This is an Author's Original Manuscript of an article whose final and definitive form, the Version of Record, has been published in the Australasian Journal of Philosophy Volume 90, Issue 3, 2012, copyright © 2011 Taylor & Francis, available online at: http://www.tandfonline.com/10.1080/00048402.2011.611150.
SINCERELY ASSERTING WHAT YOU DO NOT BELIEVE
Alexander R. Pruss
I offer examples showing that, pace G. E. Moore, it is possible to assert ‘Q and I don’t believe that Q’ sincerely, truly and without any absurdity. The examples also refute the following principles: (a) justification to assert p entails justification to assert that one believes p (Gareth Evans); (b) the sincerity condition on assertion is that one believes what one says (John Searle); and (c) to assert (to someone) something that one believes to be false is to lie (Don Fallis).
Key words: assertion, language, sincerity, lying, belief
According to Moore [1942: 543], it is absurd to assert ‘Q and I don’t believe that Q’. According to Evans [1982: 225-6], justification to assert that Q entails justification to assert that one believes Q. Searle [1969] holds that the sincerity condition on assertion is that one believes what one asserts. Fallis [2009] analyzes lying as assertion of something one believes to be false. [1] I offer counterexamples to all four claims.
Recently, Turri has offered a counterexample to the Moore and Evans principles. Consider Ellie the eliminativist who has a justified belief that propositional attitudes are not real. And so Ellie says both that eliminativism is true and that she doesn’t believe it. Turri grants that eliminativism is false, but nonetheless Ellie’s statement is justified and sincere, and does not appear to suffer from Moorean absurdity.
Turri’s example, however, is less than completely convincing. First of all, believing in eliminativism is thought by many to be an absurd sceptical attitude or even ‘pragmatically incoherent’ [Baker 1988]. The eliminativist has a response to this charges of absurdity, but Turri was assuming eliminativism as a whole is false, and so it may well be—at least Turri has not shown it is not so—that it is absurd to believe eliminativism. It could, for instance, be absurd to believe eliminativism because sceptical attitudes are in general absurd. Certainly, ordinary folk think sceptics to be absurd people. And, similarly, assertions of eliminativist-inspired ‘I do not believe’ claims could be held to be absurd. Moreover, the obviousness of our having beliefs might defeat any argument for eliminativism so that no one is justified in being an eliminativist.
I do not know if such objections against Turri succeed, but they complicate his case. Moreover, the Moore principle is so plausible on its face and the issues surrounding professions of eliminativism are sufficiently complex that one might reasonably argue by modus tollens: if affirming belief in eliminativism is not absurd, then Moore sentences aren’t always absurd; but Moore sentences are always absurd; so, affirming belief in eliminativism is absurd.
Turri’s example also leaves intact Searle’s sincerity principle and Fallis’s account of lying while my examples refute those as well.[2]
The following are the principles to be counterexampled[3]:
Moore: It is absurd to assert ‘Q and I don’t believe that Q’.
Evans: Justification to assert that Q entails justification to assert that one believes that Q.
Searle: One’s assertion that Q is sincere only if one believes that Q.
Fallis2009: If one asserts that Q to X and one believe that Q is false, then one lie to X.
One can weaken Moore by replacing ‘don’t believe’ with ‘uncontradictedly disbelieve’ and Searle and Evans by replacing ‘believes’ with ‘does not uncontradictedly disbelieve’, where x uncontradictedly disbelieves p just in case x disbelieves p and x does not believe p. It is left as an exercise to the reader to verify that the counterexamples to the strong principles adapt to the weak ones. Moreover, at least the second and third of my counterexamples will be cases of true assertions.
I shall use ‘<Q>‘ as shorthand for ‘the proposition that Q’. To avoid awkwardness, I will talk both of asserting a sentence and asserting a proposition, though, strictly, instead of ‘asserting the sentence “Q”’, one should speak of ‘asserting that Q by saying “Q”’.
Case 1. An expert (perhaps my analyst) who knows both me and the subject matter of <Q> correctly tells me: ‘Q and you don’t believe that Q. Work out the consequences for yourself.’ I’m a little slow on my feet, but agree with what the expert said and make a sequence of assertions (and one interjection): ‘Q and I don’t believe that Q. By conjunction-elimination, Q. Oh, now I see! I didn’t believe that Q, but now, thanks to your testimony, I do.’ The first assertion is true, justified, sincere and non-absurd. This refutes Moore.
Typically in a case like this, ‘Q and I don’t believe that Q’ wouldn’t remain assertible for long, since normally one would quickly come to believe that Q. But of course the amount of time for which ‘Q and I don’t believe that Q’ remains assertible can be extended indefinitely by supposing that something comes up to distract one from realizing one believes that Q.[4]
If belief does not distribute over conjunction, both conjuncts of my claim <Q and I don’t believe that Q> can be true, though not for a long time, even while I believe the conjunction. If, on the other hand, belief does distribute over conjunction, I would simply be both correctly believing that Q and incorrectly believing that I don’t believe that Q. Normally, such a pair of beliefs wouldn’t last long, as already noted, but until I recognize I have come to believe that Q, my belief based on expert testimony that I don’t believe that Q is still justified.
However, the example leaves intact Evans, Searle and Fallis2009.
Case 2. Suppose that I am bilingual in Polish and English, and I am in an earthquake that causes me to lose my hearing, but not my ability to speak. However, I form the false belief that I have also lost my ability to speak, based on the observation that people are not reacting to what I say—though in fact, the reason they are not reacting is that they are too busy dealing with those more seriously hurt than I. Some time later, I see a sign that the rescue organizers are looking for Polish speakers to assist with their efforts. Although I believe that I am no longer able to speak Polish out loud, I realize that I could be wrong. So I come up to one of the organizers, whom I happen to know to be unable to lip-read, and attempt to assert in English: ‘I can presently speak out loud in Polish and I don’t believe that I can presently speak out loud in Polish.’ And I succeed at my attempt. In so doing, I have asserted something that is true. Moreover, I was justified in asserting it. My justification was this: I knew the second conjunct (that I don’t believe that I can speak out loud) by introspection, and I know that the first conjunct (that I can presently speak out loud in Polish) is true if I manage to assert it, since if I can speak out loud in English, I can certainly likewise do so in Polish.[5] There is, thus, nothing absurd in my assertion. The assertion is true and I am being a responsible testifier in making it, as I know that the assertion will be true if I make it. Hence, we have a counterexample to Moore.
I am justified in making the assertion, since I know that either I will succeed in making it or I will not. If I succeed in making it, I will have asserted something true. If I fail to make the assertion, then I will not have asserted anything false.[6] So I know that in neither case am I going to assert a falsehood. However, I am not justified in asserting that I believe the content of the assertion. For in fact, I do uncontradictedly disbelieve it, since in the case offered I believe I am unable to speak out loud. Hence, we have a counterexample to Evans.
Moreover, given that I know that my assertion will be true if I succeed in making it and no false assertion will be made if I fail, I do not compromise my integrity, even prima facie. But an assertion that is not sincere is at least prima facie a violation of the virtue of integrity. Hence, the assertion is sincere, even though I uncontradictedly disbelieve its content. Thus, we have a counterexample to Searle. In fact, to get a counterexample to Searle, one can simplify: all we need is that as the above-described speaker who lost his hearing I assert ‘I can presently speak out loud in Polish’ while expecting the assertion not to come off due to an inability to speak out loud. Moreover, we have a counterexample to Fallis2009 since I am not violating sincerity and I am not lying, even though I am saying something that I believe to be false.
Moreover, we can see that the utterance is an assertion by noting that if I substituted the word ‘Mandarin’ for the word ‘Polish’, while knowing that I know no Mandarin, I would surely be in violation of a norm of assertion if I still succeeded in speaking.
A referee has suggested a disjunctive account: to assert sincerely you need to either (i) believe what you are asserting or (ii) believe that if the assertion is successful, it will be an assertion of a truth (or both). Unfortunately, this also fails as the present case shows. As before, suppose I incorrectly believe that I cannot speak out loud. Let P be ‘I can speak Polish out loud.’ Then the disjunctive account would correctly classify as sincere my assertion that P, since I believe that this assertion will be true if I succeed in making it. But it would also classify as sincere my assertion that it is not the case that P, since I believe that it is not the case that P, while that classification seems incorrect. For it would be very odd indeed if both my assertion that P and my assertion that it is not the case that P would count as sincere when in fact there is no contradiction in my beliefs.
Case 3. I have programmed a robot to bring me a drink whenever I utter the sentence: ‘The robot will bring me a drink and I don’t believe that the robot will bring me a drink.’ Jones, however, has a habit of interrupting me before I finish any sentence. I thus form a justified belief that I will not manage to say to Jones any sentence that is this long, and I try to assert to Jones ‘The robot will bring me a drink and I don’t believe that the robot will bring me a drink’, expecting to be interrupted. While the robot does not care what speech act, if any, I am engaging in when I make the requisite noise, I am really trying to assert the sentence to Jones. And, surprisingly, I succeed, because the word ‘robot’ makes Jones pay attention and refrain from interrupting. The robot then brings me the drink. I said something that was true. Moreover, I knew that if I succeeded in saying the sentence, it would be true. There is no absurdity. Thus, Moore is violated. (Though note that, again, the proposition will typically quickly cease to be assertible—in this case as soon as I realize that I managed to assert it.) Moreover, just as in the previous example, Evans, Searle, and Fallis2009 are violated.
One might worry that the sentence I uttered was a command to the robot and not an assertion. But this is mistaken. The robot responds to a particular utterance, regardless of what speech act the utterance constitutes. In my Case 3, I am making an assertion to Jones while expecting the robot to recognize what I am saying. I am causing the robot to bring me a drink, and only ‘commanding’ in the non-linguistic sense of the word ‘command’ in which one ‘commands’ the TV to switch channels by pressing a button. That an utterance also causes a robot to do something does not preclude the utterance from being an assertion to someone else. I can program a robot to bring me a drink whenever I utter a sentence containing the word ‘plagiarism’, and then I can assert to you that someone has been accused of plagiarism. That I am causing the robot to bring the drink makes this no less an assertion.
A referee has also offered a clever modification of the above case in which it is perhaps even clearer that an assertion has been made. I instruct a servant to bring me a drink whenever I assert to someone: ‘The servant will bring me a drink and I do not believe that the servant will bring me a drink.’ In this case, I am clearly both making an assertion to you and causing the servant to bring the drink.
I have given counterexamples to four principles. The principles, however, do hold in normal cases, just as in normal cases one knows p if and only if p is true and one justifiably believes p. But just as Gettier has shown that the knowledge biconditional is false in exceptional cases, so my cases show that the principles about assertions are not exceptionlessly true. There are four families of responses to examples like Gettier’s: (i) bite the bullet and insist that our intuitions about the examples are mistaken, (ii) add epicycles to the theory, (iii) insist that philosophers should not try to give exceptionless necessary and sufficient conditions, and (iv) construct an innovative new theory of knowledge. The first of these leaves a bitter taste, despite Weatherson’s [2003] able defense. The second has proved a failure. The third is definitely an option, but it is the fourth that is the most fruitful, and it is hoped that the present paper will encourage readers in that direction.
Is there a satisfactory way to repair the principles about assertion? Maybe, but it will not be easy. Let me focus on Searle. The counterexamples have the structure that while the speaker does not have the relevant belief, the speaker does believe that the assertion will be an assertion of a truth, at least if it succeeds in coming off. We might try to make this the new standard of sincerity: one makes a sincere assertion if and only if one believes, at the time at which one makes the assertion, that if the assertion is successful, it will be an assertion of a truth. Or perhaps we might require for sincerity a positive intention, at the time of final acceptance of responsibility, to refrain from asserting a non-truth [cf. Pruss, 2010].
Both suggestions, however, would need to handle the objection that small children appear capable of sincere assertion before they can have the complex higher-order beliefs or intentions the suggestions require.[7]
Baylor University
Baker, L. Rudder 1988. Cognitive Suicide, in Contents of Thought, ed. R.H. Grimm and D. D. Merill, Tucson: University of Arizona Press, 1–30.
Evans, Gareth 1982. The Varieties of Reference, New York: Oxford University Press.
Fallis, Don 2009. What is Lying?, Journal of Philosophy 106/1: 29-56.
Fallis, Don 2011. Lying and Grice’s Maxims of Quality, Workshop: Lying, Saying, and Meaning, University of Oslo, URL = <http://www.csmn.uio.no/podcast/FallisLying.html>.
Moore, G. E. 1942. A Reply to My Critics, in The Philosophy of G. E. Moore, ed. P. Schilpp, La Salle, IL: Open Court, 533-677.
Pruss, Alexander R. 2010. Lies and Dishonest Endorsements, Proceedings of the American Catholic Philosophical Association 84: 213-222.
Searle, John R. 1969. Speech Acts, Cambridge: Cambridge University Press.
Turri, John 2010. Refutation by Elimination, Analysis 70/1: 35-39.
Weatherson, Brian 2003. What Good are Counterexamples? Philosophical Studies 115/1: 1-31.
Williams, John N. 2004. Moore’s Paradoxes, Evans’s Principle, and Self-Knowledge, Analysis 64/4: 348-353.
[1] Fallis [2011] now defends a different account of lying.
[2] However, Turri’s example would also refute Williams’ principle [2004] that justification in believing p entails justification in believing that one believes p, while it is not clear whether my examples can be made to do so.
[3] In some cases, an author has a stronger ‘if and only if’ where I have only a one-sided conditional.
[4] I am grateful to a referee for highlighting the short duration of assertibility as well as its relevance to belief distribution in the next paragraph.
[5] I assume I knew that I wouldn’t be asserting it in any way other than out loud. Here I am using the assumption that my interlocutor can’t lip-read.
[6] Or maybe anything not true, if bivalence is false. I will ignore such complications.
[7] I am very grateful to three anonymous referees for a number of comments that have greatly improved the paper. I am also grateful to Trent Dougherty, Don Fallis, Jonathan Kvanvig and Heath White for discussion of these topics.