Flagship

From initial submission to acceptance, this one took eleven months. That’s not bad at all. But as many of you know, it sure seems like it takes forever, even when things move rapidly by journal standards. Now finally the wait is over! “Epistemic Invariantism and Speech Act Contextualism” will soon feature in a volume of The Philosophical Review near you.

The short abstract: This paper shows how to reconcile epistemic invariantism with the knowledge account of assertion. My basic proposal is that we can comfortably combine invariantism with the knowledge account of assertion by endorsing contextualism about speech acts. My demonstration takes place against the backdrop of recent contextualist attempts to usurp the knowledge account of assertion, most notably Keith DeRose’s influential argument that the knowledge account of assertion spells doom for invariantism and enables contextualism’s ascendancy.


Comments

Flagship — 55 Comments

  1. It’s a bit odd to congratulate someone you don’t know, but — congratulations, John! I just read your paper, and I think it’s terrific. I hope it gets the wide attention it deserves.

    I wonder, though, whether you think that the epistemic contextualist still has the following advantage over the invariantist. Consider the following (in my view quite realistic) variation on your “High Door” case:

    Our family has just pulled out of the driveway, on the beginning of a week-long vacation to New Hampshire. We live in a relatively safe neighborhood, but there has been a rash of burglaries lately. As we near the corner of our block, my wife asks, “Do you know if the door’s locked?” I respond, “Yes, I know it’s locked. I remember turning the key and feeling it click.” My wife reminds me of the recent string of burglaries and points out how devastating it would be were we to return from our trip only to find our home ransacked and our belonging stolen. She continues, “It’s rare, but sometimes the bolt hits the strike plate in such a way that it clicks but doesn’t lock. So … do you really know that it’s locked?” “I’m *think* it’s locked,” I reply, “but I don’t know. I’d better go back and check.”

    Here, the high stakes and the salient error possibility make the speaker say that he doesn’t know that the door is locked. It seems to me that on the view suggested by your paper the speaker is mistaken. If the dynamics of this conversation are parallel with those in your original case, what’s going on is (something like) this: in this context saying “I know that P” counts as guaranteeing that you know that P, and the speaker doesn’t satisfy the very stringent epistemic criteria for appropriately guaranteeing that he knows that P. But instead of saying, “I think I know that P, but I’d better go check”, he flat-out denies that he knows.

    A slightly more general way of putting the question: on the view suggested in your paper there are contexts in which a speaker knows P but it’s epistemically inappropriate for her to say “P” because doing so in that context would count as guaranteeing that P, and she doesn’t satisfy the epistemic standards necessary for appropriately guaranteeing that P. When I try to imagine such a situation, it seems clear to me that the speaker could perfectly appropriately say “I don’t know that P” — perhaps as part of her explanation for why she’s unwilling to say “P”. If that’s right the view saddles speakers in such situations with speaking falsely (and I assume with corresponding false beliefs).

    That’s not the end of the world, of course, but epistemic contextualists can explain these cases without saddling anybody with false utterances or beliefs. And certainly that’s an advantage.

  2. Thanks, Geoff, for the kind words and the thoughtful criticism.

    You’re right, it wouldn’t be the end of the world. But let me see if I can’t avoid ceding the ground.

    About your modified case, I agree that “P?” and “Do you know if P?” are pretty much interchangeable, so the wife’s first question in Modified High Door is perfectly natural. But my instinct is to *not* respond to a question of the form “Do you know if P?” by saying, “Yes, I know that P,” but rather, “Yeah, P.” In short, I think it’s more natural to keep the answer first-order, because the force of the question seems to be first-order, rather than genuinely about what I know. I’d be interested to hear if others feel the same.

    Now, when we get to the second part of the dialog, after the wife’s speech, it’s still not entirely clear to me that the force of the question is second-order (i.e. about whether I know P, as opposed to being just about P). I’m inclined to respond by saying, “Well, I’m not sure. I’d better go back and check,” or again just “I *think* it’s locked, but I’d better go back and check.” I’m not inclined to say, “Actually, no, I don’t know.”

    Why think my instincts are right here? Well, here’s one bit of evidence. If we let the film roll a little bit longer, you’ll see me go back, double-check the door, return to the car and say, “Yeah, it’s locked.” I don’t say, “Yeah, I knew it was locked.”

    Another bit of evidence: when I come back and say, “Yeah, it’s locked,” it’d be perverse for my wife to then respond, “That’s not what I was asking about. My question wasn’t about the door being locked — it was about whether *you knew* it was locked.”

    I’m not saying that you can’t genuinely ask second-order questions. But I do think it ordinarily takes some work to get that effect.


    BTW, I seem to remember Jim Stone making a similar point about the force of questions in his 2007 PPQ article, “Contextualism and Warranted Assertion.” I just double-checked, and it’s on p. 97.

  3. I just wanted to quickly clarify something. I do think it’s perfectly natural to respond by saying “I don’t know” in many cases. But when you have lots of evidence, as in the Door cases, I do find it unnatural. Not sure what to make of the difference.

  4. Hey John,

    Thanks for posting this very elegant paper!

    Geoff asks you to consider a case in which “My wife reminds me of the recent string of burglaries and points out how devastating it would be were we to return from our trip only to find our home ransacked and our belonging stolen. She continues, “It’s rare, but sometimes the bolt hits the strike plate in such a way that it clicks but doesn’t lock. So … do you really know that it’s locked?” “I’m *think* it’s locked,” I reply, “but I don’t know. I’d better go back and check.”

    You find something inappropriate in the example, as Geoff describes it, namely that, in response to your wife’s repeating her question “do you really know that it’s locked?”, it’s appropriate to say “Well, I’m not sure. I’d better go back and check,” or “I *think* it’s locked, but I’d better go back and check”, but it’s not appropriate to say, “Actually, no, I don’t know.”

    Now, why think that it’s inappropriate to reply to your wife’s question by saying “I don’t know it’s locked”? (I don’t find it to be inappropriate, though I’m willing to be convinced.) You offer two pieces of evidence. But I’m afraid I don’t understand how the evidence that you offer serves to confirm the hypothesis that you want. Consider each piece of evidence:

    (1) “you’ll see me go back, double-check the door, return to the car and say, “Yeah, it’s locked.” I don’t say, “Yeah, I knew it was locked.” Of course you’re right. But isn’t that because you know that, prior to performing the double-checking and acquiring the additional evidence, you did NOT know that the door was locked (at least relative to the epistemic standards in place given the high stakes)?

    (2) “when I come back and say, “Yeah, it’s locked,” it’d be perverse for my wife to then respond, “That’s not what I was asking about. My question wasn’t about the door being locked — it was about whether *you knew* it was locked.” But of course, if you go back and double-check the door lock, aren’t you tacitly admitting (or perhaps, as in Geoff’s case, explicitly admitting) that you didn’t know that the door was locked prior to doing the double-checking? If all that your wife was interested in was your epistemic state (and it’s not clear why she would care about your epistemic state, unless it was for the sake of gaining assurance about the safety of your house), her curiosity would have been satisfied as soon as you started walking back to the house to double check. She was interested in your epistemic state, but she was interested in it only because she was interesting in gaining assurance about the safety of the house. Once you admit that you don’t know that the door is locked, then it remains for you to do something (like double-checking) that will put yourself in an epistemic state that can offer her the desired assurance.

    Am I missing something about how the two pieces of evidence bear on your reply to Geoff?

    Ram

  5. Hi Ram,

    Thanks for weighing in.

    I’ll answer your numbered items first, and then come back to the more general issue of the propriety of self-denying knowledge in these cases.

    Re 1, the reason I wouldn’t say “Yeah, I did know it was locked” is not that it was *false* that I knew it was locked. Rather, it’s that I’m being cooperative. She wasn’t really asking about my epistemic state, but rather about the door.

    Re 2, you asked, “if you go back and double-check the door lock, aren’t you tacitly admitting … that you didn’t know that the door was locked prior to doing the double-checking”? I interpret it as tacitly admitting that I wasn’t positioned to guarantee her that it was locked (which is what she wanted).

    Now, on to the general question, about which you’re open-minded. It’s hard for me to imagine myself flat-out saying “I don’t know whether I locked it” because I’d naturally take myself to know that I had locked it. So it would be insincere to deny that I know. That would make it inappropriate.

    The more I think about this, the more it seems to me that if someone were to genuinely insist on the second-order question — that is, when I answered by saying something about the door, they said, “No, no, I’m not asking about the door, I’m asking whether you know the door is locked” — then after I went and double-checked the lock, I can imagine coming back and saying, “See, I told you so. It was locked, and I knew it.”

  6. Hi John,

    When I said that “I don’t understand how the evidence that you offer serves to confirm the hypothesis that you want”, I didn’t mean to suggest for a moment that you can’t offer some explanation of the evidence that you offer. What I meant is that I don’t understand why you think that your explanation is better than the contextualist’s explanation, where the latter is perfectly consistent with Geoff’s intuitions about the modified High Door case he proposes.

    So here’s the situation, as I understand it: Geoff says that your view can’t accommodate the intuitive datum concerning the modified High Door case. You deny that the putative datum really is a datum, and then you propose two pieces of evidence, allegedly in favor of the claim that it’s not a datum. But the contextualist can explain those pieces of evidence too. So is the contextualist under any pressure to admit that Geoff’s datum really isn’t a datum (and, indeed, a datum that your view can’t accommodate)?

    Help?
    Ram

  7. Ram,

    The way you describe it makes it sound like a stalemate: neither side blinks. And I think you’ve given a basically accurate description of the dialectic (though I’d say I went on to offer a third and fourth bit of evidence — including the one about being insincere, and what I’d say, after double-checking, to someone being very persistent about the second-order question.)

    And here’s a fifth. Suppose with Williamson that remembering that P is a way of knowing that P. (I actually don’t think this is quite right when taken as a necessarily true generalization, but it’s very close.) Now in Modified High Door, given the very clear seeming memory of having just locked the door, I am not in a position to know that I don’t remember that I locked the door. And if I don’t know that I don’t remember P, then I don’t know that I don’t know P. Thus, by the knowledge account of assertion, I am not positioned to assert that I don’t know that I locked the door.

    So, in the context of the present debate, where all sides agree on the knowledge account of assertion, we shouldn’t accept Geoff’s claim that it’s “clear … the speaker could perfectly appropriately say ‘I don’t know that P’.”

    Convinced yet?

  8. Hi John,

    About pieces of evidence (1) and (2), I agree that it’s a stalemate (so far as I can see). Now let’s consider your pieces of evidence (3), (4), and (5).

    (3): “It’s hard for me to imagine myself flat-out saying “I don’t know whether I locked it” because I’d naturally take myself to know that I had locked it.” But in saying that you would naturally take yourself to know that you had locked it, you presuppose your own view concerning precisely what’s at issue between you and Geoff, viz., whether, in the modified High Door case, you would naturally take yourself to know that you had locked the door. So I don’t see that (3) is a problem for the contextualist.

    (4): “if someone were to genuinely insist on the second-order question — that is, when I answered by saying something about the door, they said, “No, no, I’m not asking about the door, I’m asking whether you know the door is locked” — then after I went and double-checked the lock, I can imagine coming back and saying, “See, I told you so. It was locked, and I knew it.” Well, you might say that, or you might not. But contextualism doesn’t issue any prediction concerning whether or not you can truthfully say it after you’ve double-checked the door. The contextualist can allow that your use of “knows” in this context expresses the relation of high-standards knowledge (because the costs of being wrong are high) or that your use of “knows” in this context expresses the relation of low-standards knowledge (because, even though the costs of being wrong are high, no one in this context, after you’ve double-checked, is still worried about incurring those costs). Stalemate again.

    (5): “Suppose with Williamson that remembering that P is a way of knowing that P. (I actually don’t think this is quite right when taken as a necessarily true generalization, but it’s very close.) Now in Modified High Door, given the very clear seeming memory of having just locked the door, I am not in a position to know that I don’t remember that I locked the door. And if I don’t know that I don’t remember P, then I don’t know that I don’t know P. Thus, by the knowledge account of assertion, I am not positioned to assert that I don’t know that I locked the door.” Putting aside the unnecessary and controversial step of your argument that involves “remember”, I want to ask: Why does your very clear seeming memory of having just locked the door prevent you from knowing that you don’t know that you just locked the door? Of course, make the memory clear enough, and it won’t make sense for me to go back and double-check the lock even given the high stakes. But if the memory is shy of that level of clarity, then why can’t I know that I don’t know that I just locked the door?

  9. Hey Ram,

    About 3, I don’t think I’m begging the question against anyone here. I’m just reporting how things would normally seem to me in such a case.

    About 4, I agree that the contextualist could say something along those lines.

    About 5, I’m not seeing how my claim about remembering was either “unnecessary” or “controversial.” Could you say a bit more about that?

  10. Hi John,

    Re (3): My point about what you’re “presupposing” was poorly expressed. What I meant was that I was wondering is if the way things seem to you about this case is the way they would seem to someone who didn’t have a horse in this race.

    Re (5): You argue as follows.

    (a) Since the subject’s memory is so clear, she doesn’t know that she doesn’t remember that p.
    (b) So she doesn’t know that she doesn’t know that p.
    (c) So she can’t properly assert that she doesn’t know that p.

    Right?

    My point was that the step from (a) to (b) is controversial, since it is controversial that remembering that p entails knowing that p. (Dretske’s cognitive “Cul-De-Sac” article from Mind has a nice illustration of the trouble we get into when we suppose that this entailment holds.) And the step is unnecessary, since, once we put aside the controversy surrounding the putative entailment from “remember” to “know”, the argument above is no more compelling than the following:

    (d) Since the subject’s apparent memory is so clear, she doesn’t know that she doesn’t know that p.
    (e) So she can’t properly assert that she doesn’t know that p.

    And (d) is plausibly denied, at least if the clarity of the apparent memory is not sufficient for the subject to dismiss the possibility that the door is not locked.

  11. I think I introduced a red herring by making the wife’s original and follow-up questions explicitly about your knowledge, and not the door. I agree that when somebody asks “Do you know whether P” they are typically concerned with P and not (at least not directly) with whether you know that P. So compare the end of your original High Door case with a modification that I submit is *completely* natural:

    High Door (original). She continues, “It’s rare, but sometimes the bolt hits the strike plate in such a way that it clicks but doesn’t lock. So . . . I ask again: Is it locked?” “I think it’s locked, but I’d better go back and check,” I reply. “All right,” she says.

    High Door (modified). She continues, “It’s rare, but sometimes the bolt hits the strike plate in such a way that it clicks but doesn’t lock. So . . . I ask again: Is it locked?” “I think it’s locked, but now that you mention that possibility, I don’t know. I’d better go back and check,” I reply. “All right,” she says.

    On your view, what’s going on in the first case is that the context has shifted in such a way that alethically (?) uttering “it’s locked” counts as guaranteeing that it’s locked, and you don’t meet the demanding epistemic standards required for appropriately guaranteeing that it’s locked. If this is also what explains what is happening in the second case, then you are speaking falsely when you say, “I don’t know” in that case. Alternatively, you could say that something different is going on in the second case. But the cases seem *very* similar, and I shouldn’t expect you to want to explain them differently.

    By contrast, KAA + contextualism can explain both cases in exactly the same way without saying that you’re speaking falsely in the second case. Advantage KAA + contextualism.

    Now you could maintain that my modification is *not* natural, or that we wouldn’t really say what I’ve underlined. One of your responses to Ram (comment 9 above) suggests that you would go this route; you say you would “naturally take myself to know that I had locked it”. So you’re thinking to yourself I know I locked it but saying, “I’d better go check.” Is that right? This strikes me as psychologically a bit strange, at least in the situation as you’ve described it. The only situations in which I can imagine myself doing this are (a) extremely high stakes circumstances (and if I’m remembering correctly this is Ram’s example?) like the doctor who checks which kidney needs to be removed even though he knows that it’s the right one, or (b) cases in which you yourself take checking to be unnecessary, but are doing it to please your interlocutors (I’m imagining you huffing and puffing on the way to the door, thinking Grr, I know I locked it, but she just never takes my word for it!). Neither of those things is relevant here, though: the stakes are high but not astronomical, and you yourself take it that checking is required.

    But that’s just my impression.

    In any event, if you want to go that way, I suppose there’s not much we can do to resolve the difference. What can I say — my modification seems perfectly natural to me!

  12. Ram,

    Thanks, I understand better now. I agree that the entailment doesn’t hold, but it’s only in strange cases that it breaks down, so I figured the inference was safe enough. But you’ve provided another argument, from (d) to (e), that serves my purposes well enough.

    You say (d) is plausibly denied “at least if the clarity of the apparent memory is not sufficient for the subject to dismiss the possibility that the door is not locked.” I don’t see why.

    Is it your view that if the memory isn’t good enough for the subject to know that he knows, then it’s bad enough for him to know that he doesn’t know?

  13. Geoff, yes, it’s true, my view implies that someone could be saying something literally false by self-denying knowledge in some high-risk cases.

    As you mentioned, there’s no point belaboring our difference over the naturalness of self-denying knowledge in such cases. But I’m wondering what you think about the argument that in Modified High Door: (1) you don’t know that you don’t remember, so (2) you don’t know that you don’t know, so (3) you may not assert that you don’t know.

  14. I haven’t read the paper yet (btw, it sounds very interesting, John, and I ope to be able to read it soon), but I’m quite surprised that so many of those who have commented on this thread so far seem to assume that the husband’s double-checking that the door is locked in the example is conclusive evidence that he didn’t *know* that the door is locked (before double-checking). As far as I can see, it can only be conclusive evidence that he does not know if one takes indubitability as a necessary condition for epistemic justification. And, since most epistemologists seem to adopt more permissive standards, the husband’s double-checking is only conclusive evidence that he’s not sure that (he knows that) the door is closed not that he doesn’t know that the door is closed. Am I completely off-track?

  15. Hi John,

    So you are willing to endorse (d) above:

    “Since the subject’s apparent memory is so clear, she doesn’t know that she doesn’t know that p.”

    I say that (d) is plausibly denied for certain kinds of cases in which the clarity of the apparent memory is not sufficient for the subject to dismiss the possibility that the door is not locked.

    You wonder if I say this because I am committed to the view that if the memory isn’t good enough for the subject to know that he knows, then it’s bad enough for him to know that he doesn’t know.

    My response: no. I am not committed to the (obviously false) view that, if the memory isn’t good enough for the subject to know that he knows, then it’s bad enough for him to know that he doesn’t know. (Nor am I committed to the equally obviously false view that knowledge requires indubitability.) What I am committed to, though, is:

    (X) If you can truthfully claim to know that p, then you can (at least normally) reasonably act on the assumption that p.

    (Y) If you are a normal speaker who is not conceptually or otherwise disabled from making claims about her own epistemic state, then: you cannot truthfully claim to know that p if and only if you can truthfully claim to not know that p.

    Also, I think that some tacit appreciation of the truth of (X) and (Y) is mirrored in our ordinary ways of talking and acting. So, if I know that I cannot reasonably act on the assumption that p, then I will at least take that as some good evidence that I cannot truthfully claim to know that p. And so if the husband knows that he cannot reasonably refrain from double-checking the lock on the door, he can take that as a reason to believe that he cannot truthfully claim to know that the door is locked, and so (given that he is fully aware that he satisfies the conditions specified in Y) it is a reason to believe that he can truthfully claim to not know that the door is locked. Is it good enough reason that it can enable him to know that he doesn’t know? Sometimes it is.

  16. Gabriele, I agree with you that the husband’s willingness to double-check doesn’t imply that he doesn’t know, either beforehand or at the time of double-checking. I believe that Geoff and Ram would both agree that it doesn’t imply that the husband didn’t know beforehand.

  17. I’m not sure.

    Ram said, for example:

    (1) “you’ll see me go back, double-check the door, return to the car and say, “Yeah, it’s locked.” I don’t say, “Yeah, I knew it was locked.” Of course you’re right. But isn’t that because you know that, prior to performing the double-checking and acquiring the additional evidence, you did NOT know that the door was locked (at least relative to the epistemic standards in place given the high stakes)?

    Maybe the qualification in parenthesis changes everything, but my intuition is that the guy knew that the door was locked and that, if the context shifts at all, it shifts when his wife brings to his attention an additional piece of evidence and a potential defeater of his prior justification. But since the defeater is not actually one, I would be inclined to say that (if the husband does not know about either the quirk of the lock or the string of robberies), he was sufficiently justified before the wife brought it to them to his attention and is sufficiently justified afterwards. So, it is only at one point that he might seem not to know between the wife utterance and his double-checking.

  18. Sorry, Ram, I didn’t mean to suggest that you were supposing anything obviously false. I was just puzzled about why you rejected (d), and the suggestion I made seemed to be the right sort of principle that would motivate a rejection of (d).

    Related to X, I think the husband can reasonably act on the assumption that the door is locked, at least for many actions he might undertake.

    I’m curious: do you think the husband can truthfully (or appropriately) say, “I don’t remember locking the door”?

  19. Hey John,

    I think we can truthfully say of the husband that he remembers locking the door. And if we can truthfully say it of him, then he can truthfully say it of himself.

    But, from the fact that he can truthfully say it of himself, it of course doesn’t follow that he can warrantedly say it of himself. He might not be warranted in saying it of himself because, say, he doesn’t know whether he remembers locking the door, or merely seems to remember doing so. (I assume that “remember” is a success verb, in your usage: I can’t remember seeing my wife lock the door if she didn’t actually do so.)

    Do you disagree?

  20. Sorry, my previous comment was written very quickly and I didn’t really get to the point. What I don’t understand is what exactly is meant by context here. In particular, I don’t understand whether the context in question is “external” or “internal”. If it is external (so that, say, by entering fake barn county, the standards of epistemic justification for barn-related beliefs change independently of whether or not I have any idea about the presence of fake barns in the county), then, if the husband is justified before the wife brings those considerations to his attention, he is also justified afterwards, as the “external” context does not changes in the example. What shifts, at most, is his “internal” context, i.e. the relevant features of the context he is aware of, but, if the husband didn’t know about the quirk of the lock, isn’t that change in “internal” context just a change in his total evidence? And, if he did know, was he really justified in believing the door was locked before he double-checks?

    Now, I guess, the real question here is whether the standard of epistemic justification are affected by the stakes being higher, but why does it need to be the standards of justification? If one is to proportion one’s belief to one’s evidence, for example, it would seem that the husband should have a stronger degree of belief than he had before in the door being locked if he doesn’t want his house to be burglarized but one could argue that the higher stakes require his degree of belief to be higher for him to be practically rational not his degree of epistemic justification to be higher for him to be epistemically rational.

    (As I said, I’ve not read the paper yet, so I may be missing the point of the whole discussion and wasting your time, in which case, I’m apologize, but just let me know and I’ll stop making stupid comments ;-))

  21. Ram,

    We agree that ‘remembers’ is a success term, and that there’s an important difference between truthful and warranted speech.

    Here’s an argument that the husband can neither warrantedly nor truthfully self-deny knowledge in this case. For convenience, call it “He Knows” or “HK” for short, and let ‘P’ stand for the proposition that he locked the door.

    1. The husband remembers P. (Premise)
    2. If the husband remembers P, then the husband knows P. (Premise)
    3. So the husband knows P. (From 1 and 2)
    4. So the husband can neither truthfully nor warrantedly say that he does not know P. (From 3 + knowledge account of assertion)

    The argument is valid. (I’m assuming the knowledge account of assertion in the present context, but any account that requires truth would do just as well.) We agree on 1. So that leaves only 2. Why accept it?

    We agree that there is no strict entailment from remembering P to knowing P. But the exceptions are very strange, and nothing of the sort is happening in High Door.

    Moreover, in the present case, remembering that he locked the door is the only way for him to know that he locked the door. He can’t, as he sits in the car, see that he locked it, intuit it, learn it from testimony, or feel it, etc.

    So, in the end, the question whether the husband knows that he locked the door just is the question whether he remembers that he locked the door. And I’d be willing to upgrade 2 to a biconditional, although strictly speaking all I need is the left-to-right direction.

  22. Gabriele,

    The context here is conversational context. Presumably participants are generally aware of most of the stuff relevant to the conversational context. But they won’t always be aware of all of it. So, if I’m understanding the way you’re using ‘internal’ and ‘external’ here, then the context will be a mix of internal and external elements.

  23. John,

    Your premise 1 says that the husband remembers that P. But that’s not at all what I was conceding in 24. I was conceding that the husband remembers locking the door. That’s not the same as the husband remembering THAT he locked the door. (Compare: I can see the fish swimming even if I don’t see that the fish is swimming.)

  24. Ram,

    Okay. He remembers locking it. And that memory trace is (nondeviantly) causing him to believe that he locked it. So what’s stopping him from remembering that he locked it?

    [P.S. The html list-numbering apparently didn’t work originally for my argument in 26. You figured it out anyway. But it’s fixed now.]

  25. I wish I had more time to blog!

    Re John’s 21: yes, I don’t think the husband’s willingness to double-check implies (if that means “entails”) that he didn’t know whether it was locked. For reasons similar to those Ram mentions in 20, I think a subject’s willingness to double-check is good evidence that she takes herself not to know, and that this, in turn, is good evidence that she doesn’t know. But we can have lots of other (perfectly good!) reasons for double-checking.

    Now on the remembering thing. It’s very tricky. I don’t have a clear sense about whether the husband remembers that he locked the door. I agree with Ram – he remembers locking the door. But does he remember that he did? I don’t know. There’s a considerable bit of slip-and-slide in our ordinary talk between statements like “I remember phi-ing” and “I remember that I phi-ed”. Since “I remember phi-ing” very obviously entails that you phi-ed, it would be highly misleading to assert it if you didn’t remember that you phi-ed. And the normal way that someone remembers that he phi-ed is by remembering phi-ing (maybe “as such”). (Not always, of course; I remember that I got seven stitches in my arm when I was three but I don’t remember getting them.) So typically one is assertible iff the other one is, but if we accept Williamson’s claim about FMSOs and knowledge there is plenty of space opened up for you to remember phi-ing without remembering that you phi-ed.

    But anyway we’ve subtly changed the example here. My original claim was that the husband could appropriately say (1):

    (1) I don’t know [whether the door is locked].

    For your argument in 26 to undermine this claim, it would have to be uncontroversial that he remembers that the door is locked. And — to me — it seems not at all clear that he does, *especially* when I focus my attention on the purported remembering –> knowing entailment (I’m not actually sure I believe in this entailment, but it’s theoretically attractive).

    I have another, entirely unrelated and, I think, in fact more interesting question about your paper, too. But it’s been a long day and I am getting to that “things aren’t really making sense to me any more” point… So tomorrow!

  26. Congrats on getting it half right in the right place, John! (I prefer my strict invariantism w/o the knowledge account).

    I am teaching a seminar on these issues these days. Will it be OK to distribute the linked paper for the purposes of the seminar?

  27. Hi John,

    You write: “He remembers locking it. And that memory trace is (nondeviantly) causing him to believe that he locked it. So what’s stopping him from remembering that he locked it?”

    Remember that the reason why we’re talking about “remembering that p” is that you suggested that, at least normally, remembering that p suffices for knowing that p. (That’s why you brought up remembering back in 12.) So whatever someone thinks would suffice to prevent the husband from knowing that p could also suffice to prevent him from remembering that p. So what’s stopping him from remembering that he locked it is just whatever is stopping him from knowing that he locked it (e.g., the stakes have gone up, he hasn’t ruled out some relevant alternatives, he doesn’t have the right to be sure, he isn’t entitled to act on the premise, or what have you).

  28. Ram,

    That’s certainly one way to go. But I think it creates other difficulties. In particular, it will make it more difficult to make the case that any of the stuff you mentioned really does stop him from knowing that he locked it.

    As Moore might have said: it’s just much more plausible that he remembers, and thereby knows, that he locked the door, than that any of the stuff you mentioned prevents him from remembering that he locked the door.

    So it’d be good to have some independent argument against the claim that he does remember that he locked it — one that doesn’t go from “no knowledge that” to “no remembrance that.”

  29. Geoff,

    Good catch about the slip between “locked” and “I locked it.” It’s fairly safe to assume, though, that the door is locked just in case the husband locked it, right? (We’re not wondering whether gremlins unlocked since, or whether the lock melted, or anything like that.)

    You said, “For your argument in 26 to undermine [the claim that the husband could appropriately say, “I don’t know whether it’s locked”], it would have to be uncontroversial that he remembers that the door is locked.” That seems too strong. I’m pretty sure I only need it to be more plausible than not, not uncontroversial.

  30. John,

    Ha ha — fair enough! I guess I have pretty high standards for what counts as a controversy, so for me “more plausible than not” isn’t much different from “uncontroversial”…

    Anyway, to the more substantive point. So, if the husband remembers that he locked it, he knows he locked it. And if he knows he locked it, well then it seems that he’s pretty much in a position to know that it’s locked. Thus if he remembers that he locked it, he’s (almost certainly not) in a position to assert that he doesn’t know that it’s locked, on KAA (or any “factive” account of warranted assertion, for that matter). How to respond?

    I’ve got to say that I just have no idea whether (1) is true:

    (1) He remembers that he locked it.

    It’s not even that I give it credence .5; it’s that my credence-o-meter is not giving me a reading at all. But right now I’m looking less-than-favorably upon (2):

    (2) Remembering –> knowing

    Here’s an example. As he’s leaving the office, John glances at the clock and sees that it reads 6:00. In fact it *is* 6:00, but unbeknownst to him it’s a busted clock. My strong sense is that (3) is false:

    (3) John knows that it’s 6:00.

    But let’s continue the story. When he gets home, John’s wife asks him what took him so long. He says, “Well, I left the building at 6:00, but the traffic was terrible.” Now I don’t get anything like the strong sense that (4) is false; on the contrary, it almost seems true:

    (4) John remembers that he left the building at 6:00.

    But if remembering –> knowing, then (given some very plausible assumptions) (4) entails (3).

    It’s these kinds of things that make me wonder about the entailment.

  31. You’re right, Geoff — you’ve identified one sort of case where the entailment fails. There are similar cases where the entailment [perceives that Q –> knows that Q] fails too. Arguably fake barn cases are like this.

    I may have, at one point or another in this thread, given the misimpression that I thought there was a genuine entailment from remembers to knows. But when I’m more careful, I say things like, “I agree that the entailment doesn’t hold, but it’s only in strange cases that it breaks down, so I figured the inference was safe enough” (in 17 above).

    So, in the end, I’d be content to point out that the cases we’re dealing with here (High Door and such) bear no resemblance to the counterexamples. So it’s a very safe inference from remembers to knows. In fact, we could just run the argument probabilistically, utilizing something like:

    (2*) If the husband remembers P, then it’s very likely that he knows P.

    ********************

    A bit of a tangent from the main issue, but it’s related to your nice watch example. Suppose that there was a company celebration that day. They randomly handed out prizes — watches, as it turns out. JR got the only prize that wasn’t busted — a new Rolex. (No one even noticed the other ones were busted.) He puts it on. Then when he leaves, he looks at his functioning watch, sees that it’s 6:00, gets in the car, and goes home. Do you have a stronger intuition that, in this case, JR (a) remembers that he left at 6:00? And are you any more tempted to think that JR (b) knows that he left at 6:00?

    (Forgive me for changing the name to ‘JR’. I fear it would get too confusing if we name the protagonist ‘John’!)

  32. Sorry for joining the party unfashionably late. But I’ve now read — albeit quickly — your fine paper. So, I’ll chime in. While I like the paper, I still think that the invariantist should explain the phenomena by upholding a warrant account of assertion. Very roughly, such an account has it that S’s conversational context determines the degree of warrant for believing p S must have to appropriately assert that p.

    This approach has at least three advantages over the speech act contextualism. First, I blocks whatever mileage the contextualist can get out of KAA. For according to the warrant account, KAA is false. Assertion in super-lax contexts can require less warrant than what is required for knowledge. In extremely hard contexts, assertion can require more warrant than is required for knowledge. Second, since warrant is gradable, the account need not postulate dramatic shifts in type of speech act or the like for every problem case. Thus, it is robust against the (serious) referee worry that you mention in fn. 24. Third, it contributes to an account of the relevant data — including, pace DeRose, hard context assertions of ‘I don’t know that p’ and ‘S does not know that p.’ (These are, of course, *very* distinct cases. (Btw. let me take the opportunity to scorn you — all of you — for deploying self-ascription cases when discussing attributer contextualism. But I’ll play along below : ) )

    OK — the above was just so you have an idea about where I’m coming from. Now a couple of remarks on the paper, the tread and the third point above: Early on in your paper, you say that according to invariantism

    “In the high-stakes case it might even be appropriate for him to falsely assert that he does not know P.”

    As far as I can see you don’t abandon this view in the paper (I’ve only read it once, quickly). Likewise, you seem concede to Geoff that “I don’t know that p” can be assertible in a hard context in comment 18 above. But I don’t see how this fits your view.

    First, if such an utterance is an assertion — empathetic or otherwise — and false, it violates KAA due to factivity.

    Second, if the heightened context renders the utterance a different false alethic speech act — let’s say a guaranteeing — then the epistemic constraint on that speech act is violated. For, as you note, the epistemic constraint has to be *stronger* than KAA. If so, it has to preserve factivity. If so, it will be violated. For example, your KK account of guaranteeing will be violated.

    In sum, the speech act contextualism does not seem to address DeRose’s problem for the generality objector that ‘I don’t know’ can be OK in hard contexts. Similarly for ‘S does not know’.

    Perhaps, I’m misunderstanding your stand on ‘I don’t know.’ But I agree with Geoff and Ram that (i) there can be cases in which a hard context utterance of ‘I don’t know’ — be it an assertion or another alethic speech act — is assertible and (ii) that it is an important desideratum for invariantism to account for (i). So, better to abandon KAA in favour of the warrant account of assertion : )

  33. Hi, Mikkel. It’s never too late to join!

    About your three initial points.

    First, I don’t think that invariantists adopting a warrant account of assertion (WAA) will stop contextualists from getting mileage out of KAA. The case for WAA over KAA will need to be made for that to happen.

    Second, I think I put that worry to rest in note 24. And unless you’re using ‘warrant’ stipulatively so that it’s continuously gradable, there’s no indication that (WAA + invariantism) will outperform (KAA + invariantism) here. And even if you do stipulate it, I seriously doubt that there will be effective test cases that tilt the field to WAA by exploiting the difference.

    Third … well, I didn’t very well understand the third one! But it had something to do the preservation of the intuition that sometimes you may self-deny knowledge. But I don’t deny that this is sometimes permissible. Obviously, sometimes you know that you don’t know. That’s consistent with KAA, and indeed everything else I say.

    What I do reject, though, is that the husband in High Door (and Modified High Door) may self-deny knowledge. But that’s just one case. Ask me who nowadays is the leading marine biologist studying orcas, and I’ll tell you straightaway that I don’t know — and properly so!

  34. Mikkel,

    I appreciate and deserve the scorn heaped upon me (I’m assuming I’m intended as part of the extension of your “all of you”) for using first-person cases. But I think the problem with first-person cases is that they fail to generate data that distinguish between old-school (DeRose/Cohen-style) contextualism on the one hand and more new-fangled views like IRI and relativism on the other. I take it that for John’s purposes this distinction isn’t important, so I don’t think there’s much harm in using first-person cases in discussing his view.

    Also, I would be *very* interested to see what you have in mind when you say that WAA “contributes to an account of the relevant data – including, pace DeRose, hard context assertions of ‘I don’t know that p’ and ‘S does not know that p.’” I endorse *something* like WAA over KAA and this is a problem I’ve been struggling with for a while so I’d like to hear what you think the solution is.

    John,

    First, with respect to your watch case, (a) I don’t get any different ‘read’ on whether JR remembers that he left at 6:00, but (b) my Gettier intuition about whether he knows that he left at 6:00 pretty much vanishes. It’s not really replaced with anything else. Rather, my faculty of intuition gives me the equivalent of a shoulder shrug. But I’ve long thought that I’m intuition-impoverished relative to other epistemologists (e.g., I have no intuitions about Norman the clairvoyant and it took me a while to get the “correct” fake barn country intuition) so I’m probably not a very useful source of information!

    Hopping on Mikkel’s point 3 and your earlier response to me, I really am curious about your reaction to High Door. Do you *really* think it would be inappropriate for the husband to respond, “I don’t know” upon being asked the second time? I guess in a similar vein, do you think Keith is speaking inappropriately when he responds “no” to his wife’s question about whether he knows in DeRose’s high-stakes bank case? Or that Mary (I might be getting the name wrong here) speaks inappropriately when she says that she doesn’t know in Cohen’s second airport case? Because if you do then your sense of what’s appropriate and not appropriate is very different from that of contextualists. Do you think they’re mistaking the impropriety of saying “I know P” for the propriety of saying “I don’t know that P”?

    Anyway, I can see how, if you thought it was inappropriate for the High Door speaker to say “I don’t know”, that you’d regard that case as bearing “no resemblance” to the clock example I provided. But from where I stand the two cases have a lot in common. I.e.: in each case, we have strong evidence that the speaker doesn’t know (or, not to beg the question against contextualism, that the speaker couldn’t truthfully say “I know”) but no analogous evidence that he doesn’t remember, and indeed at least some positive evidence that he *does* remember. So to me both cases *seem* to be exceptions to the remembering-knowing entailment.

  35. Hi John,

    Re 1: Yes – I think the case for the warrant account over KAA is considerable. For example, there are lots of problems with KAA’s standard reply (namely: the primary/secondary assertability move) to the obvious counterexamples to the account (very warranted false belief in low contexts, Gettier-style cases). But let’s not grind that axe here.

    Re 2: I agree that the worry is not devastating. But if it were my paper, I’d give the fn. 24 worry more consideration. I think that what you need to argue, if your account is to cover all the cases, is that we never perform type-identical alethic speech acts in low and high contexts. But given that we the speech act is non-adverbial assertion in normal (low) contexts, it seems that we never assert appropriately w/o adverbial modification in hard contexts. If that is a consequence of your view, I think it’s worrisome. But let’s not grind that axe here.

    Re 3: Let’s grind! The objection is certainly not that your view entails that it is *never* appropriate to assert ‘I don’t know that p.’ Rather, the worry is that the utterance seems appropriate in the face of high stakes + salient error possibilities that the speaker cannot rule out. The assertion (let’s suppose) ‘I don’t know that p’ *in the case-type under consideration* does not seems to me to be conversationally odd. This is, what I think is the intuitive judgment that needs to be explained. As far as I can tell, I am far from alone.

    Anyhow, the fact that that if you don’t know, then ‘I don’t know’ is appropriate is irrelevant here. In a case where S does know (by your favorite invariantist standards), S’s response to a challenge that makes a counter-possibility that S can’t rule out salient may be ‘I don’t know’. This need not be S’s response. But if it S’s response, it is not an odd one.

    Note that in DeRose’s paper you respond to, most of section 3 is devoted to this point. I think that it provides the paper’s strongest challenge to invariantism. So, I think it should be (and can be) responded to.

    (I’m not sure whether or not I am stalking the same prey as Geoff and Ram here).

  36. It looks like MIkkel and I posted almost simultaneously, though I got the edge by a minute. WRT to point 3, Mikkel, you are definitely stalking the same prey as I am. (And I definitely still want to know how you think invariantists should deal with this problem.)

  37. Hi Geoff (we posted at the same time, it seems).

    Scorn: The failure to distinguish between SSI and attributer contextualism is one thing. However, the issue does matter for the present purpose because the invariantist response to the ‘I don’t know’ problem differs somewhat from the (my) invariantist response to ‘S does not know’.

    Solution: I don’t want to jack the tread with promotions of my own account. So, drop me a line at gerken(at)hum.ku.dk.

    PS: I appreciate the company on the warrant account!

  38. Mikkel,

    I’ve not yet seen an obvious counterexample to KAA. But as you say, we can set that aside here, since the project in the paper assumes it for the sake of argument.

    You said, “I think that what you need to argue, if your account is to cover all the cases, is that we never perform type-identical alethic speech acts in low and high contexts.” No, that’s surely not required. If my account predicted that, then it would be a non-starter.

    Of course it might be appropriate all things considered to say ‘I don’t know’ even when you do know. I’m focused solely on the epistemic propriety. Since self-denying knowledge would be false (I gave an argument for that above, and I’m happy to stick by it), then it’s clearly inappropriate, in the relevant sense, to self-deny.

    I’d like to hear what sense of ‘warrant’ you have in mind such that, even with the evidence the husband has that he remembers locking the door, he is nevertheless warranted in believing that he knows that he didn’t lock the door, thus making his self-denial of knowledge appropriate.

  39. Geoff,

    You’ve switched on the fake barn cases? Too bad — you should have stuck with your erstwhile view, because Henry does know that it’s a barn!

    As I mentioned to Mikkel, I definitely agree that it could well be all-things-considered appropriate to self-deny knowledge even when such a denial would be false and epistemically inappropriate (because false and/or unwarranted).

    I guess I’m just not seeing the “strong evidence that the speaker doesn’t know” in High Door.

    Are you at all sympathetic to this line of thought? After double-checking and re-discovering that he did lock the door, the husband could then appropriately say that he knew all along that it was locked. So he did know all along, even if he wasn’t entirely sure that he did. (And if he had self-denied knowledge, he might even feel some pressure to retract that earlier statement.)

  40. John,

    I think that a cases of very well warranted but false assertions in low contexts and Gettier-style cases are counterexamples to KAA. Notably, proponents of KAA seem to recognize that some response to such cases is required. Thus, Williamson says that in such cases the knowledge norm is violated but excusably so. DeRose follows suit by drawing the primary/secondary assertability distinction. I think this ‘excuse maneuver’ is riddled with problems (and have a paper where I argue so that you are welcome to see).

    As you can gather from the above, I don’t accept that S’s false assertion is ‘clearly inappropriate’ in the relevant *epistemic* sense. But I should emphasize that we are both concerned with the specifically epistemic constraint on assertion (or alethic speech acts) and *not* an ‘all-things-considered’ condition.

    I certainly don’t think that the warrant account should appeal to S’s higher-order warrant that he knows that he *didn’t* lock the door. What makes ‘I don’t know that p’ appropriate in a hard context is typically not higher order warrant regarding one’s own knowledge.

    Rather, the idea is – very roughly – that in the hard context S’s warrant is contextually inadequate for asserting that the door is locked. That is what I get from the warrant account. Since S can’t assert that the door is locked, S can’t assert the logically stronger ‘I know the door is locked either.’

    That does not, as DeRose notes, entailt that ‘I don’t know’ is appropriate. However, one – among several – ways for S to convey the complex fact that his warrant does not address the salient error-possibility is to *reverse* the previous knowledge self-ascription and say ‘I don’t know that it is locked.’ That does not sound odd because he is thereby responsive to the error possibilities salient (and the stakes) in the hard context.

    Ordinary speakers who don’t have terms like ‘warrant’ available will often use ‘knows’ in a manner that is literally inaccurate to convey complex epistemic stuff. In particular, when re-evaluating an epistemic judgment that ‘I know that p’, one will often use ‘knows.’ So, when ‘I know that p’ becomes unassertable because ‘p’ does, ‘I don’t know that p’ is quite natural. This is just a way for S to convey that he recognizes the stakes and the salient uneliminated counter-possibility that compromises his previous self-ascription of knowledge.

    The third-person case of ‘S does not know that p’ is somewhat harder for the invariantist. But I think it can be accounted for in a relevantly similar manner (and I have another paper arguing so that you are also welcome to see : )

    The main point I wanted to make, however, is threefold: (i) S’s ‘I don’t know’ response in the light of a hard context (of the sort contextualists appeal to) does seem natural (ii) this should be accounted for by the invariantist (iii) the speech act contextualist invariantism seems to be unable to do so.

  41. Gerken,

    So are you now agreeing that the husband speaks falsely when he self-denies knowledge?

    You said, “I certainly don’t think that the warrant account should appeal to S’s higher-order warrant that he knows that he *didn’t* lock the door.” So when it comes to higher-order assertions, what’s relevant isn’t your warrant for asserting *the proposition in question*. That’s very surprising. If this is how the view works — a disparate account of first-order vs. higher-order assertability — I’m inclined to count that as a big strike against it.

    As for (iii) at the end of #46: nothing in my view prevents me from saying, along with you, that the husband in this case could self-deny knowledge “in a manner that is literally inaccurate to convey” something else. I’ve been saying all along that the self-denial is literally false. But I of course agree that speaking falsely might nevertheless be useful in all kinds of ways.

    Good night (morning!) there in Copenhagen.

  42. Hi again John,

    Yes — I’ve been assuming throughout (see e.g., post # 38) that all parties, expect the contextualist ones, agree that that the self-denial in the hard context was false. Indeed, that is why I think there is a challenge to invariantism to explain why it is conversationally ok and, moreover, that this challenge is hard to meet for someone who upholds a factive epistemic requirement on assertion (guaranteeing etc).

    My remark that you cite concerns the case in question. I don’t think S in the hard context has a positive measure of warrant for believing that he knows hid didn’t lock the door. That would require a positive measure of warrant for believing that he did not lock the door. He does not have such warrant. (Note that if I he did, the case would fail to parallel, LowDoor and so it would fail to be member of a good case-pair for contextualism).

    Rather, S’s warrant for believing the imbedded p, that he locked the door, remains constant in Low and High. But in high that degree of warrant is in High not strong enough to render the assertion that p or any assertion that entails it appropriate. In this case, ‘I don’t know’ is an effective, albeit false, way of conveying that his warrant does not rule out the raised challenge. Of course, this is, as I said only a rough sketch of the account. The details are very hard, I think, to get right. Differences between first and third person ascription become important etc.

    Back to your paper and my main worry with it: You say that you can just agree “that the husband in this case could deny self-knowledge in a manner that is literally inaccurate to convey something else.” I still fail to see how you can embrace this approach given that you uphold a factive account of assertion (and stronger speech acts). Since the declarative utterance is false, it violates KAA (if it is an assertion). So, crucially, it will also violate the relevant stronger epistemic condition (granting you that it constitutes another speech act such as guaranteeing).

    So, although the ingenious speech act move addresses DeRose’s “lethal” argument of 2.4, it does not address his ‘I don’t know’ demise of 3.1. I think the latter is more challenging (since I find KAA antecedently implausible). Fortunately, adopting a warrant account instead of KAA addressed the two challenges in a uniform manner.

  43. Mikkel,

    What’s wrong with this: the husband’s self-denial of knowledge is literally false but nevertheless a useful way of indicating that he’s willing to double-check to reassure his wife?

    What, exactly, in [KAA + speech act contextualism] prevents me from saying this?

  44. Hi John,

    I would have thought that to reconcile any view with KAA, one would have to accept that KAA explains intuitive judgments regarding conversational propriety (that are not about relevance etc.) Roughly, if an assertion is judged to be appropriate (epistemically speaking), it is in accordance with KAA. If it is judged to be inappropriate (epistemically speaking), it violates KAA.

    However, the intuitive judgment (that is not about relevance etc.) is that ‘I don’t know’ is appropriate in the hard context. So, by reconciliatory commitment noted above, it has to be explained by KAA. But if so, ‘I don’t know’ cannot be false. So, the false-assertion-that-coveys-something-true account of the propriety does not seem to be available to someone who is in the business of reconciling KAA with his view.

    My worry (and Geoff’s) is not that the speech act contextualism makes the matter worse for the invariantist who is committed to KAA. It is that it does not seem to help.

  45. Thanks, Mikkel. I deny that any account has to explain the epistemic propriety of the self-denial, because intuitively it’s not proper. What you’re insisting on is simply not an uncontroversial datum. But I won’t belabor the clash of intuitions. I’m willing to rest my case for the epistemic impropriety of the self-denial on the arguments I gave above.

  46. Hi John,

    Yeah, I still don’t know about the fake barn case; that’s why I put scare quotes around “correct.” I now have the standard intuitive response to it. I was helped to get with the program in this regard by DeRose’s suggestion to imagine that Henry’s been seeing fake barns all morning and thinking, in each case, “There is a real barn over there.”

    With respect to the strong evidence that the husband doesn’t know, I should have been more careful. There is strong evidence that the husband could truly say that he doesn’t know. This evidence is that the husband could appropriately say that he doesn’t know. In the absence of any other explanation for this fact, we ought to assume that it indicates that he could truly say it.

    Now if you want to deny that the husband’s utterance *would* be true, you need to come up with some explanation for why it’s appropriate. Mikkel appears to be suggesting that if we replace KAA with a non-factive norm we can make it come out epistemically appropriate (I think that’s what he’s suggesting in 49 — maybe I’m misunderstanding, though). That would be cool. But if you want epistemic propriety in the relevant contexts to require truth (which I think I do, though I think the norm shifts with context), you’re going to have to couch the explanation in terms of something else — the utterance is going to be epistemically inappropriate but *somehow* appropriate. Your suggestion in 50 is a start, but given that there are lots of true things the husband can say to indicate his cooperation (“I’m not positive,” “I’ll check,” “Lemme go make sure,” etc.) it would be odd for him to say something false (and hence epistemically inappropriate) for that same purpose. Now maybe you think it *would* be odd for him to say “I don’t know,” but now we’re just up against our initial divergence in intuitions.

    We may have hit the point of reflective disequilbrium.

  47. Hi John,

    Thanks for your article, which I thoroughly enjoyed. I’m not an epistemologist, so my apologies if my question strikes you as naive/confused.

    I’m having trouble clearly differentiating the (putatively invariant) truth-conditions on knowledge ascriptions from the (shiftly) epistemic norms governing alethic speech acts.

    In your invariantist interpretation of High Door, you write “Now uttering ‘It’s locked’ is to perform an alethic speech act higher up the credibility and epistemic indices… on my view what shifts is which speech act one performs in uttering a declarative sentence… The standards for appropriate assertion and knowledge remain invariant.” (all p. 89) But how could guaranteeing that P shift us up the epistemic index without ipso facto shifting the truth-conditions on knowledge ascriptions?

    Happy holidays!

    Yours,

    Confused Philosopher

  48. Hello CP (Anon 12:43 am),

    Happy holidays to you to! I’m glad you liked the article.

    On the view I proposed, the epistemic norms of alethic speech acts aren’t shifty. They remain constant across contexts. So assertion always requires knowledge, and guaranteeing always requires, say, knowledge of knowledge.

    Does this help clear things up?

Leave a Reply

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