Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 18 October 2019

Lying, Belief, and Knowledge

Abstract and Keywords

What is the relationship between lying, belief, and knowledge? Most philosophical accounts of lying define it in terms of belief, namely telling someone something that one believes to be false, often with the intent to deceive. This chapter develops a novel account of lying by deriving evaluative dimensions of responsibility from the knowledge norm of assertion. On these dimensions, we can distinguish reasonable assertions, negligent assertions, and vicious assertions. On the definition of lying defended here, lies are best understood as special cases of vicious assertion; thus lying is the anti-paradigm of proper assertion. This enables an account of lying in terms of knowledge rather than belief: roughly, lying is telling someone something you know ain’t so.

Keywords: lying, knowledge, belief, assertion, norm of assertion, knowledge norm, vicious assertion, negligent assertion

9.1 Introduction

What is the relationship between lying, belief, and knowledge? One natural first-pass idea is to understand lying in terms of knowledge: to modify a line from Mark Twain, “Lying is telling someone something you know ain’t so.”1 But many will regard this natural idea to be incorrect from the start, insofar as they judge that one can lie even when one simply believes, falsely, the negation of what one asserted: on their widely-accepted view one can lie even if one mistakenly tells another the truth. In what follows I shall proceed as if this widely held view should be respected as orthodoxy, though in section 9.5 I shall revisit a version of the Twain-style knowledge account of lying to explore how plausible it might actually be.

Several philosophers debate the exact conceptual analysis of lying. Many accept an analysis with the traditional three-clause structure:

  • Lying, Belief, and Knowledge

Much of the debate over such an analysis concerns the exact nature of the third condition, in particular, whether the speaker intends for S to infer something false about the (p. 121) proposition asserted,2 or about what the speaker herself believes, or perhaps that the former be inferred from an inference about the latter. However, many argue that a third such clause, requiring something about the speaker’s intent to deceive, is not even necessary for an assertion to be a lie; for example, Sorensen (2007), Saul (2012), and Stokke (2013a) defend definitions of lying which retain the first two clauses above, but abandon any such third condition. Even if such a third condition is not necessary, most commentators are interested primarily in lies which are intended to deceive, and one might approach the project of analyzing what a lie is by focusing on deceptive lies (cf. Fallis 2010a: 4 ff.). While such lies are worthy of such interest, we shall see that an account of lying is available which can explain with other resources the deceptive intent typical of lies.3

The main focus of this chapter is the relationship, if any, between knowledge and belief, which figure in the epistemic norm of assertion, and the moral/social norms against lying. I develop a broad account of the injunction against lying according to which, roughly, lying involves intentionally misrepresenting oneself at the level of what one believes or knows, which in turn represents one’s epistemic relationship to the proposition asserted. Two such accounts will be articulated, one in terms of belief, and one in terms of knowledge. Approaching the project in this way enables us to see how a norm against lying may be derived from the epistemic norm of assertion. And it opens up a new perspective on how to understand lying which mirrors the epistemic expectations which hearers have toward speakers who make assertions in testimonial exchanges.

9.2 The norm of assertion

First, some terminological preliminaries. Assertion is the linguistic act of committing outright to the truth of a proposition: by one’s utterance one declares or claims or states4 that something is so. As such, assertion is regarded as the default speech act for outright utterances made in the declarative or indicative mood. Speakers use assertions to make unhedged claims about the way things are, such as when one states London is the capital of the UK, or Joanna is in her office.5 Amongst speech act theorists (e.g., Searle (p. 122) 1979a: 2–5), it is common to distinguish assertion from other types of speech act (for example from interrogatives, imperatives, or performatives, among others) by noting, first, that the point of assertion is to represent the world as being a certain way, namely, the way represented by the proposition asserted; second, that assertion has a word-to-world direction of fit (the utterance is made true by being made to conform to the way the world is); and third, that assertion of a proposition expresses the psychological state of belief in that proposition. On this approach, assertion expresses belief even when the speaker asserts insincerely by not believing what she asserts.6

What is meant by an epistemic norm of assertion? Assertion’s epistemic norm is a social linguistic rule that one satisfy some relationship to that proposition for properly asserting: typically, it specifies a necessary condition on (epistemically) proper assertion, such as that one must know it, or believe it, or have adequate evidence for believing it, or that it be true. Philosophers often think of belief as the mental correlate of assertion: it is the mental act of outright commitment to a proposition as true, which normally triggers dispositions to act on that belief. But beliefs, and assertions, can be false; whereas knowledge, though normally regarded as involving belief (one’s belief can amount to knowledge), is distinguished from mere belief in that one can only know facts, that is, true propositions. Though one can hold a false belief, one can only know truths. Because of this, it is often said that knowledge is ‘factive’.7 (Moreover, knowledge requires some normative condition(s), for example that one’s belief be well-formed, be the product of a safe method, or fit one’s evidence, etc. Epistemologists disagree on the exact nature of these conditions, which typically fall under the label of ‘epistemic justification’; we shall not consider these issues here.)

Grice’s maxim of Quality offers a candidate epistemic norm of assertion: its supermaxim is ‘Try to make your contribution one that is true’ and its two specific submaxims are

  1. 1. Do not say that which you believe to be false.

  2. 2. Do not say that for which you lack adequate evidence.8

Grice’s first submaxim is sometimes called the sincerity condition, the second the evidence condition. Strictly speaking, Grice’s Quality supermaxim does not require that one’s assertion be true. But if we interpret it as being concerned with truth (and not merely trying to assert truths), we may understand its submaxims as corollaries the following of which will help one fulfil the supermaxim.

(p. 123) Timothy Williamson (2000: ch. 11) and others9 have defended knowledge as the epistemic norm of assertion (KNA), such that this norm is constitutively related to the speech act of assertion:

  • Lying, Belief, and Knowledge

The Knowledge Norm of Assertion imposes a strict duty, forbidding the combination that: one asserts p and one does not know that p. Because knowing involves believing, KNA forbids insincere assertions; and because knowing is factive, KNA forbids false assertions. On KNA, one does something epistemically improper when one asserts without knowing what one asserts, even if one thereby does something appropriate in, for example, the prudential or the moral sense. Much of what follows invokes KNA, but very little turns on this; many of the ideas developed here generalize to most rival accounts of the norm.

Lying is a category that most naturally applies to assertion, rather than to presupposition or conversational implicature, or to other speech acts, for example, to suggesting that a proposition is or might be true, or asking questions concerning whether a proposition is true.10 Thus, a natural route into the connections between the communicative practice of assertion and the norms against lying is to consider the epistemic norm of assertion.

9.3 Norms and evaluative dimensions

We can gain some purchase on the relationship between KNA as the norm of assertion and lying by considering how KNA will affect conversational patterns, and how it will generate expectations of conversational participants who implicitly grasp the relationship of KNA to assertion. In particular, if KNA is correct, assertions will typically represent their speakers as knowing what they assert;11 for by engaging in the practice of (p. 124) assertion whose norm according to KNA its speakers implicitly grasp, a speaker would represent herself as conforming to KNA. Even if we allow that asserting instead represents something weaker about oneself—perhaps only representing oneself as believing, or as having some adequate evidence for, what one asserts—then still, asserting what one knows one does not believe will involve intentionally misrepresenting oneself in some way.12 Does lying always involve misrepresenting what one believes? One gloss might be this:

The parenthetical clause of condition (ii) is needed to rule out as not being lies cases where one believes a proposition and its negation; this possibility, though typically overlooked, can occur (on a certain view of structured propositions) when one believes a proposition and its negation under different guises. Suppose I see the same object through two different windows, but do not realize that it is the same object. Through the right window, it looks gray, and through the left window it looks green. I come to believe both: that object [seen through the right window] is gray, and also that object [seen through the left window] is not gray. But if the object is in fact gray, and I tell someone (thinking of what I see through the left window) that the object is not gray, I have not lied. But without the parenthetical clarification in (ii), Lie would count this as lying.

More generally, it might be possible for someone to know a proposition but, perhaps by misremembering or by habit or by some irrational process which they do not recognize, they come to believe its negation. (My wife accidentally leaves her mobile phone behind, and I find it; it looks much like my own. Later I need to call her at a moment when I happen to be holding her phone, and out of habit, using her own mobile phone which looks like mine, I begin to try to call her mobile number. For that moment, (p. 125) I plausibly believe that she has her phone, even though I know that she does not have it. If during that moment I tell someone that she has her phone, I am not lying.)

Some may suppose that lie’s condition (iii) is superfluous because they think of assertion as an intentional speech act requiring a speaker to intend to represent herself as being sincere, that is, as believing what she asserts. This way of thinking about assertion might well be correct; but it seems committed to the idea that one cannot ironically or metaphorically or playfully assert a falsehood so as to mean or convey instead its negation. On this view of assertion, irony involves cases where one utters that p, and ‘makes as if to’14 assert that p, but one is in fact asserting (roughly) that ¬p. Those who understand assertion in this way may leave off condition (iii), and insist that with ironic utterances a speaker does not lie, for she does not even fulfill condition (i).15 But for those who prefer a thinner conception of assertion according to which one can assert something without intending to communicate it, or without intending to represent oneself as believing it, (iii) will be needed. On the thinner picture of assertion, (iii) can itself provide an interpretation of why standard cases of irony are not lies: for the hallmark of irony is that it typically does not involve the speaker’s intending to represent herself as believing the falsehood uttered (mutatis mutandis for cases of metaphor, hyperbole, and so on).

Condition (iii) need not involve the intent to deceive B concerning either p or what A believes, because A might know that B knows what A believes. So this gloss properly counts many of Sorensen’s bald-faced lies and ‘knowledge-lies’16 as lies. For bald-faced lies and knowledge lies are plausibly speech acts which are nevertheless used to represent oneself as believing what one asserts, even if everyone involved knows or thinks it likely that one does not believe it.17 Clause (iii) also enables a satisfying pragmatic account of why, typically, lies do involve the intent to deceive: most scenarios in which one wants to misrepresent oneself are also occasions on which one does so for deceitful purposes. On those occasions, (i)–(iii) obtain because one endeavors to deceive someone into thinking that one believes p, and typically as a means to getting them to believe that p.18 But because what is typical of lying may not be required for its definition, I suspect that a definition of lying need not include a clause concerning the intent to (p. 126) deceive. So I will proceed as if lie is plausible enough to begin with as a general characterization of lying.19

Recall that KNA forbids the combination that one asserts a proposition but one does not know that proposition. That is, KNA is logically equivalent to: One must not: (assert that p & not know that p). If KNA is the norm of assertion, then we can plausibly derive a secondary norm of assertion on which one should assert only if one reasonably takes oneself to know. This follows from a quite general schema connecting norms of a certain structure to the conditions under which one should do what the norm permits. The schema reads:

  1. (1) Lying, Belief, and Knowledge

where the “transition from ‘must’ to ‘should’ represents the transition from what a rule forbids to what it provides a reason not to do” (Williamson 2000: 245). If we suppose (as seems plausible) that evidence is what makes belief reasonable,20 and if we prefer talk of reasonable belief over merely one’s having evidence, we may substitute these in thus:

  1. (1*) Lying, Belief, and Knowledge

Substituting ‘assert p’ for ‘φ’ and ‘one knows p’ for ‘C’ yields the proposed derivation from KNA, which we may call KNA*:

  • Lying, Belief, and Knowledge

Given KNA, one may apply modus ponens to KNA* to derive the following norm (call it KNA**): one should assert a proposition only if one reasonably believes that one knows it.

Note that KNA**, derived in this way from KNA and KNA*, might seem to impose an extremely strong standard on what it takes to be in position to assert reasonably.21 (p. 127) But KNA** is much stronger than what we need to generate a useful derivative norm. For given KNA, one who endorses it (tacitly or explicitly) will tend to accept the following:

  • Lying, Belief, and Knowledge

Assuming normal cognitive agents who tend to believe or accept what their evidence suggests, (G) will serve as a guidance norm for one who aims to conform to KNA, because following (G) will make one less likely to violate KNA.22 Agents who accept KNA will violate (G) on pain of either incoherence or akrasia: it would be incoherent if they accept KNA, prohibiting assertion of what one does not know, but also accept (against G) that they may on this occasion assert that which they believe they do not know; whereas it would be akratic if they accept KNA but owing to weakness of will, do anyway what they know that (G) forbids. Either way, they would be choosing to assert a proposition which they believe or accept to be in violation of a norm which they endorse.

9.4 Lying and vicious assertion

One may distinguish evaluative dimensions of responsibility built up from any norm with a certain structure. Consider any moral, prudential (etc.) norm enjoining one to undertake an action φ only in circumstances C. If one endorses that norm on φ-ing, then one may be said to φ reasonably when one believes one is in C; one may be said to φ negligently when one does not even consider whether one is in C; and one may be said to φ viciously when one believes that one is not in C. For example, if a moral norm enjoins one to act only if one causes no one harm, then: one reasonably acts when one believes that by so acting one will cause no harm; one negligently acts when one acts without consideration of whether one’s action will cause harm; and one acts viciously when one acts while believing that one’s action will cause harm. Or for a legal example, suppose there is a legal rule enjoining one to send someone to prison only if they are guilty of breaking a law. A sentencer would thus act reasonably by sending someone to prison whom she believes to be guilty; she would act negligently by sending someone to prison when she does not consider whether or not they are guilty; and she would act viciously by sending someone to prison whom she believes is not guilty.

Given (G), and again on the simplifying assumption that a speaker grasps and endorses KNA, we can also distinguish between evaluative dimensions of responsibility (p. 128) in the domain of assertion. These layers of responsibility turn on one’s reflective stance toward one’s epistemic position concerning the asserted proposition p:

  1. (E1) Lying, Belief, and Knowledge

  2. (E2) Lying, Belief, and Knowledge

  3. (E3) Lying, Belief, and Knowledge23

These evaluative terms (negligent, vicious) aim to label the distinctively epistemic responsibility at play in assertion. For example, (E2) aptly captures something akin to Frankfurt’s (2005) notion of bullshitting, that is, asserting without regard to whether one’s claims are true.24 Likewise, the viciousness of (E3) is primarily epistemic; so-called ‘prosocial’ lies, though undertaken with high moral or social aims, nevertheless exhibit epistemic vice.

According to (E1), one can reasonably assert even if one violates KNA (and thereby asserts improperly): by asserting when one believes that one knows, one is in some sense less subject to blame if one violates KNA, for one does something right by virtue of trying to conform to KNA.25 According to (E2), one negligently asserts when one fails to evaluate one’s epistemic position with respect to p but goes ahead and asserts p anyway; even if one in fact fulfills KNA, one has done something a bit irresponsible. Finally, given (E3), one viciously asserts when one also believes that one asserts in violation of KNA. In the case of vicious assertion where one asserts that p in order to express what one thinks that one does not know, because one knows that one believes the opposite, we plausibly have a lie. That is, where a speaker asserts that p in order to assert what she herself realizes that she does not believe (because she knows she instead believes that not-p)—and where she does not flag by any other means (such as by winking to her interlocutor) that something special is going on—she lies. Lying is then a special case of vicious assertion.26 (p. 129)

9.5 Knowledge and lying

We have considered a subject who lies by asserting p while knowing that she believes that not-p. Because she knows that she does not believe p,27 she thus knows that she does not know p. By asserting p, she misrepresents herself as believing p when she does not; and (E3) classified this as a vicious assertion, which our gloss Lie above counts as a lie.

Lie can seem plausible as a general characterization of lying, in part because its third clause invokes the intent to misrepresent oneself even though that intent may not extend to deceiving others. But perhaps the Twain-inspired knowledge slogan—that lying is telling someone something you know is not so—should be revisited, for it makes possible an attractive account of lying which parallels the knowledge norm of assertion:

Like Lie from section 9.3, this knowledge account of lying counts bald-faced lies and ‘knowledge-lies’ as lies. Lie-K also makes good on the idea from section 9.4, that lying is a special case of vicious assertion. In fact, it makes characterizing that special case even easier: one viciously asserts p when one asserts it to express p when one also believes that one does not know that p. But when one knows that p is false it will typically be quite easy to form the belief that one does not know that p. So the special case of viciously asserting occurs when one believes that one does not know what one asserts because one knows its negation. Lying is thus the anti-paradigm of conforming to the norm of assertion: asserting what you know follows KNA; intentionally asserting the negation of what you know is lying.30

This mirroring relationship between lying and the KNA upholds the intuitive idea that lying involves violating a conversational norm (Fallis 2012: 577). It also suggests an elegant account of why it would be that lying is possible for assertion but not for (p. 130) conversational implicatures or other non-explicit messaging, which themselves often depend on assertion.31 If lying is conceptually related to the knowledge norm of assertion in the way envisioned above, it explains why one cannot lie that p by conversationally implicating that p: because conversationally implicating that p does not represent one as knowing that p, one does not undertake the same epistemic responsibility by conversationally implicating that p as when one asserts that p with full force. This is partly because conversational implicature and other implicit messages make possible denying what was conveyed, something which is not available with outright assertions. Claiming “I never said that p” when one has merely conversationally implicated that p can typically get one off the epistemic hook for having conveyed that p; but clearly, when what one said was that p, one cannot (truthfully!) deny that one said it.

Lie-K has the virtue of being more simple by dint of needing no intent to deceive clause, a clause invoked by many traditional accounts of lying (compare the third clause of the Trad schema with which this chapter began). And when combined with KNA, it enables an account of why typical lies are misleading as well as epistemically problematic, even without an intent to deceive clause. Typical lies are misleading because, given KNA, they represent their speakers as knowing, and so believing, something which their speakers know to be false. And they are epistemically problematic because they frustrate hearers’ cognitive efforts at gaining knowledge from others by way of testimony: indeed, when lies are taken up by hearers as the truth, they generate false beliefs which their hearers take on trust from their interlocutors to be knowledge. When testimonial uptake leads one to think that one knows something which is in fact false, and which one took on trust from someone else who was in fact the opposite of trustworthy, it is not just that one’s epistemic position is compromised; the social goodwill we have toward others, especially those with whom we have ongoing relationships, is strained.

Notice that preface paradox scenarios can generate a case judged to be a lie by a belief account, but which is not judged a lie by the Lie-K account.32 Think of a writer who has painstakingly researched and written a non-fiction book, in which she makes numerous claims. She accepts each claim as true, though she recognizes that given the book’s length and her own fallibility, at least one of her claims is likely to be false; and in the preface she admits as much. She thus has asserted (by writing the whole book) a large series of claims, but has also claimed that (at least) one of them is false (cf. Makinson 1965). Similarly, suppose S believes each of a large set of propositions p, q, r, … n, but which is a set such that S does not believe their conjunction. That is, S believes each of these propositions individually, but does not believe or assent to their conjunction: for S knows that one of the set’s conjuncts is likely to be false, though S knows not which. For these reasons, S believes that the conjunction (call it ‘CON’) is false. And we shall stipulate that one of the conjuncts in that set is in fact false. If S asserts the conjunction CON (p. 131) (‘p and q and r and … n’) to someone, S is asserting a proposition which S does not believe (though S believes each of its conjuncts); indeed, S believes that CON is false.

Let us suppose that the intuitive response to such a case is that S’s conjunctive assertion of CON is not a lie. On many belief accounts of lying, such as Lie from section 9.3, the assertion of the conjunction CON will presumably, but counterintuitively, count as a lie. This is because S believes that CON is false. Even traditional schemas for defining lying, such as Trad, will either have to deem it (counterintuitively) as a lie, or will have to explain the case as one in which S does not, by asserting the conjunction, intend to deceive the hearer into believing the conjunction. Yet the latter option seems a tall order, for asserting the conjunction looks like an act which is undertaken to put forth that conjunction to the hearer, even though the speaker believes that conjunction to be false.

Lie-K appears to have an advantage here: for Lie-K does not even count the assertion of CON as a lie. This is because S believes, but does not know, that CON is false. Now, S’s asserting CON would count as a vicious assertion, because S regards herself as believing that she does not know CON. But it does not meet Lie-K, nor is S’s asserting CON one of our special (lying) cases of vicious assertion. This is because S’s assertion of CON is not undertaken in order to express what S knows that she does not believe (S would know this by knowing that she instead believes that CON is false). Rather, S undertook asserting CON merely by, and for the purpose of, asserting individually the conjuncts of CON, each of which S believes.

9.6 Objections

lie-K will be opposed by those who insist that one can lie while (mistakenly) telling the truth.33 These traditionalists always invoke belief rather than knowledge in the second clause of such accounts because they have the strong intuition that it is still a lie when one wrongly believes a falsehood, and tells someone its negation: if I really believe that the grocer is closed today, and I tell you it is open because I want to frustrate you by sending you to a closed shop (and you go and find it open and happily get on with your shopping), these traditionalists maintain that I have still told a lie. Some recent empirical studies suggest, however, that this intuition is wrong: results from Turri and Turri (2015) suggest that in fact, most people require falsity for a telling to be a lie.

Such findings are not the last word; indeed, the case developed for Lie-K above proceeded on independent grounds. But consistent with such findings, the proponent of Lie-K can claim that in the above case, when I told you the grocer was open, I merely tried (and failed) to tell a lie. To know whether you have lied, the Lie-K proponent will suggest, we must look not only at the speaker’s inner mental life, but also at the way the world is.

(p. 132) The Lie-K proponent can thus be concessive here to the traditionalist who prefers an account along the lines of (Trad) or our (Lie) above: what traditionalists get right is that the attempted liar has done something wrong (and it might be that the traditionalist is conflating the judgment that the attempted liar has done something wrong with the judgment that she has lied). In particular, the attempted liar has engaged in vicious assertion, for he tells someone something which he himself believes he does not know, because he knows that he does not believe it (instead, the would-be liar believes its negation). That would be a vicious speech act the likes of which, when one chooses to engage in it, involves choosing to misrepresent one’s own state of mind to someone else.34 But it need not follow from this that the attempted liar has in fact lied.

Given the Lie-K account, is it possible to lie to someone even when one is not engaged in vicious assertion? If our claim above is correct, that lying is a special case of vicious assertion, then no, it would not seem possible. Suppose that Bill and Ashley are talking about France, where Bill would like to visit. And suppose that (i) Ashley asserts that Lyon is the capital of France to Bill, (ii) Ashley knows that this is false (and does not believe that it is true), and (iii) Ashley intends by so asserting to represent herself as knowing that it is true. For it to be plausible that Ashley is not engaging in vicious assertion, it would have to be the case that Ashley does not believe that she knows Lyon is the capital of France to be false. Now, it might be possible for Ashley to assert this to Bill without even forming a belief about what she knows concerning France’s capital. She might be asserting this to Bill simply because it was the first proposition about France which, though false, popped into her head, and Ashley was just not reflective enough to assess her epistemic position with respect to Lyon is the capital of France. In our taxonomy of evaluative dimensions (E1)–(E3), Ashley here is asserting negligently (E2) rather than viciously. Note however that (E1)–(E3) are only sufficient conditions for reasonable, negligent, and vicious assertion, respectively; as such, some negligence might itself be vicious, particularly if one is intentionally acting recklessly. If one interprets the evaluative notions in this way, Ashley’s lie might be merely negligent given (E2), but viciously so; and so the lie is still a case of vicious assertion, even though it may not be vicious in the sense given by (E3).

But even if we regard Ashley’s assertion as negligent without being vicious, it is hard to see how Ashley could be unreflective about her epistemic position toward this proposition yet also (iii) intend by her assertion to represent herself as knowing that Lyon is the capital of France. For by intending through that assertion to represent herself as knowing this, it looks as if she at least has a cognitive perspective on it, which would involve enough reflection to see that she does not know it, and in fact knows its negation. More generally, I suspect that it will be hard to find cases in which a speaker does intend (p. 133) to represent herself as knowing something when she knows its negation, yet does not actually believe that she knows its negation. One reason why it will be hard to find such cases is that coming to form an intention to represent one’s epistemic state as being a certain way plausibly engages one’s reflective capacities on whether one’s epistemic state is that way. This is plausible because cases of lying involve doing the opposite of what we normally do. Most of us assert knowledgeably, and thus truthfully, most of the time, and even when we do not, our default aim is truth rather than falsehood. Cases in which we opt against this habit and tell falsehoods require us to note mentally that we are telling someone something against what we believe, and such mental judgments will involve reflection on both what we think is true and why not to tell someone the truth.35

9.7 Conclusion

In this chapter, I have considered the relationship which belief and knowledge have to our understanding of lying. That relationship turns on the epistemic norm of assertion, and in particular, on the way in which belief and knowledge are required for proper assertion. Most of the results considered may be applied whatever is the correct epistemic norm of assertion, though the knowledge norm, KNA, has been widely defended. As such, I have used KNA to provide the background for articulating the normative epistemic dimensions to which we hold one another responsible in conversation. Epistemically reasonable, epistemically negligent, and epistemically vicious assertion provide a taxonomy for understanding the epistemic and social aspects of the wrongness of lying, as well as the wrongs of other speech which are not lying, but which are nonetheless misleadingly deceptive. The norms which may be derived from KNA offer a new perspective on how to understand lying: lying is knowingly doing, because one knows the opposite, that which the norm of assertion forbids. By intending one’s assertion to represent one as knowing, or at least believing, the opposite of what one knows to be true, one lies. Lying is the inverse of an assertion which is beyond epistemic reproach; lying is as bad as knowingly asserting is good.


Thanks to Max Baker-Hytch, Rachel Fraser, John Hawthorne, Jörg Meibauer, John Turri, and Andreas Stokke for discussion and helpful comments. I am also grateful to two anonymous referees for useful feedback. This chapter was made possible through the support of a grant from the John Templeton Foundation. The opinions expressed in this publication are those of the author and do not necessarily reflect the views of the John Templeton Foundation.


(1) See the quote about faith from the school boy, in Twain’s Following the Equator, ch. 12, “Pudd’nhead Wilson’s New Calendar” (1898).

(2) Different accounts specify the intent-to-deceive clause differently: see e.g., Mahon (2008a), and Lackey (2013). See Fallis (2010a) for an overview.

(3) For the purposes of this chapter, I shall largely set aside concerns about irony, hyperbole, or otherwise inaccurate claims involving scalar considerations: for example, when someone claims that he just swam a mile, or that it is 9 o’clock even though he knows that it is 8:59. Whether these ought to be classified as lies is a difficult question (though see Marsili 2014); but one might think that if so, this gives motivation for including a very specific clause about intent to deceive.

(4) Some theorists distinguish statements from assertions (Meibauer 2014a: 71), citing the German distinction between Feststellungen and Behauptungen. In English, however, there is little to distinguish them.

(5) For lengthy overviews, see Jary (2010a) and Pagin (2015); for shorter overviews, see MacFarlane (2011) and Goldberg (2015: ch. 1).

(6) Searle (1979a: 4). This importantly updates Searle’s original position (Searle 1969: 64 ff.), on which sincerity was required in order even to make an assertion.

(7) For more on factivity, see Benton (draft, ch. 1).

(8) Grice (1989c: 27). Benton (2016) argues that Grice’s maxim of Quality is best understood as the knowledge norm, ‘KNA’ in what follows.

(9) See especially Turri (2010, 2011, 2014a), Benton (2011, 2016, draft), and Buckwalter and Turri (2014), among many others. For an overview of this literature and its critics, see Benton (2014: §1), and Pagin (2015: §6.2).

(10) Does one do something akin to lying if one predicts that p when one knows that not-p? This will depend on whether one treats the force of predicting as on a par with asserting (e.g., for Stokke 2013a, who thinks of assertion along Stalnaker’s lines of proposing that a proposition be added to the conversation’s common ground, predicting might well have the same force as asserting). For an epistemic norm of the speech act of prediction in terms of proper expectation, see Benton and Turri (2014: esp. 1862 ff.).

(11) Moore (1962: 277) and Unger (1975: 251 ff.) endorse the thesis that asserting that p represents its speaker as knowing that p. The case for that thesis turns on conversational patterns and the kinds of criticism available, which any norm of assertion ought to well explain. Williamson notes that the KNA norm subsumes the assertions-represent-knowing thesis under more general principles: “In doing anything for which authority is required (for example, issuing orders), one represents oneself as having the authority to do it. To have the (epistemic) authority to assert p is to know p” (Williamson 2000: 252 n. 6). Because the norm itself explains why asserters do represent this of themselves, the case for the representation thesis is also the case for the norm. Pagin (2016: §6 n. 20) disputes this by claiming that even here, ‘authority’ has a non-normative reading, namely “satisfying of sufficient conditions, which need not be related to any norms”; but sufficient for what? Presumably, an answer must be given in normative terms: that is, sufficient for acceptable or appropriate action undertaken as an authority.

(12) This is because people do typically believe what their evidence adequately supports; and plausibly in virtue of this, speakers express belief, or represent themselves as believing, when they assert. This creates a problem for Lackey’s (2007a) alternative Reasonable-to-Believe Norm of Assertion (which does not even require belief for proper assertion), particularly because she also endorses a ‘Not Misleading Norm of Assertion’ (2007: 615–17). For criticism, see Benton (2016: esp. n. 27) and Turri (2014b).

(13) This gloss bears some resemblance to one considered by Davidson (1985: 88).

(15) See Stokke (2013a) and Chapter 10 by Andreas Stokke for helpful discussion.

(16) See Sorensen (2007, 2010; cf. also Staffel 2011). The above account will count such lies as knowledge-lies when they are uttered with the intent of preventing the addressee from knowing that what is asserted is untrue, but without the intent of deceiving them into believing it (see Sorensen 2010: 610).

(17) An example of a bald-faced lie: a student who is caught red-handed cheating on an exam nevertheless denies it to the headmaster in his office, because denying it results in a punishment that the student deems a preferable to what he would receive if he confessed to the cheating (for similar examples, see Carson 2006: 290 and Lackey 2013).

(18) One might think that one can lie while satisfying lie’s (i) and (ii) by asserting p whilst only intending to represent oneself as being in a good epistemic position with respect to p (that is, regardless of whether one believes p). I do not think that is a lie, but I do find something objectionable about it: I shall describe what is objectionable about it in section 9.4, under the notions of negligent and vicious assertion.

(19) Fallis (2010a: 13 n. 43; 2013b) seems to agree, commenting on Davidson (1985).

(20) Epistemologists think of evidence in a variety of ways (some think of it as consisting of beliefs, or experiences, or propositions, or some collection of these; Williamson thinks of it as knowledge (Williamson 2000: ch. 11). But epistemologists tend to agree that however it is understood, evidence is the kind of thing (perhaps not the only thing) which somehow supports hypotheses (typically, by making them more probable). So they tend to agree that a belief that p is reasonable if p is supported by one’s evidence.

(21) Though it is one which Williamson (2000: 256) seems to accept; how strong it actually is will depend on how hard it is to reasonably believe that one knows when one in fact knows; for example, it might be that knowledge that one knows is not that uncommon. Nevertheless, violating KNA** is presumably less important than violating KNA itself: one might develop virtuous cognitive habits such that one tends to assert only what one knows, even though one rarely forms a belief about whether one knows.

(22) I am also assuming that for such normal cognitive agents, their practice of believing what their evidence suggests makes the propositions believed objectively more likely to be true. For strange agents who (perhaps for some domain) constantly believe falsehoods more often than they believe truth, (G) will not help them much.

(23) These distinctions owe much to Unger (1975: 260–2) and Williamson (2000: 245 and 255–6). Some, particularly those who think that KNA is too strong, will likely regard (E3), for example, as much too strong. Note however that (E3) fits well with our intuitive reaction to the knowledge version of Moore’s paradox, namely assertions of the schema “p but I don’t know that p.” And (E2) fits well with our reaction to an assertion of “p but I haven’t considered whether I know that p.”

(24) For more, see Chapter 20 by Andreas Stokke.

(25) Compare Williamson’s (2000: 256) distinction between reasonably asserting and permissibly asserting, and also DeRose’s (2009: 93–5) distinction between “primary propriety” and “secondary propriety.”

(26) Unger (1975: 260–2) gestures at such an idea, utilizing the asserting-represents-knowing thesis. Marsili (2014) claims that graded-belief lies, wherein one asserts that p but only believes that p is probably false, intuitively also count as lies. On the accounts developed here (Lie and Lie-K), they are not full lies, but they are still special cases of vicious assertion: one asserts the negation of what one thinks is probable, in order to assert what one realizes one does not (outright) believe.

(27) We are assuming that she realizes that she does not believe both that p and that not-p.

(28) For reasons delineated in §3 above, discussing Lie.

(29) Turri (2016) considers a similar ‘known-false’ gloss on lying, but his does not include a third clause. Recall however that the third clause might be dropped on a thicker conception of what assertion is.

(30) Notice also that endorsing the derived KNA** from section 9.3 itself would provide a strong norm against lying in the sense given by Lie-K: for conforming to KNA** will ensure that one does not meet the conditions of Lie-K.

(31) See Fricker (2012) and Hawthorne (2012) for discussion. Adler (1997) argues that false conversational implicatures are not lies. Meibauer (2014a: 113 ff.) demurs, though he also allows that false implicatures could be called deceptive rather than lies (2014a: 135).

(32) Thanks to John Hawthorne for suggesting this case.

(33) Contrast Carson (2006), who is rather exceptional in arguing that lies must have false subjects.

(34) Notice that Lackey’s (2007, 2013) (see also Chapter 18 by Jennifer Lackey) cases of “self-less assertions” are classified here as vicious assertions which are not in fact lies (on Lie-K) if they are true. I find this result satisfying, for selfless assertions are intuitively not epistemically appropriate assertions, even if they are in some respects admirable; for criticism of relying on such cases, see especially Turri (2015).

(35) Lying engages more areas of the brain, including areas involved with suppressing the truth, and with anxiety. See Langleben, Schroeder, et al. (2002) and Langleben, Loughead, et al. (2005).