doxastic attitudes by not having any of them toward the proposition in question. If we suppose for some reason you can’t permissibly fail to form any of the attitudes, I’m happy saying you ought to believe in the relevant cases. To make progress in our exchange, I suspect we’d have to work in some detail what we have in mind by “suspension of judgment.”

    Clayton, I should say also that I’d be more than happy just to drop all talk of “oughts” and do it in terms of reasonableness. Fantl and I could have defined up a notion of “unique reasonableness” (believing p is uniquely reasonable for one iff, of the three attitudes, only believing is reasonable for one wrt p) and then argued in the very same way for an equivalence thesis about it (i.e., one is knowledge-level justified iff believing p is uniquely reasonable for one). The main cost would be another technical term!

  17. Hi Matt,

    My point was only that if you want to refute the optionalist view, you’d have to show that for the n-options we have (believe, disbelieve, suspend judgment, 4th option), there’s n-1 options that are permissible. I didn’t think that something other than the three options were options in the cases where you were considering whether p, but that’s not really what’s at issue. I’m happy with the idea of unique reasonableness, by the way, since I don’t think that what’s permissible and what’s reasonable come to the same thing. Perhaps the main sticking point left over is justification. If the reasonable isn’t the mark of the permissible (which it can’t be if we drop the ‘should’ stuff), what happens to justified? I thought that X-ing could be justified for you if X-ing is permissible for you. So, can we say that there are multiple options available to S that S would be justified in even if there is only one option available to S that is reasonable? I like that view. Let’s go with that one!

  18. Clayton, just to be clear: in the book, we never argue against what you’re calling optionalism. Nor do I want to argue against it. We do claim that there are (many) cases in which the following holds: if you are to have one of the three attitudes, it ought to be belief. This is compatible with the claim that it is always permissible not to have any of the three attitudes. (The footnote Jon mentioned is crucial — even though we drop the conditionalized formalization for convenience in much of rest of the book.)

    We should talk about reasonable vs. justified vs. permissible/obligatory one of these days. At the present moment, I have a sort of “don’t care” attitude: if you give me “reasonable,” I can do everything I want to do with that notion, applied to both belief and action.

    On the other hand, your concerns about whether one can be justified in thinking one is justified in PHI-ing without being justified in PHI-ing do rearise for reasonableness: if one is reasonable to think that one is reasonable to PHI, could one fail to be reasonable to PHI? I think the answer is yes, and I would *think* that your negative arguments would apply here just as well as for permissibility/obligation. Anyway, we can’t hope to solve that one here!

Leave a Reply

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