The reason-giving argument against pragmatic encroachment

In the wake of Kvanvig’s series on pragmatic encroachment, here is an argument against the view that rests on assumptions having to do with reason-giving: Roughly, if a pragmatic encroachment theory of knowledge were true, it would be reasonable to cite practical factors as epistemic reasons for ascribing or denying knowledge. But it is not. So, pragmatic encroachment theories are is false. To make the argument less rough, I will use this broad characterization of Pragmatic Encroachment About Knowledge (PEAK): 

PEAK: The true theory of knowledge is a pragmatic encroachment theory if and only if practical factors partly determine the truth-value of knowledge ascriptions

Thus, I use ‘pragmatic encroachment’ is as a generic label of the family of views that share the feature that they accept the right-hand-side of PEAK. To make the argument compatible with various conceptions of reasons, I use the ugly formulation ‘((reasonably) believed) practical factor.’ Given this terminology, here is the argument from reason-giving: 

R1:      The true theory of knowledge is a pragmatic encroachment theory only if practical factors partly determine the truth-value of knowledge-ascription.                                             [From PEAK] 

R2:      If practical factors partly determine the truth-value of knowledge-ascriptions, then ((reasonably) believed) practical factors can be good partial epistemic reasons to ascribe/deny knowledge. 

R3:      It is not part of our reason-giving practices to cite ((reasonably) believed) practical factors as partial epistemic reasons to ascribe/deny knowledge. 

R4:      If it is not part of our reason-giving practices to cite ((reasonably) believed) practical factors as partial epistemic reasons to ascribe/deny knowledge, they cannot be good partial epistemic reasons to ascribe/deny knowledge. 

C5:      ((Reasonably) believed) practical factors cannot be good partial epistemic reasons to ascribe/deny knowledge.                                                                                                                                [from R3, R4, MP] 

C6:      Practical factors do not partly determine the truth-value of knowledge ascriptions.                                                                                                                                                                                                                                    [from R2, C5, MT] 

C:        The true theory of knowledge is not a pragmatic encroachment theory.   [from R1, C6, MT] 

I’m interested in hearing what pragmatic encroachers think is wrong with this argument. I’ve been fiddling with it for a few years now, and I’m still not sure that I have it straight. If you’d like a motivation for a premise, just let me know and I’ll add it in the comments!


The reason-giving argument against pragmatic encroachment — 11 Comments

  1. Hi Mikkel. This is a neat and new argument. I guess my initial thought is that we do in ordinary practice cite practical factors when we deny knowledge. “Look, you don’t really know that the drug is safe. If you’re wrong and you take it, you’ll die!” “If that train is the express you’ll miss your stop. You better ask a second person so that you know it’s the local.” Etc. I guess I’m not entirely sure what would constitute an ordinary language practice of citing practical factors in attributing or denying knowledge, but these look like plausible candidates. If there were a nice locution that would count as exemplifying such a practice, we could Google it.

  2. One more thought, Mikkel: whenever you fail to know partly because of practical factors, it will also be partly because your evidence is insufficient. So, when you can’t reach a shelf, it’s partly because you’re too short and partly because the shelf is too high. When you lose/don’t finish a race, it’s partly because you were too slow/didn’t have enough energy, and partly because the others were too fast/the finish line was too far away. Sometimes it’s more appropriate to cite one of these factors rather than the other. And I don’t see a principled reason for thinking that in some domains there aren’t practical reasons for (almost) always citing only one of kind of factor. So, it might be that, when you fail to know because of practical factors, it’s only practically appropriate to cite insufficiency of evidence, even if what makes the evidence insufficient is that it wasn’t enough to reach the practically determined threshold.

    Why might it be almost always more practically appropriate to cite lack of evidence? Perhaps because that’s really what you can control most directly (you can just decide to know, but you can decide to acquire more evidence). You can practically appropriately cite both your running out of energy and the finish line being too far away, because you can choose to run in shorter races, and you can choose to train better. So, even if we don’t find many practical-factor-citing knowledge-denials/attributions, I don’t know how much evidence that is that practical factors aren’t reasons to deny/attribute knowledge (so, this is a worry about R4; of course, McGrath and I routinely point to knowledge-attributions/denials as defenses/criticisms of action, and I’m not sure how the points in this comment bear on those).

  3. Hi Jeremy – thanks a lot for these suggestions (contra R4). I have a point on your first comment before turning to the second one. I agree that citing practical factors can provide a context in which a denying knowledge will in some sense be more conversationally apt. But, as you note, the question is whether this sense counts as citing the practical factors as an epistemic reason for denying knowledge. And I am not quite ready to let the encroacher off the hook on the basis of your examples.

    There are, in fact, some locutions that are interesting to consider here. One is explicit reasons talk such as “p. The reason why is R.” Likewise, there are argument-markers “R. Therefore, p”. Such terms serve to make it clear that something is cited as a reason. To see this, consider some uncontroversial cases of factors that partly determine the truth of a knowledge ascription.

    (1) S does not believe that Obama is American. Therefore, she don’t know that he is American.

    (2) While S believes that Obama is a good point guard, he does not know it. The reason why is that his belief not based on any evidence but on stereotyping.

    Clearly straightforward factors that partly determine the truth of a knowledge ascription can, given a bit of context, by cited as an explicit epistemic reason. I.e., within the scope of “the reason why is…” or after an argument marker such as ‘therefore.’ But this is a major asymmetry with the practical factors. Just consider the cases you provide:

    (3) Look, you don’t really know that the drug is safe. The reason why is that if you’re wrong and you take it, you’ll die!

    (4) If the train is express you will miss your stop. Therefore, you don’t know that it is local until you have asked a second person.

    As opposed to (1) and (2), (3) and (4) are not felicitous. So, it seems to me that practical factors cannot
    be cited *qua* reasons in the same explicit manner that purist factors can. If you agree, we can move on to the interesting suggestion in your second comment. If not, why not?

  4. Really interesting post, Mikkel! My initial reaction was similar to Jeremy’s, but the more I thought about this the more I started to feel the pull of the intuition that I think you’re hoping us to have, which is that while we might say, ‘He doesn’t know’ having pointed out that the target is mistaken, doesn’t have good reasons, etc., we wouldn’t say ‘He couldn’t know that because [describes something of practical significance]’

    The worry that I do have might be similar to a worry Jeremy raises in his second post. If we think of practical factors as factors that determine what strength of epistemic position would be necessary for knowledge, those factors might function like factors that help determine the extension for some gradable adjective. It seems that there might be facts that help determine the extension in some particular context and that wouldn’t be appropriate to mention in explaining why the adjective doesn’t properly apply in the relevant context (e.g., if we’re talking about whether someone’s tall or rich and someone challenges that, I don’t think I’d say, ‘He’s tall, we’re not talking about basketball players’, but whether he counts as ‘tall’ in the context would depend upon whether we were talking about basketball players). (Fwiw, this is off the top of my head. I’m sort of inclined to think that there’s something to your argument.)

  5. I don’t find 3 and 4 obviously infelicitous. Even less obviously infelicitous (to me) are:

    5) She doesn’t have enough evidence to act as if this is the local. So, she doesn’t know it’s a local.

    6) You don’t have enough evidence to act as if the drug is safe. So, you don’t know the drug is safe.

    But I’m willing to grant I’m not the best judge of what’s felicitous in this matter.

  6. Hi Jeremy (and hi Clayton, I’ll try to address your comment in a bit).

    I’m happy with (5) and (6) since they can be given a purist reading. But it looks like we have a clash of intuitions re (3) and (4). That might be bad for the arguments dialectical force. But I’m happy that you think (5) and (6) are sounds more felicitous than (3) and (4).

    Let me say a bit about your interesting proposed explanation why we cite purist factors (such as lack of evidence) rather than practical factors as reasons. Roughly, the idea is that we can do something about the former (gather evidence) but not the latter. But this is not always true. What’s worse, in cases where it is not feasible to gather evidence but where it is feasible to work on the practical factors, it *still* seems odd to cite the latter as a reason. Here is a candidate case:

    The scene is a windswept station with no-one but S and S* around. There are no printed schedules, the internet is down and the empty train is just about to leave. So, there is no way of gathering further evidence before a decision about whether to board must be made. But, fortunately, there is a way of improving the practical matters! After a brief phonecall S* proclaims:

    (7) I just called and postponed the important meeting in Foxboro until tomorrow. Therefore, we know that the train is the local one.

    This strikes me as just as bad as (3) and (4). But since we disagree about the former, we might disagree about (7) as well.

  7. Hi Clayton – it looks like some of what I said in response to Jeremy’s second post address your worry as
    well. But let me see if I can swing you around on the gradable adjectives like ‘tall’ or ‘expensive.’ It seems to me like we can, in the right conversational context, cite the sort of practical factors that determine the relevant contrast class as explicit reasons for ascribing some gradable adjective. Consider the coach and the travel guide:

    (8) You have skills, Joe, but at 6´3´´ you just can’t play center position in the NBA. An average
    NBA center is 6´11´´. So, you’re not tall.”

    (9) People in this country live on less than two dollar a day. So, your watch is expensive.

  8. I don’t think Matt and I would like to say that there is knowledge in your windswept case, anyway, so I’m pretty happy not to allow practical factors as reasons to ascribe knowledge a la your 7. In any case, the point wasn’t that lack of control was the explanation. That was just a possibility. The point was that even when p entails q we won’t always expect q-factors to be cited as reasons to deny or assert p. (It might also depend on order of explanation or whether we tend to only be interested in p because of its bearing on q, etc.)
    Notice that it isn’t controversial that when p is certain for you, then you can act as if p. But citing practical factors as reasons for attributing/denying certainty seems no less odd than for knowledge:

    8) You shouldn’t board the train. Therefore it’s not certain it’s a local.
    9) You can’t act as if the drug is safe. So you’re not certain it is.

    Both are no more felicitous than the corresponding claims about knowledge. But it’s still true that if p is certain for you, you can act as if p.

  9. Thanks for this Jeremy! Setting the case aside for a minute, consider your claim:

    (CiteStrong): “when p entails q we won’t always expect q-factors to be cited as reasons to deny or assert p.”

    (CiteStrong) seems right, for the reason you mention. It is only in some conversational contexts that it makes sense to cite q as a reason for claiming that p. But I think that I only need the weaker claim:

    (CiteWeak): For any partial determiner of the truth-value of a knowledge ascription, R, there are conversational contexts in which R may be cited as an explicit epistemic reason for ascribing/denying knowledge.

    Paradigms of the conversational contexts in which R may be cited, are contexts in which all other necessary conditions are reasonably presupposed to obtain. In such contexts, it typically makes sense to cite R as a reason for ascribing/denying knowledge. This is what (1) and (2) illustrate. However, (CiteWeak) still gives rise to an asymmetry between the purist factors and the practical factors. For it seems (to me) that there are no conversational contexts in which a practical factor can be cited as an explicit reason for ascribing/denying knowledge. That is, even in the cases where all other factors are agreed upon, citing a practical factor strikes me as infelicitous. At any rate, I have yet to see a convincing case that might serve as a counterexample to R3.

    Return to the windswept station case, I failed to specify that the case should be construed such that S and S* had some background evidence for believing that the train was the local one. How much background evidence? Enough to make it natural to ascribe knowledge in a low-stakes case, but not enough to make it natural to do so in a high-stakes case! Given this specification, I thought that an encroacher seems committed to say that since S’s purist epistemic position remains constant, the variation of practical factors determine the truth of knowledge ascriptions. That is, the case is meant to be one of the paradigm cases in which all other conditions on knowledge are agreed upon. But, if so, the remaining practical factors should, according to pragmatic encroachment accounts, be citable as reasons.

    Cheers and thanks again,

    PS: I’ll have to think about the impact of the cases of certainty.

  10. Pragmatic encroachers are not committed to saying that knowledge returns when the stakes get low. They are committed to saying that knowledge can be destroyed when the stakes get high. That’s because the main knowledge-action links that demonstrate pragmatic encroachment impose only pragmatic necessary conditions on knowledge. It has always seemed to us implausible to allow that knowledge returns after the stakes return to LOW. It’s harder to say why knowledge wouldn’t return, and this is a problem for the pragmatic encroacher. Salience-considerations are a possible option, but not one we’re particularly happy with. But Matt and I, at least, want to deny that you would get knowledge back in your Windswept case, and I don’t think we’re committed to saying that you would get knowledge back.

    CiteWeak does seem more plausible than CiteStrong, but I still don’t see why it’s guaranteed that, for all partial determiners, there will be an ordinary context in which it is appropriate to cite the determiner as a reason. It has to be an ordinary context, of course, because there are non-ordinary contexts in which it’s no problem: in philosophical contexts, for example, it’s no problem; Matt and I do it all the time :). But in ordinary contexts, it just might always be the case that there’s no point in citing practical factors as reasons to deny knowledge, because, in ordinary contexts, it just might always be the case that we’d take the importance of knowledge-denials to indicate something about what we should do, not the other way around. . Or it might always be the case that, in ordinary contexts, knowledge is taken to be explanatorily prior to practical factors. I do agree, though, that there is a burden on the pragmatic encroacher to come up with a good story about this.

    When I try to think of counterexamples to R3, I find myself wanting to say things like this:

    “She should have gone to check if her dad’s air conditioner was on. I guess she didn’t know it was.”

    But the difficulty here is that I would never cite this reason to deny knowledge, not because there’s no connection between whether she knows and whether she should have done this, but because whether she should have done this depends, in this case, on whether she knows. But that still leaves the conditional true: if she knew the air conditioner was on, she needn’t have gone to check.

    Finally, I guess we’ve already established that I see some of the counterexamples to R3 as more convincing than you do. That’s no surprise! Let me try a couple more:

    Here’s are two from Matt:

    A: I know that p.
    B: Do you? Our lives are on the line!

    A: I know p
    B: Did you realize our lives were on the line?
    A: No.
    B: Do you know?
    A: Maybe not.

    Matt’s cases are not ones in which practical factors are cited in knowledge-denials. They are cases, though, in which practical factors are cited as reasons to doubt knowledge attributions.

    Here are two that involve actual knowledge-attributions and denials:

    “Wait, you’re saying she SHOULD have stayed at home instead of going out to get yams? I guess she knew we had them all along.”

    “Wait, you’re saying she SHOULDN’T have just stayed at home? I guess she DIDN’T know we had them, after all.”

  11. Thanks for sticking with it Jeremy (+Matt) – very helpful!

    I take the point that there may be an asymmetry between the claim that knowledge may be generated by lowering stakes and the claim it may be destroyed by elevating them. However, confess to worry that it is ad hoc in the absence of a principled rationale for postulating the asymmetry. But we should set that aside since it is a problem that’s independent from the reason-giving issues.

    I like CiteWeak because, I think that for every partial determiner of knowledge that I believe in, I can cook up a context in which it can be cited as an explicit reason for ascribing/denying knowledge. I agree with your idea that the context must be an ordinary one (at least insofar as the condition on knowledge is not a technical articulation of a principle. It is not a problem for safety principles that the folk do not cite them, as stated by the epistemologists). Anyhow, if you agree that the burden of proof is on pragmatic encroachers to explain why we don’t cite practical factors as explicit reasons, that is as much as I can reasonably hope that the argument can establish.

    Fun methodological fact: The interesting lines of explanation you suggest resemble the sort that many pragmatic encroachers eschew. Arguing that we don’t cite practical factors as reasons for knowledge denials because the latter have a certain conversational functions looks like what DeRose would call “chalking it up to pragmatics.”

    To be sure, I think that such explanations should be taken seriously. But I don’t think the pragmatic encroachers should be able to cherry-pick her methodology by the case. In the original trains, banks and planes cases, it is assumed the knowledge ascriptions track truths. But in the cases of our reason-giving practices are assumed to track pragmatic features of knowledge ascriptions. Some principled account is owed for why the ordinary talk is treated differently (this methodological constraint applies equally to the purists, of course).

    As for the proposed counter-examples to R3, I suspect that you can guess my line by now. I agree that those are natural ways of talking and so I accept that there are stakes-effects to be explained by the purist. But the cases lack the markers of explicit reason-giving (“Therefore…”, “For this reason…” etc.) that determiners of the truth of knowledge claims generally exhibit. This I think we can agree on. The disagreement is over whether invoking practical factors remain felicitous when paraphrased as explicitly providing epistemic reasons. For example:

    A: I know that p.
    B: Our lives are on the line. So, you don’t know.

Leave a Reply

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