the boundaries of language

It's generally seen as an objectionable consequence of an account that it entail a robust form of doxastic voluntarism. For instance, take epistemic deontologicalism, which has it that we should understand epistemic justification in deontic terms such as 'requirement', 'blame' and 'obligation'. This view has been criticized on the grounds that it is committed to the appropriateness of assessments of beliefs in these terms, even though such assessments seem to presuppose that we are responsible for our beliefs - that we have voluntary control over what we believe. It is taken as pretty obvious that we have no such control.

Feldman, Heironymi, and others have offered a partial response to this argument by questioning whether the appropriateness of assessing beliefs in deontic terms really does commit us to doxastic volutarism. My concern here is not with this dialectic, however, but with a parallel issue that arises once one adopts certain aspects of Grice's philosophy of language. In 'Meaning', Grice writes:

'Suppose I discovered some person so constituted that, when I told him that whenever I grunted in a special way I wanted him to blush or to incur some physical malady, thereafter whenever he recognized the grunt (and with it my intention), he did blush or incur the malady. Should we then want to say that the grunt meant(nn) something? I do not think so. This points to the fact that for x to have meaning(nn), the intended effect must be something which in some sense is within the control of the audience, or that in some sense of "reason" the recognition of the intention behind x is for the audience a reason and not merely a cause. [...] It looks, then as if the intended effect must be something within the control of the audience, or at least the sort of thing which is within its control.'
(p221 in Studies in the Way of Words)

Now, Grice's account of imperatives looks okay here; the view he comes to in the William James Lectures is that the intended effect should be that one's audience forms a particular kind of intention, and a sufficiently robust voluntarism about intention-formation doesn't seem problematic. But in the case of 'informative' utterances, Grice suggests the intended effect is that a particular belief is induced in one's audience. So, assuming that doxastic voluntarism really is a bad consequence, we seem to have a straightforward objection to Grice on our hands.

Some contemporary Griceans have suggested weakening the intended result in the case of informative utterances to the following: one intends one's audience to entertain a particular thought, rather than to form a particular belief. (For example, see Ray Buchanan and Gary Ostertag's contribution to the Mind volume celebrating the centenary of 'On Denoting'.) Here again a sufficiently robust voluntarism doesn't seem too objectionable - at the very least, the weakening seems to do better with respect to the problem I've sketched than Grice's original suggestion, so some progress seems to have been made.

But the problem still seems to remain for Gricean accounts of assertion. Consider Bach and Harnish's familiar Grice-inspired proposal (Linguistic Communication and Speech Acts: 42):

In uttering e, S asserts that p if S expresses:

1. the belief that p, and

2. the intention that her audience believe that p.

For the purposes of Gricean accounts of meaning, saying and communicating, it might be satisfactory to suggest that the intended result is that one's audience entertain a particular thought. But I'm worried that an attempt to amend Bach and Harnish's (2) along these lines is too weak - that something of the force of assertion, in contrast to mere saying, gets lost. Clearly some work needs to be done clarifying and defending this last point, but hopefully the basic idea is clear enough for now.

Additionally, if this inchoate line of thought is to be developed into an objection to Gricean accounts of assertion, Grice had better be right that the intended result must be under the audience's voluntary control, and doxastic voluntarism had better be genuinely problematic. So there are a lot of interesting issues to be considered before the force of the objection I'm trying to mount can be properly assessed.

Labels: Assertion, Philosophy of Language

I just got back from a pretty spectacular weekend in St Andrews, during which I saw lots of people, and attended eleven - count 'em - eleven presentations. On Friday morning my good friend Marcus Rossberg, who kindly put me up for the bulk of the time I was there, discussed the options one has for reconciling inferentialism about the logic constants with the non-conservatism of third- over second-order logic.

Jonathan Schaffer kicked off the Arche Assertion Workshop that evening with a talk arguing that if you put together the Stalnakarian thought that what an assertion does is aim to reduce the contrast set with the Williamsonian thought that knowledge is the norm of assertion, the outcome is a contrastivist notion of knowledge; in slogan form, knowledge in the image of assertion is contrastive knowledge.

The next morning, Ofra Magidor presented joint work she's done with John Hawthorne which aims to show that the non-transparency of certain notions appealed to in Stalnaker's framework prevents him from being able to retain the following principle, we they label Uniformity: In cases of rational communication, the same proposition is asserted at each world in the context set. (Very roughly indeed: A notion N is transparent just in case Np implies NNp and ~Np implies N~Np.) Jessica Brown's talk noted that the thesis the knowledge is sufficient for warranted assertion is frequently appealed to, but never defended. She explored some arguments analogous to Williamson's arguments for the necessity claim, and suggested that none of them were good enough. In the questions, I tried to argue that her point is even more significant than she'd suggested, since it's actually the sufficiency thesis that Williamson appeals to when defending the necessity condition in the face of cases where a subject asserts something false but on the basis of impeccable evidence. These seem to be cases in which we just as asserter to have asserted well in some sense, yet he clearly did not know that which he asserted. Williamson writes (257, emphasis added):

'The case is quite consistent with the knowledge account. Indeed, if I am entitled to assume that knowledge warrants assertion, then, since it is reasonable for me to believe that I know that there is snow outside, it is reasonable for me to believe I have warrant to assert that there is snow outside. If it is reasonable for me to believe that I have warrant to assert that there is snow outside, then, other things being equal, it is reasonable for me to assert that there is snow outside. Thus the knowledge account can explain the reasonableness of the assertion.'

The same line will get run in response to Gettier cases. So, I suggested, it seems like it may be difficult to defend necessity without appealing to sufficiency - but, as Jessica pointed out, sufficiency has never been remotely adequately defended.

In the afternoon, John MacFarlane explored four different styles of accounts of the nature or purpose of assertion; the Stalnakarian account, the more or less Gricean account, Williamson-style accounts which hold that assertion has a constitutive norm, and commitment accounts. He explored how each of these accounts could make sense of the phenomenon of retraction, and whether they are genuinely competing, ending with a tentative suggestion that the kind of commitment account he has defended in print can make the best sense of retraction, and may enable one to recapture what each of the other accounts seems to have right. Lastly, Jason Stanley argued that defenders of the knowledge norm, his earlier self included, aren't free to appeal to Moore-paradoxical sentences of the form 'p, but I don't know so' to support their view. The reason is that sentences of the form 'p, but I'm not certain that is so/it's not certain that is so' are, as Unger pointed out, just as weird, yet seem to require even more demanding norms of assertion.

I only attended three out of the four talks on the Sunday, the final day of the workshop. Sandy Goldberg kicked things off, arguing that accepting that assertion has an evidential norm can help us explain certain phenomena in the epistemology of testimony. Jennifer Lackey also argued against the sufficiency of knowledge for warranted assertion, though in the question session most of the audience seemed convinced that her counterexamples should really be regarded as cases in which one has violated Gricean maxims. Bob Stalnaker finished things up with a talk on how to accommodate self-locating beliefs within his general framework.

All in all, this was about the best conference I've ever attended. The talks I went to were all very interested, and I learned a lot from them, and the whole thing was superbly organized. Thanks to Jessica Brown, Herman Cappelen, and last but not least, Sharon Coull for putting together such a great event.

Yesterday we had two further talks in the Basic Knowledge seminar from Baron Reed and Jennifer Lackey. Baron tried to defend a new argument for scepticism which would be immune to the charge, commonly leveled against standard sceptical arguments, of having presupposed an internalist picture of knowledge and justification. Jennifer defended and elaborated her 'justificationist' view of what to do in the face of peer-disagreement.

All in all, a fantastic weekend. As well as attending the talks, I got to see a lot of old friends, and meet a bunch of new people. Now I just have to recover in time for Duncan Pritchard's workshop on Basic Knowledge and Scepticism this coming weekend...

Labels: Assertion, Conferences, Norms of Assertion

I found myself very unhappy with how I put my earlier objection to Lackey's treatment of Moore's Paradox, and I managed to put the worry much better in conversation with Sandy Goldberg this weekend. Towards the end of the earlier post, I wrote:

'it may be that Lackey conceives of the challenge to the reasonable belief account as follows; the account rules as permissible assertions which strike us as paradoxical and/or odd'

I now think this is spot on. Likewise, Lackey takes herself to be playing defense against the following objection; we hear assertions of lottery propositions as conversational fouls, but it's reasonable to believe them, and so Lackey's norm doesn't rule them as infelicitous. The heading that the section in which she discusses these issues is called 'counterexamples to the RTBNA', (i.e. the reasonable to believe norm of assertion), the conclusion of which is:

'neither lottery propositions nor Moorean paradoxes--the two central objections to norms requiring anything less than [the knowledge norm]--pose a problem for the RTBNA.'

But now my worry can be stated. If all the evidence Williamson adduced in favor of the knowledge norm can be explained away with appeal to a principle to the effect that we shouldn't assert what we know will mislead our audience, what reason do we have for thinking there's a norm of assertion at all? The argument for the existence of a norm is an inference to the best explanation, but what's left for Lackey's RTBNA to explain?

Labels: Assertion, Epistemology, Norms of Assertion

I've been reading Jennifer Lackey's new book 'Learning from Words' in preparation for meeting her this weekend, and I've really enjoyed it so far. It's superbly written, and it's both a useful overview of the main issues at stake in the recent debate over testimony, and an interesting and original contribution to that debate. The book is based on a series of papers Lackey has published since 1999, but it's useful to have the material collected together, and there's a significant amount of new stuff and revision.

Here I want to raise a worry about her explanation of Moore-paradoxical sentences in the chapter on norms of assertion. Williamson, taking his cue from Unger, has noted that the knowledge account of assertion offers a particularly elegant account of why there's something wrong with asserting sentences of the form 'p, but I don't know that p'. But an issue immediately arises about what it is that needs explained. Here are three answers that one might endorse:

1. The usual answer is that what needs an explanation is why we hear assertions of such sentences as contradictory or paradoxical, despite the sentences themselves being perfectly consistent.

2. Following DeRose's influential '92 paper on epistemic possibility, some participants in the debate have suggested that all that we need to explain is why such sentences sound odd. (See Douven's 2006 Phil Review article, for instance.)

3. Lastly, we might think that all that really needs accounting for is that such sentences ought not be asserted.

It seems that the tasks involved get less demanding as one moves from (1), through (2), to (3). And, relatedly, it seems like to have successfully carried out one of these tasks is also to have carried out the tasks below it, but not above.

So what does Lackey take to require explanation? It seems that she initially takes it to be (1). Here's how she introduces the topic:

'Moore famously noted that assertions of these general forms--that is, of either the form "p, but I don't believe that p" or of the form "p, but I don't know that p"--seem quite paradoxical.' (130)

It is this paradoxicalness that Lackey takes Williamson and other proponents of the knowledge account (KNA) to trying to explain:

'Advocates of the KNA are in an excellent position to account for the paradoxical nature of asserting Moorean sentences.' (130)

Lackey wants to defend a rival to the KNA, the reasonable belief account. Moorean sentences seem to be a problem for this account, since Lackey herself has described cases in which it is reasonable for a subject to believe p, and that they don't know p. These are cases of selfless assertion, in which a subject asserts p on the basis of great evidence - evidence which makes it reasonable for her to believe p - and yet isn't willing or able to bring her beliefs into line with the evidence, and so either fails to form any belief on the matter, or continues to believe ~p.

In response, Lackey argues for a second norm, the NMNA, which rules an assertion as improper if it is reasonable for the asserter to believe that it will be misleading in the context of utterance. (She suggests this second norm might get subsumed by Gricean maxims; I think it would strengthen her position if she were so.) It's the NMNA that allows her to deal with Moorean assertions:

'the NMNA rules out the permissibility of asserting such paradoxes in most circumstances, even when they involve selfless assertions,' (134)

How? Well, consider one of Lackey's examples of selfless assertion. Sebastian is a well-respected doctor who has seen excellent evidence showing that there's no link between vaccinations and autism. However, his own child was recently diagnosed with autism not long after receiving vaccinations. As a result, he cannot bring himself to believe that there's no connection, despite recognizing that the evidence all points in that direction (he recognizes that the case of his own child isn't evidence to the contrary, and that his reluctance to belief that there's no connection stems from his emotion state). When asked by a patient whether there's a connection between vaccination and autism, he sets his own feelings to the side, and asserts on the basis of the excellent evidence that there is not. Of this case, Lackey writes:

'such an assertion is most likely to lead Sebastian's hearers either forming no relevant beliefs at all, because they are confused by its oddity, or to forming false beliefs, because they are trying to plausibly explain its oddity away.' (134)

Notice the shift. Not only are we now engaged in task 3 rather than task 1, Lackey's account, far from providing an explanation of the oddity of assertions of Moorean sentences (task 2), actually helps itself to the fact they sound odd in order to explain why they should be ruled impermissible by the NMNA.

Now, it may be that Lackey conceives of the challenge to the reasonable belief account as follows; the account rules as permissible assertions which strike us as paradoxical and/or odd. But if so, we should wonder why this lack of ambition wasn't flagged more clearly, and we should wonder whether we can rest content with an account of assertion that can only help us with such an unambitious (type 3) explanation of the Moore cases, when there are rival accounts which enable us to offer type 1 and type 2 explanations.

Labels: Assertion, Epistemology, Norms of Assertion