Lying versus misleading, with language and pictures: the adverbial account

Abstract

We intuitively make a distinction between lying and misleading. On the explanation of this phenomenon favored here—the adverbial account—the distinction tracks whether the content and its truth-committing force are literally conveyed. On an alternative commitment account, the difference between lying and misleading is predicated instead on the strength of assertoric commitment. One lies when one presents with full assertoric commitment what one believes to be false; one merely misleads when one presents it without full assertoric commitment, by merely hinting or otherwise implying it. Now, as predicted by the well-supported assumption that we can also assert with pictures, the lying/misleading distinction appears to intuitively show up there too. Here I’ll explain how the debate confronting the two accounts plays out both in general and in that case, aiming to provide support for the adverbial account.

We intuitively make a distinction between lying and misleading. If, to answer Rebecca’s question ‘Are you going to Paul’s party?’ (hoping to thereby discourage her from attending), Dennis replies ‘No, I’m not going to Paul’s party.’, when he intends to go, and ends up going, Dennis is lying to her. If he had answered instead in the same circumstances ‘I have to work.’, implying that he doesn’t plan to go, then he is misleading her but not lying (Stokke, 2016, 85).

On the account that I favor—the adverbial account, as I’ll call it, a specific saying account—the distinction tracks whether content and truth-committing force are intuitively (as fully as language allows), literally or explicitly conveyed: not just thus with what is done, but with how it is done also, hence ‘adverbial’; an explicit definition is offered in Sect. 1. Several writers (Borg, 2022; Michaelson, 2016; Saul, 2012; Stokke, 2016) have pointed out that on saying explanations the intuitive distinction provides evidence on the theoretical notion of what is said and the theoretical distinction between semantics and pragmatics. On an alternative (assertoric) commitment account (Meibauer, 2014; Viebahn, 2017, 2020, 2021; Reins and Wiegmann, 2021; Wiegmann et al., forthcoming), the difference is predicated instead on the strength of assertoric commitment. One lies when one presents with full assertoric commitment what one believes to be false; one merely misleads when one presents it without full assertoric commitment, by hinting or otherwise implying it.

Now, it is a well-supported assumption that we can also assert with pictures (Eaton, 1980; Greenberg, 2018; Kjørup, 1974; Korsmeyer, 1985; Nöth, 1997).¹ Viebahn (2019) argues that the lying vs. misleading distinction is intuitively available there, even if perhaps less sharply. He uses this claim to defend the commitment account, on the grounds that it deals better with the distinction in that medium. Here I’ll critically examine the debate confronting the two accounts. Section 1 articulates the adverbial account, drawing on Grice’s notion of what is said. It is flawed for well-known reasons when it comes to its main goal—to distinguish between semantics and pragmatics; but it is closer to what we need to capture the intuitive notion of lying, and thus it may after all further the methodological goals that Grice expected from it. Section 2 presents the alternative commitment account; it summarizes my critical take on it (García-Carpintero, 2021a), and it discusses recent empirical work on the issue, arguing that it is consistent with the adverbial view. I’ll argue that by distinguishing along Gricean lines assertoric commitment and explicitness as separate ingredients in the account of lying, unlike other saying accounts the adverbial view improves on commitment accounts.

I’ll then provide additional support for it and its methodological implications by critically engaging with Viebahn’s (2019) arguments on lying with pictures (Sect. 3). I’ll argue for the conditional claim that, to the extent that there is a lying vs. misleading distinction that applies to pictures, the adverbial account also captures it better than alternatives. The view I promote thus makes the issues we’ll be discussing relatively independent of the medium, linguistic, or pictorial. I’ll emphasize the (striking, to my mind) structural and methodological parallels.

1 Lying versus misleading: the adverbial account

Let’s start with further examples of the intuitive distinction between lying and misleading. Williams (2002, 96) illustrates it with an utterance of ‘Someone has been opening your mail’, when it is the speaker who has been doing it. Another traditional example is Athanasius’ reply to his enemies’ query about his whereabouts, ‘He’s not far away’ (Williams, 2002, 102). In both, speakers convey conversational implicatures (Grice, 1975) believed to be false, someone but not me is opening your mail, Athanasius is nearby but not here, while what they literally say is true. They intuitively aim to mislead their audiences by steering them to form wrong beliefs, without strictly speaking lying to them. Lay people deploy much ingenuity to perform this trick; politicians intent on deceit are a notorious case in point. Both philosophers and lay people take it to be axiologically significant (Pepp, 2019; Shiffrin, 2021). Language being multiply polysemous (Ludlow, 2014), we also use ‘lie’ in a broader sense on which any potential deception, even non-intended, counts as such (Saul, 2012, 1). Nonetheless, the extent of the phenomenon and its intentional character testifies to the reality of the distinction; and there is now significant empirical evidence that people are indeed sensitive to it, see below.²

I’ll present now the Grice-inspired adverbial version of saying accounts that I prefer (García-Carpintero, 2021a).³ I follow Williamson (1996/2000) in assuming that we have a pre-theoretical grasp of a specific kind of speech act that Williamson calls flat-out assertion, which he takes to be what we do by default when uttering sentences in the declarative mood (Williamson 1996/2000, 258).⁴ Proposals in the literature provide different accounts of the nature of flat-out assertion. Expressivist views invoke psychological states of their agents such as Gricean reflexive intentions (Bach and Harnish, 1979). Brandom (1983) and Geurts (2019) develop social normative accounts in terms of commitments that the agent incurs, to justify the assertion if challenged, retract it if poorly supported, etc. Williamson offers a lucid version of an Austinian social constitutive rules approach, on which a flat-out assertion with content p is by its very nature subject to the rule KR below. Alternative rules have been advanced, including the audience-involving norm I’ll assume here KPR, cf. García-Carpintero (2004), Hinchman (2020); on this proposal, flat-out assertions are acts of testifying:⁵

(KR)::

One must ((assert p) only if one knows p)

(KPR)::

One must ((assert p) only if one’s audience gets to be thereby in a position to know p).

We do things other than testifying by literally using declarative sentences: we make guesses, suppositions, we put forward propositions for the consideration of our audiences, we present fictional scenarios for the imagination, and so on. The aforementioned views characterize those other practices in their proprietary terms. Moreover, the views allow that the acts they characterize (flat-out assertion, in particular) can be made indirectly, as other speech acts can. An utterance of ‘Thanks for not browsing our magazines’ found in the bus station kiosk is not the expression of gratitude it literally conveys, but a request. Grice’s (1975) conversational implicatures are a particular case of indirect speech act—one in which an assertoric act is indirectly made by means of a declarative sentence.⁶ That act can be one of testifying—one constitutively beholden to KPR, on my assumptions here.

Thus, consider Grice’s (1975, 52) famous recommendation-letter example, and, for later use, let’s put it in a specific context. You are contemplating hiring a former student of mine, X, for your department; you tell me about the circumstances and ask me for my opinion on X’s philosophical capacities. I answer as Grice specifies: “X's command of English is excellent, and his attendance at tutorials has been regular. Yours, etc.” Let’s assume that KPR offers the right account of flat-out assertions—but the point could equally well be grounded on other views. It intuitively seems that the letter-writer has made one vis-à-vis the implied content—one beholden to the KPR rule. My performance would be wrong if I am not putting you in a position to know that X is no good at philosophy and not a good hire for your department, because he is in fact a first-rate philosopher for his career stage whom I intend my own department to hire at the next opportunity. If you find out later, you may angrily reprove me: “Why did you put it to us that X is no good at philosophy? You knew perfectly well that X is a first-rate philosopher, whom you yourself have hired!” The same applies to the false implicatures conveyed in the Athanasius and Williams examples above; on the sort of accounts I have discussed, they can be taken as (faulty, questionable) flat-out assertions.

By the same token, in the extended sense mentioned above my angry audience may well describe the implicature as a lie. In the stricter sense that the lying vs. misleading intuition manifests, however, it is a paradigmatic case of misleading without lying. This would be theoretically accounted for if (as saying accounts have it) lying in the strict sense requires not just to flat-out assert what one believes to be false—which on the outlined accounts may also obtain when the speaker misleads without lying—but to do it explicitly, by means of a sentence that conveys the content believed to be false and its force as close to the intuitively literal as it is feasible.⁷ Now, what is literally, explicitly conveyed by an utterance has good claims to count as the, or a semantic content of it.⁸

On Grice’s (1989) view, the semantic content of declarative sentences in a privileged sense is what is said by their utterances, and he offered two criteria that jointly help identifying it, formality and dictiveness. The first has to do with the extent that “the items or situations signified are picked out as such by their falling under the conventional meaning of the signifying expression rather than by some more informal or indirect relationship to the signifying expression” (Grice, 1989, 359). He admits that the second criterion is less clear-cut (Grice, 1989, 363), but, given his examples and indications, I take it that it corresponds to the degree or strength of speakers’ assertoric commitment relative to the given content; as he puts it, “a speaker’s alignment with an idea or thesis” (Grice, 1989, 367). Speakers can explicitly indicate that their epistemic standing with respect to a proposition is short of knowledge (and hence their assertoric commitment weaker) by embedding the proposition after ‘I guess—think—conjecture—imagine that …’, or hedging appositive phrases (Benton and van Elswyk, 2020); it can also be contextually conveyed. Grice also assumes that assertoric commitment is presented as stronger with respect to at issue (“ground-floor”, as he puts it, 1989, 362) content than backgrounded (conventionally implicated or presupposed) content.⁹ Standardly (even if defeasibly) speakers “stick out their necks” more with respect to what is at issue. Backgrounded, conventionally implicated content is assumed to be less directly relevant to addressing the issues under discussion. Presuppositions are taken to be already established material, shared knowledge for whose epistemic status the speaker doesn’t take individual responsibility.¹⁰ Grice takes the two criteria to be independent of each other (cf. Carston, 2002, 113–114). The adverbial account of lying helps itself to (versions of) them.

As an account of what is said—semantic content—Grice’s view is inadequate, however. Semantic theories fundamentally aim to articulate the meaning-properties of natural languages (what words literally mean), accounting in so doing, among other things, for their productivity and systematicity. It is wrong to rely for these purposes on a notion that requires Gricean dictiveness (García-Carpintero, 2021b). The literal content of utterances used ironically is not said in Grice’s sense. When ‘He is a good friend’ is uttered ironically to mean that the referent is disloyal, for Grice the speaker only “makes as if to say” that the referent is a good friend, there being zero assertoric commitment to it; but this is what a semantic theory should ascribe to the utterance.¹¹ It is doubtful that the literal content that semantic theories should ascribe to malapropisms (‘The vote was anonymous’, meaning that the vote was unanimous, Bach and Harnish, 1979, 33) is said in Grice’s sense.¹² The backgrounded contents of conventional implicatures are for him not said, but a semantic theory should countenance them; the same applies to the meanings literally conveyed by presupposition triggers. For one final example, semantic theories ascribe meaning to moods and other conventional force indicators that don’t contribute to Gricean sayings. Following Bach (1994), I think we should resort instead to an elaboration of the Austinian notion of locutionary act (cf. García-Carpintero, 2021b; Recanati, 2013), ultimately relying on what semantic theories have to offer to attain their proprietary explanatory goals for a precise specification of semantic content.¹³

Stokke (2017) convincingly argues that one can lie with conventional implicatures—by uttering ‘Lance Armstrong, an Arkansan, won the Tour de France’ in a context in which the audience wants to be told where Armstrong was born, and the speaker knows that he is Texan. Backing it up with empirical data, Viebahn (2019) and Viebahn et al. (2021) argue the same about lexically triggered presuppositions. For instance, one lies according to Viebahn (and his experimental subjects) by uttering ‘Did you know that John had a Mercedes?’ in a context in which it is at issue what brand John’s car is, and the speaker knows that it is not a Mercedes.

Both lying with conventional implicatures and lying with presuppositions are possible on the adverbial account. Like saying on Grice’s view, lying on this account requires (i) to present a content literally, explicitly, and (ii) to present it with full assertoric commitment. I’ll assume that this comes to present a content as known, so that the speaker can be challenged as follows: ‘Do you/we know p?’, ‘How do you/we know p?’, ‘You/we cannot know p!’¹⁴ On the locutionary notion of saying, this can obtain in flat-out asserting and also with both kinds of non-at-issue content (García-Carpintero, 2020); it fails to obtain when one hedges with ‘I think’, ‘I guess’, and so on.¹⁵ This is thus the adverbial account of lying:

(AL):

A lies in communicating proposition p to B if and only if:

• (1) A assertorically commits to p

• (2) A’s utterance says/makes explicit p

• (3) A believes p to be false

Saul (2012), Stokke (2016), and Marsili (2021), among others, offer alternative saying accounts; I won’t go here into my reasons to prefer AL (García-Carpintero, 2018, 2020), but some come up below. On saying accounts, the lying vs. misleading intuitions provide data contributing to delineating the semantic content of utterances (Borg, 2022; Michaelson, 2016; Saul, 2012; Stokke, 2016). Given AL, intuitive data about cases of lying vs. misleading are evidence constraining theoretical views on the semantics vs. pragmatics divide, through this criterion: a content that a speaker conveyed by uttering S while disbelieving it is not part of its semantic content if the speaker didn’t thereby lie to, but merely misled, her audience (Michaelson, 2016, 482). I believe that data including this favor a moderate contextualism, in contrast with minimalism and other views (García-Carpintero, 2021b).¹⁶ Of course, such data are no more than purported evidence; theoretical proposals have considerable leeway, including rejecting some of the relevant intuitions, or explaining them away.

The adverbial account predicts intuitive conflicts in cases of contents involving what Grice called generalized conversational implicatures which, in contrast to particularized ones, are conveyed in most contexts, and thus appear close to what the words themselves mean—and in fact are taken as such by theorists who give prominence to convention in fixing semantic content, cp. Lepore and Stone (2015), Stojnić (2021). They involve conventionalized indirection as in ‘Could you pass the salt?’ which, although it requires calculation from speakers unfamiliar with them, is interpreted by most as automatically as prototypically literal discourse; the Athanasius and Williams examples above are illustrations.¹⁷

In an empirical investigation of these issues, Weissman and Terkourafi (2019) show that, indeed, ordinary speakers’ intuitions about the lying/misleading distinction in these cases are conflicting. It is unfortunate that in this, as in other studies discussed in Sect. 2, it was not made clear to subjects that it was lying vs. misleading that they were supposed to tell apart—potentially allowing them to classify the cases with the extended use of ‘lie’ in which it doesn’t contrast with ‘mislead’. It may not be easy to experimentally prevent this. I find it encouraging however that results in this work appear anyway to confirm speakers’ sensitivity to the distinction, and that results align well with the predictions of saying accounts (Weissman and Terkourafi, 2019, 238–242).¹⁸ For subjects were doubtful that most cases of generalized implicatures count as lying; and the authors offer good reasons to doubt that cases that their subjects tend to count as such (for instance, those depending on cardinals) in fact involve indirection.¹⁹ I’ll come back to empirical data at the end of Sect. 2. Be this as it may, our discussion should concentrate on particularized implicatures as in Stokke’s Paul’s party example or the recommendation letter case; they are the prototypical cases of misleading without lying that all parties should accept.

2 The commitment account and its predicaments

I’ll now present the alternative account of the lying/misleading distinction and my critical take on it. I have explicated the Gricean criterion of dictiveness in terms of assertoric strength, and we saw that in his characterization of this notion, Grice assumed that in presenting a content by using a declarative sentence in default contexts we indicate a stronger degree of assertoric commitment than by merely hinting or implying it. As he puts it, dictiveness is an aspect of signification “connected with what the signifying expression (or its user) says as distinct from implies, suggests, hints, or in some other less than fully direct manner conveys” (Grice, 1989, 360). On the adverbial account, this is only true about how contents are presented by the choice of direct as opposed to indirect means. On the one hand, on account of irony, or malaprops, the act literally indicated might not be actually performed with any assertoric commitment; this is why AL includes (1), for otherwise speakers would lie vis-à-vis literal content when they are sarcastic, malapropian, or tell fictions.²⁰ On the other, the act actually performed in a paradigm case of indirection—a particularized implicature—may well be one of testifying, subject to KPR on the account assumed here—hence one with the highest strength of assertoric commitment; this is why AL has condition (2). Grice’s recommendation letter case and Dennis’s implied message to Rebecca illustrate this.²¹

The alternative commitment account purports to explain the difference between lying and misleading in terms of strength of assertoric commitment: its “central tenet” is that “liars take on a commitment that misleaders avoid” (Viebahn, 2021, 291).²² The lying/misleading distinction is on this view not sensitive to the semantic properties of expressions. I have granted some correlation between semantic content and degree of assertoric commitment; but the points in the previous paragraph show that it breaks down in many cases. This makes for a prima facie indictment of the commitment account as an acceptable explanation of the intuitive lying vs. misleading divide, because particularized implicatures should fall squarely in the second class in the sense of ‘lying’ we are elucidating.²³ How can commitment accounts like Viebahn’s, which relinquishes a condition like (2) in AL, line up with the intuitive contours of the lying vs. misleading distinction? As said, this is all that there is to it; this is just a superficial, intuitive distinction—which may play, as such, an evidential role in theoretically delineating real kinds like assertion or semantic content. I’ll now sum up my reasons against Viebahn’s response to this worry (García-Carpintero, 2021a).²⁴

Some philosophers (Borg, 2019, 2022; Fricker, 2012) argue that no implicature can come with true assertoric commitment, because they can be cancelled—this is one of several criteria that Grice (1975, 57–58) offers for them. For instance, in Paul’s party vignette Dennis might have cancelled the implicature by consistently uttering instead ‘I have to work; but I plan to attend Paul’s party nonetheless/but I don’t mean to suggest that I don’t plan to attend.’ The cancellation availability is supposed to allow the speaker “plausible deniability” of the implicature, which is taken to show the absence of proper assertoric commitment.²⁵

Viebahn (2017, 2019, 2020, 2021, Sect. 1) disagrees with Fricker and Borg; he grants that conversational implicatures may convey full assertoric commitment.²⁶ But he also enlists plausible deniability as his criterion for the lying vs. misleading distinction: lying, unlike misleading, requires assertoric commitment incompatible with plausible deniability. Viebahn contends that, in response to an accusation by Rebecca of having lied to her by uttering ‘I have to work’ in the Paul’s party example, Dennis “can offer the following sincere (albeit pedantic) response: ‘I didn’t lie. I didn’t claim that I wasn’t going to go to Paul’s party. I merely claimed that I had to work, which I did’. By contrast, Dennis cannot sincerely reply in this way if he lies by uttering [‘No, I’m not going to Paul’s party’]” (Viebahn, 2020, 3; cf. also Viebahn, 2017, 1370, Viebahn, 2019, 246).²⁷

Viebahn (2021, 300) turns the plausible deniability criterion into his characterization of lying. Lying in a communicative exchange relative to a content p that the speaker believes false, as opposed to misleading, occurs when the speaker cannot “consistently dismiss” the audience’s challenge to justify that he knows p. On the view about full assertoric commitment I am assuming, such challenges are in order when the commitment exists, being indeed a telltale sign of it. As Viebahn says, they can be properly answered by offering an acceptable justification; or the commitment may be retracted. The alleged distinguishing issue is for him whether they can be “consistently dismissed”: if they cannot, we have a case of lying; if they can, one of misleading intent. Viebahn doesn’t explain what he means by ‘consistent dismissal’; neither he, Fricker or Borg define ‘plausible denial’ either. It cannot just be narrow, formal logical consistency; but it doesn’t seem to be broadly conceptual consistency either. I’ll argue that neither plausible deniability nor consistent dismissal are up to the task of tracing an acceptable distinction between lying and misleading coincident with intuitions.

The problem lies in that ordinary speakers’ intuitions on the distinction concern contents of utterances in particular contexts, to which the contribution of context-dependent expressions and other sources of (roughly speaking, which suffices for my purposes) “ambiguity” like polysemy and homonymy have been settled. But such “ambiguities” in literal discourse open roads to “plausible deniability” and “consistent dismissal”.²⁸ I’ll illustrate this with force-indicators, which are “ambiguous” enough; but any other source would equally sustain the point.²⁹ Lying Dennis utters ‘No, I’m not going to Paul’s party’; but he can still (even if, as Viebahn puts it, “pedantically”) try to save face by making an utterance of the kind that Viebahn offers as his test: ‘I didn’t lie. I didn’t claim that I wasn’t going to go to Paul’s party. I merely expressed my intention at that time, which I had’. He can equally dismiss challenges: “I didn’t claim that I wouldn’t go to Paul’s party, I merely expressed my intention not to go at the time, which I had; I can’t see into the future any more than you can.”³⁰

Viebahn qualifies face-saving misleaders like Dennis as “pedantic”; more accurately, he also describes their retorts as “disingenuous” (Viebahn 2021, Sect. IV.2), which lying Dennis’s certainly are. Why are they? On my own diagnosis, because both had in fact claimed what they now say they hadn’t—on the account assumed here, becoming thereby beholden to KPR. Both lying Dennis and misleading Dennis had thus incurred assertoric commitment of the highest strength. Rebecca might thus angrily reply to Dennis in both cases: ‘Don’t bullshit me, of course you told me that you weren’t going to go to Paul’s party; that is what I had asked you about, what I wanted to know and what was only relevant to me, as you very well knew. You are dishonestly reinterpreting your words.’³¹ I already made the same point above about Grice’s recommendation letter case, when placed in the particular context I described.

This shows that the assertoric commitment account is inapt to capture the intuitive lying vs. misleading distinction. For Dennis was intuitively lying in the case for which, as I have just shown, he nonetheless retains “(pedantic and disingenuous) plausible deniability” and has a “consistent denial” available to him, as much as he did when he was aiming to mislead Rebecca. The (weak) sense in which in both cases they retain plausible deniability, and challenges can be dismissed, is this: affording an outwardly face-saving response, even if deceitful, and open to moral or legal reproof.³² Alternatively, we can say that the availability of Rebecca’s scolding reply shows that liars don’t really keep plausible deniability and consistent dismissal, in a stronger sense than this; say, affording a face-saving response, as appraised by an informed judge.³³ But then some misleaders aren’t allowed them either.

Like Stokke (2016), Viebahn (2020) presents the Paul’s party case as prototypical of lying—as one on whose classification all accounts should agree to stay on topic. If my diagnoses of the cases are right, by Viebahn’s own lights (i.e., by applying his own test) we can see why the commitment account doesn’t adequately capture the intuitions behind the lying/misleading distinction. In the weak sense just offered, plausible denial and consistent dismissal just come to the putative availability of an ostensible face-saving retort; as shown, the many wrinkles in literal discourse also allow lying speakers a good measure. A more demanding interpretation of plausible denial and consistent dismissal would also leave assertorically committed but merely implying speakers without them. On the demanding interpretation, these notions track only the first condition in AL. They are thus inadequate for the job that Viebahn wants from them because, as he agrees, some implicatures convey full assertoric commitment.³⁴

I have been insisting that the lying/misleading distinction is an intuitive one; it is hence eminently amenable to empirical study, and there are now some interesting results, cf. Wiegmann and Meibauer (2019) and Wiegmann et al. (forthcoming, Sect. 5) for good summaries. I already mentioned the one by Weissman and Terkourafi (2019), which their authors take to support saying accounts (see, however, Wiegmann, 2022). I saw his results too late to properly discuss them here. I'll just say that I find them not just consistent with AL, but better explained by it. Note in particular that one of Wiegmann's crucial additions to Weissman and Terkourafi's vignettes is to make explicit, as a question under discussion—hence a presupposition afterwards—, the implied false content as a potential answer). Other researchers (Mazzarella et al., 2018; Reins and Wiegmann, 2021; Wiegmann, 2022) take their results to go against such accounts, and to support instead commitment accounts. I’ll say something in response to close this section.

Mazzarella et al. (2018, 17) say their results show that “implicating is taken to be less committal than saying” (Mazzarella et al., 2018), which I have argued is true in general but false in some cases. They asked their subjects two sort of questions about speakers who have tended wrong information, some saying, some implying it. First, punishment questions concerning their normative appraisal. Their results here align with those of Bonalumi et al. (2020), which are consistent with AL. Their subjects take speakers to be committed as strongly to what they literally, explicitly say, as to what they convey by means of implicatures, when its significance to the audience is made sufficiently clear in the context. Bonalumi et al. target promises, but their results carry over to assertions; Mazzarella et al. (2018) similar results concern them. Now, Mazzarella et al. also asked trust questions, exploring whether their subjects would be prepared to rely on speakers afterwards. Their results here suggest that implying speakers “keep their reputation” more than saying speakers, which they take to support the claim quoted above. But Bonalumi et al. (2020, 361–362) also tested their subjects on this dimension and obtained inconsistent results, undermining the claim.

Reins and Wiegmann’s (2021) and Wiegmann et al. (forthcoming) contend that their results support commitment accounts. They do conflict with views like Fricker’s (2012), Stokke’s (2016) and Borg’s (2022) that assume that assertoric commitment must be expressed by declarative mood; but AL distinctively rejects this, and hence allows that generalized implicatures may lie someone in between. As far as I can make it out, these results otherwise align well with those in Weissman and Terkourafi (2019). Beyond contents literally conveyed by declarative sentences, their participants gave the most clear-cut diagnosis of lying to cases of deception by means of linguistically articulated presuppositions to whose truth the speakers were clearly committed given the issues at stake in the context. Subjects also in general considered generalized conversational implicatures closer to count as lies than particularized implicatures, as AL predicts. Like subjects in the studies by Weissman and Terkourafi (2019), those in these studies took cases involving cardinals—which many researchers wouldn’t consider implicatures—to be closest to lies with literal assertions.

The empirical results do not thus appear to conflict with AL; a case can in fact be made for the claim that AL fits better the data than rival accounts. Viebahn’s commitment account makes predictions inconsistent with folk intuitions for some particularized implicatures. Alternative saying accounts like Stokke’s make wrong predictions for presuppositions and fail to predict that some conventionalized implicatures yield in-between verdicts.

As I said, both philosophers and lay people find value in the lying vs. misleading distinction, moral or perhaps aesthetic (cf. Pepp, 2019; Shiffrin, 2021 and references there). The ingenuity we are prepared to put to mislead without lying witness it. Timmermann and Viebahn (2021) argue that the commitment account offers a better explanation for such value intuitions. Although I cannot go into this in any detail here, I’ll briefly mention aspects relevant to our discussion. On the adverbial account, the distinction is one of how, not of what; this might suggest that it upholds Saul’s (2012) and Williams’s (2002) revisionary view that there is in fact no real moral significance to it. But even if lying is only a matter of manner, on the social account here assumed the manner in question involves the misuse of an important social tool with a significant social function, the pooling of information. This might be elaborated along different lines to defend the default moral significance of lying—Pepp (2020, Sect. 5); see also Pepp’s (2019, 301–303) AL-friendly account of the aesthetic significance she claims for the distinction. Consistent with AL, in some cases lying and misleading may be morally on a par—say, when there is full assertoric commitment to the misleading content and inducing a false belief has high moral significance, as in Saul’s (2012, 73) peanut allergy case. Lying might even be morally better for addressees who “don’t have a claim to truth” like Kant’s murderer (Timmermann and Viebahn, 2021).

Let’s move now to consider the case of communication by means of pictures.

3 Lying and depiction

Intuitively, we can assert—we can incur the full assertoric commitment of flat-out assertions, testifying on the view assumed here—by means of pictures (Eaton, 1980; Greenberg, 2018; Kjørup, 1974; Korsmeyer, 1985; Nöth, 1997). If I draw you a map for how to get to my home from the nearest freeway exit, I present myself as telling/informing you how to locate it pictorically. If a report in the newspaper on an accident at the Grand Prix comes with a photograph of a wrecked motorbike, it tells us about how the vehicle looked after the accident. A drawing accompanying a newspaper report on a trial shows us how the defendant looked during his deposition. It speaks in favor of AL (and prima facie against, say, Borg’s, Fricker’s and Stokke’s views) that it straightforwardly allows for this: what constitutes the act is its being answerable to KPR, not the manner or vehicle by means of which it is performed.

At this point I need to say something about my assumptions on the nature of contents. There is a debate in the philosophy of mind about a distinction between mental states like beliefs and judgments—which we would naturally articulate linguistically in inner or outer speech—and states like imagistic visual experiences. Unfortunately, this has been cashed out as a debate on whether the contents of experiences are “conceptual” or otherwise, which suggests that the contents themselves of images differ from those of linguistic items. I reject such views. On an influential account propositions can be modelled as sets of worlds, and these in turn are understood by Stalnaker (1976) as determined by ways or properties the world might have. I endorse Stalnaker’s view that propositions/contents, both for pictures and linguistic items, are such properties (Kjørup, 1974, 220–221; Blumson, 2009; Greenberg, 2018, 2021; Maier, 2019; Abusch, 2020). I will not defend this view here; let me just note two points in support of assuming it for our purposes.³⁵

First, most current semanticists assume it; in particular, the view is assumed in recent ground-breaking work on the semantics of pictures just referenced, by Abusch, Greenberg and Maier, among others. Second, the view corroborates intuitively natural assumptions, such as that a literary fiction and its film adaptation share content, or that perceptual beliefs share content with the experiences that justify them. Taken as properties, propositions are finer-grained than the sets of worlds at which they are instantiated, thereby dodging well-known difficulties with the identification of propositions with such sets. They can be properties of situations smaller than worlds, as in truth-maker semantics (Fine, 2017; Yablo, 2014). They can specify impossible conditions (Berto, 2017). Pictures and linguistic expressions have different expressive properties: we don’t get double-negation with pictures, perhaps not even negation; in contrast, picture-content is rich, “analog”. But these can be explained as differences in the vehicles, not their contents (Grzankowski, 2015).³⁶

If we can assert with pictures, we can also deceive with them, as our intuitions also uphold (Nöth, 1997). Thus, in drawing the map to get to my home offered as an illustration above, I may (intentionally) deceive you, leading you astray by depicting a route that I know will lead you away from my home. Viebahn (2019, 246) reminds us how “the newspaper The Mirror lied to its readers by printing a manipulated photo of Lady Diana and Dodi Fayed on its front page” portraying them as if they were about to kiss. Korsmeyer (1985) discusses Paul Revere’s engraving of the Boston Massacre in this regard. Similar cases involve the moving image. Wilson (1986, 1, 202) mentions a montage of three shots in Welles’s 1947 Lady from Shanghai: a truck pulls out in front of a car; a woman’s hand presses a button; the car crashes into the truck. The montage induces us to “see” the pressing of the button as somehow causing the accident. This occurs in a fiction, but similar inferences are induced in assertoric cases, sometimes deceptively. Discussing the notorious case of Michael Moore’s 1989 Roger & Me, Currie (1999, 296) points out that the time frame presented in the film implies “that events had one kind of cause when in fact their actual time of occurrence made such causation impossible”. Moore’s films rank high on fan lists of deceptive documentaries,³⁷ together with classics like Flaherty’s 1922 Nanook and Algar’s Disney 1958 White Wilderness.

Not to beg any questions, I have presented these examples as cases of deception, but I find persuasive some of Viebahn’s (2019, Sect. 2) reasons to find in this medium an intuitive correlate of the distinction between lying in the more restricted sense, and misleading intent—although it seems to me that the distinction has a less clear foothold in the case of pictures, which AL explains along lines I’ll suggest shortly.³⁸ Viebahn offers the following example as an intuitive case of lying in the narrow sense: “Martha wants Nora to think that Oscar and Paula kissed. Martha knows that in fact Oscar and Paula never kissed. She carefully manipulates a photograph in such a way that it shows Oscar and Paula kissing and messages it to Nora with no further comment. When Nora receives the photo, she comes to believe that Oscar and Paula kissed”, (Viebahn, 2019, 243, like him I’ll call the exchange ‘P1’). He contrasts this with the following case of mere deceptive intent. This time “Martha’s goal is not only to convince Nora that Oscar and Paula kissed, but that they are a couple … Martha could … send several photos of Oscar and Paula standing outside their flat and carrying boxes into the flat. In this way, she could communicate to Nora that Oscar and Paula have moved in together, and mislead her into thinking that Oscar and Paula are a couple” (2019, 245–246, ‘P2’).

Viebahn offers two reasons for these exchanges to exemplify an intuitive lying vs. deceiving distinction in communication by means of pictures. The first makes by itself a good prima facie case for it. He compares P1 to a case in which Martha just says to Nora ‘Oscar and Paula kissed’, and he notes that “there appears to be no relevant difference between the two communicative acts. In both cases, Martha intends to communicate to Nora the proposition that Oscar and Paula kissed, which she believes to be false. And it would be entirely natural to report either communicative act as follows: […] Martha lied about Oscar and Paula kissing.” (Viebahn, 2019, 245)³⁹ I would add, in support of the AL account below: in both cases the content is intuitively communicated as “literally” or “explicitly” as it is feasible, given what each medium affords; more on this shortly.

Viebahn’s second consideration appeals to his unhelpful “plausible deniability” criterion and hence offers no support. He contends that, while in P2 Martha retains “pedantic but truthful” deniability, she doesn’t retain it in P1. I reject this on grounds by now familiar. On the one hand, Martha does retain in P1 a measure of the weak form of deniability. In the utterance case she can say, I didn’t lie about Oscar and Paula kissing; I thought it was clear to you that I was merely parroting what I had heard. In the picture case, I was only resending you a picture someone else had sent me.⁴⁰ Of course, she is just putting forward a (pedantic, disingenuous, and in all probability unsuccessful) attempt at face-saving; but the same applies to P2, if Martha sent the picture after having been asked explicitly whether Oscar and Paula are a couple, and then, when the deception is found out, she reproduces Viebahn’s formula: I didn’t lie about Oscar and Paula being a couple. I merely showed you that they moved in together, which is true. Alternatively, we could withhold plausible deniability in the first case using the stronger notion in Sect. 2; but then we should withhold it in the latter also.

Assuming thus an intuitive lying/misleading distinction for communication by means of pictures, Viebahn considers whether saying accounts can work here. He raises a “first, albeit minor worry: there appears to be no pre-theoretical understanding of saying that applies to presenting pictures” (2019, 250). But there is such a pre-theoretical understanding, as I’ll now establish, ‘showing’ being a better term for it than ‘saying’. It allows us to articulate a version of AL for lying in assertoric communication with pictures good enough for my purposes. As said at the outset, my aim is to establish a conditional claim: to the extent that the folk lying vs. misleading distinction has a foothold in the case of pictures, AL can be developed for depictive media to better account for it than rivals. I’ll offer three quotations pointing to an intuitive explicit vs. implicit distinction for communication with pictures.

Discussing his Imagined Seeing Thesis, Wilson (2011, 57) makes this intuitive point: “we regularly think and speak of ourselves as if we ‘see’ the fictional objects, events, and situations that movie shots and sequences depict […] These judgments about what we do and do not ‘see’ among the depicted fictions in a movie are not only ubiquitous, but they correspond to the fundamental distinction between those narrative items and occurrences that are presented on screen in a given stretch of the movie and those that aren’t. Thus, for example, it is fictionally true in the movie M (Fritz Lang, 1930) that a certain child murderer, Hans Beckert, […] meets a little girl Elsie Beckmann […] on a Berlin street, buys her a balloon, and subsequently murders her […] although viewers of M arguably ‘see’ the meeting with the murderer, Beckert, on the street and certainly ‘see’ the purchase of the balloon, they notoriously do not ‘see’ the murder of young Elsie. The latter is a fictional episode in the story, but its occurrence is merely (although quite distinctly) implied in the narration.” Similarly, a central part of what Almodóvar’s 2002 Talk to Her conveys is that Benigno rapes the comatose Alicia; but, unlike other plot features, we don’t “see” this in the film.

Eaton (2013) calls our attention to a striking example of the same contrast in Hitchcock’s 1972 Frenzy: “Early on in the film we are shown a scene of rape and strangulation that was quite graphic and vivid in its time. About thirty minutes later, we see the perpetrator escort another young woman up to his apartment. The camera follows them up the stairs and as he shuts the door, we hear him say the very same words that he uttered before attacking the first victim (‘You’re my type of woman’).” We thus infer that a similar crime occurs behind the door, and our suspicion is later confirmed; but this time we do not “see” it: “the door shuts in our face as the camera slowly tracks back down the stairs in silence and across the street” (Eaton, 2013, 378).⁴¹ Finally, Abell (2005, 55) offers a third illustration, to motivate a distinction between visible and depictive content: “black-and-white pictures do not necessarily depict black-and-white objects, and stick-figure drawings do not usually depict emaciated beings with gargantuan heads. Black-and-white pictures often depict colored objects and stick figure drawings often depict normally proportioned human beings [… but] we cannot arrive at this interpretation of their content by appeal to the way they look.”⁴²

In the quoted passages, Wilson, Eaton and Abell point towards an intuitive distinction between contents that pictures, as such, make explicit (contents that, I’ll say, are shown in them) and contents that are also communicated by means of them, but are not made “visible” by them—are not shown.⁴³ This makes it clear that, against Viebahn’s objection, the adverbial account does have a clear-cut intuitive foothold in this case too. This intuitive distinction supports the view that there is a semantics vs. pragmatics distinction for pictures worth to be theoretically articulated. The view, that is, that there are contents that are to be ascribed to pictures themselves, as such, relatively independently of the use that communicators make of them, and others that are merely speaker-meant.⁴⁴

This suggestion has in fact been taken up in debates about the meanings of pictures. Although this has not been investigated in contemporary philosophy as intensely as the linguistic case, there is already a considerably literature on it.⁴⁵ We cannot go into it here, nor is it needed for my goals. It suffices for my purposes that we find represented in these debates the very same positions as in the linguistic case. Putting aside the equivalent of “radical contextualist” views that would dismiss it, most writers would agree that there is a minimal properly depictive content. Different theories characterize it in different ways; on a view I like (Greenberg, 2021; Kulvicki, 2006), this would be the class of scenes that would project the shape depicted in the picture given some system of projection. This would capture the core of the intuition that (unlike their linguistic counterparts) pictures “resemble” the scenes they depict.⁴⁶ As in the linguistic case, there is a range of views on these core semantic contents of pictures, depending mostly on how constrained they are taken to be at the metasemantic level by facts about our visual system.⁴⁷ Minimalists would contend that this is all that there is to the semantic content of pictures; anything beyond it is just “pragmatic”.

As with corresponding views in the linguistic case, this would make the semantic content of pictures wildly remote from what intuitions like those articulated in the three quotations above, or the related intuitions about a lying vs. misleading distinction with pictures that Viebahn’s examples suggest. For—just to illustrate it—bare-bones content is even compatible with the depicted scene being a 2-D reproduction of the depictive surface. Because of this, most writers posit a still semantic, but richer, more determinate content for pictures, obtained by “filling up” minimal depictive content. Theoretical elaborations on this yield “moderate contextualist” views for depiction. I’ll illustrate the idea with a version of this.

Kulvicki (2020) assimilates minimal depictive contents to Kaplanian characters, and the sort of richer but still “semantic” depictive contents I am discussing now—which he calls pictorial content, what is shown in my terms—to Kaplanian contents; this is a view analogous to Stanley’s (2000) indexicalism. Even accepting a framework of this sort, there is still plenty of room for disagreement. Thus, for instance, Kulvicki himself opts for pictorial contents that are general, while Greenberg (2018) defends singular contents for pictures, which I take to be more in line with the intuitions just mentioned. Terrone (2021) provides another account, closer to Greenberg’s proposals (and to my own views, for what it is worth).

We don’t need to go further into these debates to appreciate that AL has the resources needed to deal with pictures. We should understand ‘makes explicit’ in the second condition relative to a notion of pictorial content related to what pictures show, on which the views just discussed theoretically elaborate, which as we have seen has intuitive support. Thus understood, and to the extent that our intuitions validate a folk lying vs. misleading distinction for pictures, the account retains its main methodological virtue: it captures a pretheoretical, to a good extent indeterminate nominal kind, which can be deployed in abductive arguments on views about depiction like those just mentioned. But, of course, by itself it doesn’t select one of them; even a radical minimalist account on which only Kulvicki’s bare bones content is the semantic content of pictures may be defended, by rejecting the intuitions or explaining them away. We already know from extensive debates about it how controversial the notion of semantic content is in the linguistic case. Perhaps this is only a consequence of the fact that there has been less debate on pictorial content, but it is my impression that the intuitive notion of what is shown by pictures, as opposed to what is merely “implicated”, is shakier than the corresponding distinctions for the linguistic case. This would explain the comparative shakiness of the lying vs. misleading distinction with pictures. As said, this should be studied empirically.

Aside from the “minor worry” I have thus just answered, Viebahn raises two objections to the sort of account I have outlined. The first is “the challenge of propositionality”, that “many theorists hold that pictorial content is not propositional” (Viebahn, 2019, 251). I have already replied to this too. ‘Proposition’ is a theoretical term in philosophy. Yes, there are accounts of propositions on which pictures don’t express them; but the one I have suggested is a coherent, available choice, which allows for them to perform the tasks in their job description. AL is not meant to be neutral on all relevant philosophical debates and clearly it is not.

Viebahn’s second objection is another challenge, to offer “a notion of pictorial content that is neither too broad, nor too narrow” (2019). He illustrates this by critically discussing from this perspective Abell’s and Blumson’s accounts of pictorial content. I don’t need to defend them to show that this objection misinterprets the role that the lying vs. misleading distinction should have in our theoretical endeavors. This is not, I have been insisting, a real kind that our theoretical categories (including pictorial content) are mandated to neatly capture. We just need to offer a philosophical characterization of its superficial features that fits well enough its intuitive profile, which is, I have argued, just what AL provides, alone among contenders. If forthrightly applied to fix what counts as a lie, minimalist views like Borg’s (I argued above, against her own claims), or corresponding ones for pictorial content, yield intuitively inadequate notions. But if they have more overall abductive support than their rivals, their proponents should at most explain away our intuitions. I don’t see any reason why they would be unable to do so: they can simply acknowledge an intuitive notion of what is “literal”, or “made explicit” by either linguistic or pictorial means. They just reject that the proper theoretical notion of semantic content should align with it.

4 Conclusion

In this paper I have articulated a particular version of a very natural view on the intuitive lying vs. misleading distinction, the saying account, on which the distinction is “adverbial”: it depends on how the assertoric acts that must be present in both lying and misleading attempts are conveyed. I have shown how the view naturally extends to the case of lying vs. misleading with pictures. I have defended saying accounts against the revisionary commitment account that Viebahn and others have been promoting, aiming to preserve its methodological virtues: it provides us with a good tool for the abductive investigation of theories on the semantic vs. pragmatic divide in different media. I have emphasized the underlying ontological issues: while we have good reasons to stick to the hypothesis that the latter distinction tracks a real kind—both in linguistic and depictive media—the lying vs. misleading distinction is just a superficial, nominal kind that doesn’t bend to substantial theoretical discernment.