‘Knowledge’ and its limits

First at Fake Barn Country and now in a paper Allan Hazlett has been arguing against the claim that knowledge is factive. Since the paper refers rather generously to my work, it would be churlish to criticise it too strongly. Nevertheless…

Allan has two arguments against factivity. The first is a variant on Sartwell’s argument that knowledge doesn’t require justification. Loosely speaking, Sartwell’s argument is that knowledge is the aim of inquiry, the aim of inquiry is true belief, so knowledge is true belief. Allan’s argument is that knowledge is the aim of inquiry, the aim of inquiry is justified belief, so knowledge is justified belief. (Allan goes on to analyse justified belief as reliable belief.) I’m more interested in Allan’s second argument, so I’ll leave that one to the readers.

The second argument turns on some ‘ordinary’ usages of ‘know’. Here are two examples from Allan, and one from Ian McEwan.

(1) I knew I was going to die, and then I was miraculously rescued.
(2) Before Copernicus, everyone knew that the sun revolved around the earth.
(3) He knew from experience that unless he made a formidable effort, a pattern was waiting to impose itself: a polite enquiry would elicit a polite response and another question … He had asked her about tea making. One more like that, and there would be nothing he could do … Rather than tolerate more silence he settled after all for small talk, and began to ask, ‘Have you lived here long?’ But all in a rush she spoke over him, saying, ‘How do you look without your glasses? Show me please.’

The last example is via one of my favourite ever short papers, Richard Holton’s Some Telling Examples: A Reply to Tsohatzidis. Richard claims that these cases are assertable falsehoods, on the model of (4) (by Billy Bragg)

(4) I saw two shooting stars last night
I wished on them, but they were only satellites
It’s wrong to wish on space hardware
I wish, I wish, I wish you cared

If (4) is true, then two shooting stars have to be satellites, which isn’t the case. But (4) can still be asserted (or at least sung) in the context of telling a story. What is necessary is what Richard calls ‘protagonist projection’. Here is his description of the phenomena.

I suggest that these sentences work by projecting us into the point of view of the protagonist; let us call the phenomenon protagonist projection. In each case the point of view into which we are projected involves a false belief. We describe the false belief using words that the protagonists might use themselves, words that embody their mistake. So we deliberately use words in ways that do not fit the case.

(It’s arguable I suppose that the last sentence is wrong and really (4) is elliptical for something less poetic like “I thought I saw two shooting stars last night…”. If so, (1) (2) and (3) should all be thought to be elliptical as well. I’m sceptical of positing elided material to get out of jams like this, so I’ll stick with Richard’s way of putting it, but I don’t think much here turns on the issue.)

Now it’s striking that in Allan’s case, and in McEwan’s, the natural background to asserting the sentences is that the speaker is trying to get us to see things from the point of view of a protagonist. (Maybe that isn’t quite as true in (2), but I find (2) extremely marginal.) So it’s plausible that in this kind of setting, the norm governing assertion is that the sentence be believed by the protagonist. Hence these cases are not evidence that knowledge is not factive.

Richard’s paper is a defence of one of the neatest little observations in formal semantics, Lauri Karttunen’s and Zeno Vendler’s observation that while ‘tell’ is not normall factive, when ‘tell’ is followed by an embedded question it is factive. To see this, imagine that although the party is in Ithaca, Billy tells Suzy that it is in Syracuse. Then consider the following questions

(5) Did Billy tell Suzy where the party is?
(6) Did Billy tell Suzy that the party is in Syracuse?

The answer to (5) is no, and the answer to (6) is yes. Similarly, (7) is false although (8) is true.

(7) Billy told Suzy where the party is.
(8) Billy told Suzy that the party is in Syracuse.

This is a nice point, and actually one that is of some assistance to the opponent of factivity for knowledge. For it provides a non ad hoc reason for thinking that knowledge-that claims are not factive, even though knowledge-wh claims are factive, and that ‘knows’ means the same thing when followed by a that clause and when followed by an embedded question. Now I don’t think factivity fails, but I think Kartunnen’s and Vendler’s observations show that we can’t argue for factivity from the behaviour of knowledge-wh.

Having said all that, I have one favour to ask readers. Also on Fake Barn Country there is a discussion of whether ‘knows’ is ambiguous. Does anyone know what the best literature on this question is? In his anti-Donnellan paper Kripke attacks Hintikka for holding such a view, but doesn’t actually cite anything by Hintikka to back up this attack. In ‘Knowledge and Belief’ Hintikka writes as if he’s presupposing that ‘knows’ is ambiguous between a strong and a weak sense (and perhaps many senses in between – I think scholars of the history of epistemic contextualism should be looking at this book for possible early traces of contextualism). But I couldn’t find a place where he argues for this. I also seem to recall some contextualists arguing against ambiguity views on the ground that there should be many more than two senses for ‘knows’, but I can’t remember who says that. Any tips for where to look here would be much appreciated!

8 Replies to “‘Knowledge’ and its limits”

  1. I also seem to recall some contextualists arguing against ambiguity views on the ground that there should be many more than two senses for ‘knows’, but I can’t remember who says that.

    Hi, Brian. Though I don’t argue the point at length, you could be thinking of me here; see section 7 of “Responding to Skepticism” and section 7 of “Contextualism: An Explanation and Defense” (on-line versions of these are available at http://pantheon.yale.edu/~kd47/OLP.htm )

    On ambiguity theories (strong vs. weak senses of “knows”), I always cite Norman Malcolm’s 1952 Mind paper, “Knowledge and Belief.” That’s a bit before Hintikka, I think. I too would be interested to learn if there are earlier versions of such a view.

  2. There definitely is ‘protagonist projection’ (those aren’t scare quotes, they’re just “I’m not pretending like I’d heard of protagonist protection before” quotes); one of my favorite examples is when secular persons say that such-and-such a day is the “holiest day of the year.” Presumably they don’t think that such-and-such a day really is the holiest day of the year, but just that that is what the people described believe.

    Two replies:

    1. This one’s not going to convince many, I can tell already, but: ‘saw’ isn’t factive either. After all, the ‘aim of inquiry’ argument that I endorse works for seeing as well, at least when we think of seeing as a way of knowing. (Haven’t thought about this a lot, so this is pretty knee-jerk.)

    The same kind of ordinary language evidence that ‘knows’ isn’t factive is there for ‘sees’, so I’d be prepared to say that they stand and fall together. Which brings us to …

    2. Are our non-factive uses of ‘knows’ explained by PP? This is something I want to think about more, but my first thought is that they don’t because you don’t get the right answers with requests for straight talk:

    A: Yom Kippur is the holiest day of the year.
    B: What? I didn’t think you believed in any of that stuff …
    A: Well, I mean that according to Jewish people it’s the holiest day of the year.

    A: I knew that I was going to die, but then I was miraculously rescued.
    B: What? You knew you were going to die? But you just said you were rescued.
    A: …

    It’s not obvious that A needs to retract her first claim to keep conversationally kosher. In talking to people about this, I’ve heard a fair number of people say that the mountain climber surely did know, that people before Ptolemy surely did know, etc., and non-philosophers often wonder what the fuss is about. (This is a species of the same genus that the phenomenon of how hard it is to convince undergrads that knowledge is factive is a species of.) If we’re speaking falsely when we use ‘knows’ non-factively, we don’t know it. But it seems like usually when we’re doing PP, we know it.

    But this is a serious concern Brian, thanks. I am prepared to bank quite a bit, however, on the ‘aim of inquiry’ argument…

  3. Hi Allan,

    I confess to not having looked at the paper. I’m just going by the above dialogue snippet, so you may have answers to this already. But does it matter that, when the non-philosophers and students say that the mountain climber surely did know, that people before Ptolemy surely did know, etc., they immediately retract when asked the following question:

    So, the mountain climber knew whether he was going to die in 5 minutes?


    So, they knew whether the earth revolved around the sun?


    So, they knew the shape of the earth?

    All of these receive (in my experience) emphatic “no“s. Is this not evidence that “know” is factive?

    If you’ve already dealt with this sort of thing, feel free to tell me to shove off.


  4. Jeremy,

    The first two of these involve embedded questions, and the third involves a so-called concealed question, and it’s certainly consistent with the claim that ‘know’ is not factive that it is factive in these cases. That is, it is consistent to claim that ‘know’ behaves just like ‘tell’, non-factive with that-clauses, factive with embedded/concealed questions.

    Having said that, I think the embedded questions cases show something. It should be uncontroversial that ‘know’ is factive with embedded questions. (I don’t know what it would even mean to have a non-factive usage of ‘know’ with an embedded question.) But we can have protagonist projection in these cases, even non-obvious protagonist projection. And that makes it (even) more plausible that protagonist projection is what is going on with Allan’s cases.

  5. In “Socrates’ Disavowal of Knowledge,” The Philosophical Quarterly, Vol. 35, No. 138. (Jan., 1985), pp. 1-31, Gregory Vlastos claims to find an ambiguity view in Socrates (as between “certain” knowledge and “elenctic” knowledge). Apology 20D6-E1 is one useful text. It’s probably worth noting that Vlastos overlapped substantially with Malcolm at Cornell. But Socrates did not.

  6. Thanks, David. Gary Ostertag e-mailed me last year to call my attention to that Vlastos paper. What’s really interesting about it (& Gary pointed this out to me) is that, while V ends up using some form of “two senses of ‘know’” theory to reconcile Socrates’ various claims, when V is warming up for his interpretation, and speaking in his own voice, at the paragraph at pp. 11-12, he sounds a lot like a current contextualist, rather than a 2-senses guy.

  7. I’m not sure how this connects the rest, but is it clear what’s going on with with expression ‘as far as S knows’? “It” is non-factive (‘As far as we knew, he was the right man for the job’, ‘Nothing ever happened here, as far as anyone else knows’) and one can easily make concessive as-far-as-knowledge attributions (‘As far as I know it’s on Sunday, but I could be wrong’). But to say that p, as far as you know, isn’t just to expression a mere opinion that p; saying that as far as S knows, p, is to say that S bears some epistmeically significant relation to p. (As in ‘As far as I know!’ as a reply to ‘Is that true?’, where it is an assertion of one’s epistemic non-blameworthyness with respect to regarding p as true.)

  8. I didn’t realise anyone else challenged the factivity of knowledge explicitly, phew! An example of an implicit non factive user of knowledge is Karl Popper, who often and freely talks of “scientific knowledge” but says that most scientists don’t think of their theories as being “true”. The protagonist projection doesn’t quite work here, but the know that/know whether difference does. I think Popper would assent to something like the following: We know that e=mcsquared, but we don’t know whether this is true. I think the analysis of knowledge should come out that knowledge is the end of enquiry, and so is truth. However, enquiries can always be reopened. When we look back at a previous end of enquiry, we can choose to refer to the state of enquiry then, or the state of enquiry now. This is not an ambiguity of the term “know”, just a shifting in the background discourse. When we won’t to refer similtaneously to two states of enquiry, then truth and knowledge come apart. Hence Billy Bragg and the Mountaineer. At that point in the enquiry, the bets were off, I was going to die. Now however, the book has reopened as to whether I will die or not.

Comments are closed.