1 The poet: a liar by profession?

Consider the oft-quoted passages above. They instantiate two rather radical, opposite views. The first one, defended by Hume, is that all fiction-writers are liars. Famously endorsed by Plato, who did not wish to include the mendacious poets in his Republic, this view exerted influence on philosophers throughout history: Pascal defined poetic imagination as the “queen of lies and error”, and about a century later Hume penned the harsh comment above.

Nowadays, tides have changed. The thesis that fiction writers are liars strikes most contemporary philosophers as risible—little more than a historical curiosity. Surely fiction and lying have an important trait in common: typically, they both involve saying something that isn’t (believed to be) true. But the current consensus is that it would be erroneous to conflate these two concepts, since they are importantly distinct. Many philosophers nowadays (like most folk) rather side with Sir Phillip Sidney’s view that fictional statements aren’t lies—however, as we shall see, exceptions abound.

There are excellent reasons to side with Sydney. It would be rather odd to claim that the falsities contained in a work of fiction are lies. Take, for instance, the incipit of Jorge Luis Borges’s The Lottery in Babylon: “Like all the men in Babylon, I have been proconsul; like all, I have been a slave. I have known omnipotence, ignominy, imprisonment.”. Nobody would argue that Borges’s opening statement is a lie, even if Borges surely believed it to be false—he was never a proconsul, nor a slave; nor did he share this fate with “all the men in Babylon” (among other reasons, because he never lived there). But why do we judge that the Borges has not lied in writing this statement? Sydney’s plausible suggestion is that lying requires making assertions (“telling things for true”), whereas authors of fiction (“poets”) don’t present their stories as true. Accusations of mendacity, then, aren’t in order. The falsities we find in fictional works aren’t lies, because they aren’t affirmed.

This explanation is simple and appealing. But it’s not the only explanation on the table. Some philosophers think that a deceptive intent is essential to lying, and an alternative explanation follows from this view. If Borges’s incipit contains no lies (the story goes), it’s because Borges has no intent to make his readers believe something false. He doesn’t want to convince them that that he was a slave and a proconsul, or that he lived in Babylon. Lies differ from fiction because lies, unlike fictions, aim to deceive.

Which of these two views better tracks the distinction between fiction and lying? And can genuine lies be found within works of fiction? This paper aims to answer these two questions. Its goal is to determine what grounds the distinction between lying and fiction, and to establish whether these concepts really are mutually exclusive.

2 Fiction versus lying

2.1 Defining lying, and distinguishing it from fiction

How does lying differ from fiction? Let’s start by considering what lying is. There is consensus that stating what you believe to be false is a necessary condition for lying: you cannot lie unless you explicitly say (as opposed to imply) something that you believe to be false:Footnote 1

Lying-Nec::

A speaker S lies only if S states that p and S believes that p is false

Lying-Nec identifies necessary conditions that aren’t jointly sufficient to determine whether an utterance is a lie. Alone, it is unable to distinguish lying from fiction. Consider Borges’s example once again. Writing The Lottery in Babylon’s incipit, Borges stated something that he believes to be false (that he was a proconsul, a slave, and so forth). The incipit satisfies Lying-Nec, but it’s not a lie.

To distinguish lying from fiction, Lying-Nec needs to be narrowed down. We need a criterion that can tell lying apart from ‘non-mendacious falsities’Footnote 2: fictions, but also ironic statements; jokes; teasing remarks; hyperboles; metaphors; euphemisms, and the like. Historically, philosophers have offered two competing solutions: Deceptionist accounts of lying and Assertionist accounts.

Deceptionist accounts complement Lying-Nec with an ‘intention to deceive condition’ (IDC below):

  • Deceptionist definitions:

    S lies to A iff:

    (a) S states that p

    (b) S believes ¬p

    (IDC) S intends A to believe pFootnote 3

For the Deceptionist, non-mendacious falsities aren’t lies because they aren’t meant to deceive. If Borges isn’t lying, it’s because he isn’t attempting to deceive his readers (he isn’t trying to convince them that it is actually true that he was a proconsul, a slave, etc.). But there is an alternative approach to explain why fictional statements of this sort aren’t lies. One can complement Lying-Nec with the requirement that the speaker must genuinely assert that p (AC below):

  • Assertionist definitions:

    S lies to A iff:

    (a) S says p

    (b) S believes ¬p

    (AC) In saying that p, S asserts that pFootnote 4

On this view, the reason why non-mendacious falsehoods (fictional, ironic, metaphorical utterances) aren’t lies is that they aren’t genuinely asserted. Assertionists side with Sydney’s explanation of the distinction between lying and fiction: if Borges’s statements are not lies, it’s because Borges “nothing affirms” (condition AC isn’t met). Whether Borges had an intent to deceive is irrelevant to determine whether he was lying.

2.2 Different grounds for the fiction/lying distinction

Deceptionism and Assertionism yield different explanations for the intuitive difference between lying and fiction. What grounds the distinction is either the deceptive intent of the speaker (for Deceptionism), or the force of the utterance (for Assertionism).Footnote 5 There’s more than one way, however, to interpret the theoretical implications of each view for the lying/fiction distinction: both Assertionism and Deceptionism admit a Weak and a Strong interpretation.

Before I move on to discuss Weak and Strong varieties, let me introduce a terminological stipulation. In what follows, I will use the expression “utterance in fiction” to refer to any utterance contained in a work of fiction. If I prefer “utterance in fiction” to the more elegant “fictional utterance”, it’s because philosophers don’t agree that all the utterances contained in a work of fiction (i.e. all “utterances in fiction”) are fictional utterances. While “unitarians” (Friend, 2008; García-Carpintero, 2013, 2020, 240; cf. Stock, 2017) are happy with the equivalence, “patchwork theorists” (Currie, 1990, 48; Konrad, 2017; Searle, 1975, 332) claim that fictional works are a mixture of fictional utterances and genuine speech acts performed by the author in propria persona. Given this controversy, it’s preferable to frame the disagreement between Deceptionists and Assertionists as a disagreement about the status of utterances in fiction (whether or not they can be lies, and under which conditions). This way of articulating each position remains neutral on the broader question of which of utterances occurring in a work of fiction are genuine fictional utterances.

With this clarification out of the way, we are ready to explore Strong and Weak varieties of Deceptionism and Assertionism. I’ll start from the former family of views. On a Strong reading, Deceptionism may be interpreted as claiming that utterances in fiction are never lies, because utterances in fiction are never meant to deceive:

SD::

Necessarily, utterances in fiction are not intended to deceive, so they cannot be lies

SD, in turn, entails that “Fictions Never LieFootnote 6” (FNL):

FNL::

Utterances in fiction cannot be lies

In the next section (§3) we shall see that there are several ways in which fictions can be designed to intentionally deceive their audiences: SD is false, and blatantly so. Luckily, Deceptionism need not be interpreted in this strong way. A more cautious Deceptionist take is the following: whenever an utterance in fiction is intended to deceive, it’s best classified as a lie. This yields a more plausible account of the lying/fiction distinction (call it Weak Deceptionism):

WD::

If a believed-false utterance in fiction is intended to deceive, it’s a lie. Otherwise, it’s not a lie.

Applied to fiction, Assertionism also comes in two varieties. On a strong reading (call it Strong Assertionism), it holds that utterances in fiction cannot be lies, because utterances in fiction cannot be asserted:

SA::

Necessarily, utterances in fiction are not asserted, so they cannot be lies.

Strong Assertionism (SA) sides with Sydney’s views about fiction. Like Strong Deceptionism, Strong Assertionism entails FNL, i.e. that fictions never lie (although this conclusion is, of course, determined by altogether different motivations). Arguably, SA is the most influential view in the philosophical and narratological scholarship on fiction; its most recent and systematic articulation is due to Mahon (2019).Footnote 7

Assertionism admits a weak interpretation as well. Weak Assertionism (WA) grants that some utterances in fiction are genuinely asserted. When an assertion in fiction is believed to be false, it qualifies as a lieFootnote 8:

WA::

If a believed-false utterance in fiction is asserted, it’s a lie. Otherwise, it’s not a lie.

Weak Assertionism admits a “patchwork-theoretic” and a “unitarian” interpretation. It’s easy to see that endorsing a patchwork theory (jointly with an Assertionist definition of lying) naturally leads to Weak Assertionism: if some utterances in fiction are genuinely asserted, some utterances in fiction can be lies. Crucially, Weak Assertionism is open to unitarian positions as well. Unitarian views do not rule out the possibility that an utterance in fiction could be both asserted and fictional.Footnote 9 The takeaway is that Weak Assertionism admits different assumptions about the mereology of fiction. While being aware of these complications is important, I shall leave them aside in what follows, since they are not of primary concern. At this point, keeping track of the different positions may be difficult: Table 1 offers a helpful summary.

In all their differences, the four views under consideration converge on two answers to the initial question raised by the comments of Pascal, Plato, Hume, and Sidney—namely, whether fictions can lie. The Strong version of each view denies (for different reasons) that fictions can lie. Both Weak versions concede that sometimes fictions lie (although, again, for different reasons). To establish whether fictions can lie (whether FNL is true), we’ll have to assess the plausibility of each of the four views summarised in Table 1. This is what I set out to do in the rest of the paper, starting from evidence that bears on the validity of Strong Deceptionism.

Table 1 Four views about the distinction between lying and fiction, summarised
Full size table

3 Deceptive fiction: an overview

I anticipated that Strong Deceptionism is false, because it incorrectly assumes that fictions are never meant to deceive. It’s now time to justify this claim. There are two main ways in which fictions can be designed to deceive their audiences. First, authors can aim to deceive their audiences about what occurs in the story (deception about the fictional world). Second, they can aim to deceive their audiences about what is actually the case (deception about the actual world). Let’s consider these two species of deception in turn.

Intended deception about the fictional world is, in a way, a puzzling phenomenon. Suppose that we endorse the naïve view that fictional utterances simply stipulate fictional truths.Footnote 10 As Culler (2004) puts it, when the novelist writes something in the fiction, “she cannot be wrong […]”; such is “the power of invention, of incontrovertible stipulation”. If Conan Doyle writes that ‘Sherlock Holmes lives at 221B Baker Street’, he stipulates that it’s true in this fiction that Sherlock Holmes lives at 221B Baker Street. This proposition is true in the fiction in virtue of the fact that there is a sentence in the book that says so. We may naively conclude that deception about the fictional world is simply impossible, for whatever is recounted in the fiction is true by stipulation—there can be no mistakes.

Yet authors can clearly deceive their audiences about what happens in the fictional world. They may deploy an ‘unreliable narrator’, who can be insincere or mistaken about the events occurring in the story (Booth, 1961, 158–159). They may also adopt multiple narratorial ‘voices’, whose accounts of the fictional events are inconsistent. Poe’s The Tell-Tale Heart and Palahniuk’s Fight Club are prominent examples of deceptive stories recounted by an ‘unreliable narrator’; Kurosawa’s RashomonFootnote 11 adopts multiple, inconsistent voices. These fictions involve intended deception: until the correct version of the narrated events is revealed, we are bound to be deceived about what really happened in the fictional world.

Intended deception about the actual world occurs when fictions are designed to cause their audiences to form incorrect beliefs about the actual world. It is thought that Walter Scott’s historical novels (together with many other Scottish works of his time, like the literary forgeries of James MacphersonFootnote 12 and the bogus essays of the ‘Sobieski Stuarts’Footnote 13), contributed to promote false beliefs about Scottish folklore and history (Hobsbawm & Ranger, 2012, 48). If this was Scott’s intention, some of his works would be an example of fiction that intentionally deceives about the actual world.

Fiction is often designed to convey misleading stereotypes and deceptive misconceptions, too. Examples abound in early American literature and cinema, especially in relation to its disturbing representations of African American (Chaleila, 2020) and Native American (Barnett & Walker, 1975; Berny, 2020; Rollins, 2011) characters. It is sadly known that in these works Native Americans and African Americans are often represented as either vicious and violent, or submissive and unintelligent. Works of fiction have been used to explicitly disseminate white supremacist misconceptions and conspiracy theories, too. The fictional novels of William Luther Pierce (most famously The Turner Diaries) promote the idea that Zionists secretly control Western governments, and that various forces are conspiring to subdue, oppress, and eliminate white people. These novels have been taken to be faithfully depicting reality by many readers: the Turner Diaries are thought to have incited numerous acts of violence, with experts estimating that over 40 terrorist attacks and 200 killings were directly inspired by Pierce’s novel (Berger, 2016).

Insofar as some of these fictional works sincerely reflected the racists attitudes and xenophobic delusions of their authors, it may be countered that, while these works contain many false and vicious statements, their authors lacked an intention to deceive: their work simply reflects their racist ideals and historical misconceptions. If we accept this premise, these works are no less objectionable, but they don’t represent a challenge to SD.

In several cases, however, a deceptive intent is harder to deny. Take the example of Thomas Dixon Jr., an American author who wrote various works of fiction romanticising white supremacism in the Southern States—famously, bestselling novels like The Leopard’s Spots: A Romance of the White Man's Burden—1865–1900 and The Clansman: A Historical Romance of the Ku Klux Klan (the latter of which was turned into the infamous motion picture Birth of a Nation). Invariably displaying realistic ambitions, Dixon’s works are plagued with historical and factual inaccuracies, which some commentators assumed to be deliberate.Footnote 14 If these commentators are right, Dixon’s works of fiction were deceptive by design, not by accident.

Be it as it may, it is undeniable that fiction can be an effective form of propaganda, and can be intentionally designed to convey false narratives meant to manipulate public opinion and to distort the historical record. Utterances in fiction can be deceptive by design (cf. Stokke 2023, sect. 4.2), both about the fictional world and about the actual world. This means that Strong Deceptionism is false, and that little can be done to redeem it. My discussion, from now on, will focus on the remaining three positions, starting from Weak Deceptionism.

4 Intended deception about the fictional world

We saw that fictions can be deceptive by design. What needs to be established, now, is whether some of these deceptions amount to genuine lying, as claimed by Weak Deceptionism. This section considers whether intended deception about the fictional world amounts to lying; the next one covers deception about the actual world. Let’s start by considering an excerpt from Chuck Palahniuk’s novel Fight Club:Footnote 15

(1):

Tyler and I met at a nudist beach. He asked me: “Do you know what time it is?”

Without doubt, (1) is a deceptive statement: the reader doesn’t know it yet, but Tyler and the narrator are the same person. Therefore, Tyler and the narrator cannot have met at a nudist beach (or anywhere else), and Tyler cannot have asked the narrator the time. Palahniuk included (1) in the story to trick his readership into falsely believing that (1) is true. According to Deceptionist definitions of lying, (1) is a lie.

But here’s a problem. Intuitively, by writing (1) Palahniuk hasn’t lied to his readers.Footnote 16 He merely set them up for a plot twist. Readers may feel disappointed and fooled when they realise (later in the novel) that Tyler and the narrator are the same person. Yet, it would be intuitively inappropriate to accuse Palahniuk of being a liar for introducing a plot twist into the story.

To better appreciate this point, it can be useful to compare (1) to the deceptive utterances found in non-fiction. Consider, for example, Lance Armstrong’s autobiography It’s Not About the Bike (co-authored with Sally Jenkins). After the book’s publication, it came to light that Armstrong had been using illegal substances throughout his career; “much of what was written [in the text] turned out to be mendacious” (Bury, 2013). Readers complained that numerous statements contained in the books were outright lies. There is little question that they are, and the outrage caused by Armstrong’s biography is surely justified (Mahon, 2019). The same outrage would be intuitively out of place if addressed to Palahniuk. It’s unsurprising, then, that Palahniuk’s book did not generate the accusations of mendacity and the outrage that accompanied Armstrong’s.

Generalising, there seems to be an intuitive difference between the deceptive statements found in non-fiction and those found in fiction. It is quite straightforward that (at least some of) the intentionally deceptive statements that we find in works of non-fiction are lies. By contrast, utterances in fictions that are meant to deceive about the fictional world, like (1), are intuitively not lies. This is a problem for Weak Deceptionism, which fails to appreciate the distinction, and classifies both kinds of statements as straightforward lies.

But Weak Deceptionism might only need a minor tweak. Perhaps genuinely intending to deceive requires an intention to make someone believe that a given proposition is actually true (true in the actual world). Clearly, Palahniuk has no intention to trick his audience into believing that (1) is true in the actual world. Reinterpreting the ‘intention to deceive condition’ in this way has the welcome consequence that (1) isn’t classified as a lie.

The problem with this solution, however, is that it rules out the possibility of lying about possible worlds in general, and fictional worlds in particular. Suppose that a friend asks me how Moby Dick ends, and I reply:

(2):

Ahab finally kills the whale.

It seems pretty clear that I would be lying. Yet, I only intend my friend to believe that (2) is true in the fictional world of the story, not in the actual world. The revised criterion is therefore incorrect, because it rules out the possibility of lying about fiction. This is undesirable, because it’s possible to lie about propositions under the scope of modal operators, including fictional operators.Footnote 17

Perhaps some further epicycle can be added to Deceptionism, to deliver a criterion that includes (2) without excluding (1).Footnote 18 I am doubtful that this would help much, however. As we are about to see, Weak Deceptionism also has trouble accommodating fictions designed to deceive about the actual world.

5 Intended deception about the actual world

5.1 A case study

Commenting on the historical inaccuracies found in recent movies (Oppenheimer, Napoleon) and series (The Crown), British columnist Simon Jenkins expressed concerns over the practice of “deliberately telling lies about the living or the deceased”, which (he argues) is becoming increasingly common in cinematic works of fiction (Jenkins, 2023). Similarly, Masha Gessen criticised the HBO series Chernobyl for “crossing the line from conjuring a fiction to creating a lie” (Gessen, 2019). As these examples illustrate, we sometimes speak of deceptive fiction as containing lies. Is the term used loosely in these cases (as when we refer to honest mistakes as lies), or can deceptive fiction contain genuine lies, just like non-fiction?

To address this question, let’s focus on explicit statements contained in literary works of fiction that purport to narrate actual events, which best illustrate the possibility of lying in fiction. Consider the following passage from Victor Hugo’s Les Miserables:

(3):

The artillery had to wait until [the soil] became a little firmer before they could manoeuvre. Napoleon was an artillery officer, and felt the effects of this.

The chapter from which this excerpt is taken, Waterloo, is a long digression on what happened on the 18th of June in Waterloo, when France faced a coalition of European forces. Here Hugo displays an unmistakable intention to convey a reliable account of what happened during the battle. Knowing this, many readers take (3) to communicate something that is true not only in the world of the story, but also in the actual world (cf. García-Carpintero, 2019, 2020; Marsili, 2023).

Can utterances of this sort be lies? Suppose that, instead of (3), Hugo had included the following incorrect statement:

(3-F):

Napoleon had never been trained to be an artillery officer

Imagine that Hugo added (3-F) (‘F’ as a reminder for ‘False’) to the story because he wanted to trick his readers into believing, incorrectly, that Napoleon’s lack of artillery expertise was the cause of his defeat in Waterloo. Would (3-F) be a lie?

Weak Deceptionism gives a positive answer. Hugo included a statement he believed to be false in the book, with the intention to deceive his readers. By WD’s light, he is clearly lying. Weak Assertionism concurs, but offers a different explanation. For the Weak Assertionist, (3-F) is a lie because it is a genuine assertion about the actual world. Since Hugo asserted what he believes to be false, WA classifies (3-F) as a lie. But whether (3-F) really is asserted is up for dispute: Strong Assertionism disagrees. While it’s true Hugo intended to convey that (3-F) is actually the case, Les Miserables is a work of fiction, so Hugo doesn’t explicitly take responsibility for the truth of its content. This disqualifies (3-F) from being a genuine assertion: (3-F) cannot be a lie.

Which of these views is right? Typically, philosophers appeal to their own intuitions to settle questions of this sort. The problem, however, is that in these crucial cases intuitions tend to diverge. Strong Assertionists report the intuition that (although deceptive) statements like (3-F) are intuitively not genuine assertions nor lies. The Weak Assertionists and the Weak Deceptionists retort that intuitively (3-F) is a genuine lie (either because it’s asserted, or because it’s meant to deceive). Intuitions alone aren’t likely to settle the disagreement here. We need to determine whether utterances like (3-F) are lies on independent grounds.

5.2 The lying-misleading distinction

To make progress, we may regard the current stalemate as involving a disagreement on whether these deceptive utterances in fictions are merely misleadingFootnote 19 (as the Strong Assertionist would have it) or genuine lies (as argued by the other two views).Footnote 20 The advantage of framing the disagreement in these terms is that we can now take advantage of linguistic tests that rely on pre-theoretical intuitions about which moves are available to the participants of a conversation, instead of theory-laden intuitions about what lying is.

There’s relative consensus between linguists and philosophersFootnote 21 that the lying/misleading distinction parallels a distinction between asserting a proposition and indirectly communicating it. A misleading statement communicates something false without asserting it (Michaelson, 2016; Saul, 2012; Stainton, 2016). Philosophers also tend to agree that the availability of felicitousFootnote 22denials is evidenceFootnote 23 that an utterance is not explicitly stated (and therefore not a lie), but at most implied (i.e. at most misleading). A felicitous denial is a statement by means of which the speaker successfully cancels a potential interpretation of their utterance (in the Gricean sense of ‘cancellation’, cf. Grice 1989). In other words, a misleader (but not a liar) should be able to felicitously use expressions like (M) and (C) to take back a misleading interpretation of their statements (cf. Marsili, 2023):

(M):

Sorry, but you misunderstood me: I didn’t mean to suggest that p is actually true

(C):

In fact, p is false

To illustrate how this works, let’s compare a mendacious assertion with a misleading implicature. Imagine that I know that (3-F) is false, and I utter (3-F) in a serious conversation about French history. Accusing me of lying would be perfectly appropriate. In reply to the accusation, I would be unable to felicitously back off with (M) or (C).Footnote 24 If I added (C), I would contradict myself; if I replied with (M), my comeback would have the paradoxical flavour of a Moorean assertion. In both cases, my denial would “misfire”: it would be conversationally inadmissible. From the unavailability of (M) and (C), we can infer that I’ve lied rather than misled.

Now suppose that in the same context I uttered the true but misleading sentence (3-M) instead, knowing that (3-F) is false, and intending to trick my audience into believing that (3-F) is true:

(3-M):

My history professor told me that (3-F) Napoleon had never been trained to be an artillery officer

If someone objects that Napoleon was actually a trained artillery officer, I could easily back off with constructions like (M) or (C). I may explain that my purpose was to expose my history professor’s utter incompetence, since (3-F) is blatantly false. The test correctly classifies (3-M) as a misleading statement, rather than a mendacious assertion.

Attending to the availability of (M) and (C) can help us determine whether deceptive utterances in fiction like (3-F) are best classified as misleading or lying. Imagine that a critic interviews Hugo and presses him about the presence of the false statement (3-F). Hugo could easily reply with something like (M) or (C): “You misunderstood me, I added that detail merely for colour—I didn’t mean to suggest that Napoleon was an incompetent artillery officer. In fact, he was a very competent one!”. On top of being felicitous, this clarification would be plausible (in the right context). Oftentimes, considerations about plot design and enjoyability outweigh the importance of historical accuracy, from the author’s perspective. And when critics point out inaccuracies in a work of fiction with realistic ambitions, it isn’t uncommon for authors to resort to replies along these lines.Footnote 25 This suggests that (3-F) is best classified as a misleading statement, as opposed to a genuine lie. The same verdict is reached for any fictional utterance designed to deceive about the actual world: authors of fiction can felicitously back off with expression like (M) and (C).

This doesn’t mean that authors of deceptive fictions are off the hook for the deceptions that they concoct. Quite the contrary: they can be criticised for purposefully attempting to deceive their audiences.Footnote 26 Going back to our example, we may suspect that Hugo is just making excuses during his interview. Perhaps adding (3-F) was not needed for the economy of the story, and we may know that Hugo’s true goal was precisely to deceive his readership. In this context, it would be appropriate criticise Hugo for his ruse. But the contention here is simply that misleaders can felicitously make excuses—not that these excuses are necessarily plausible or convincing,Footnote 27 nor that are necessarily sincere.

If (as argued) statements like (3-F) aren’t genuine lies, Deceptionism is in trouble. This view turned out to clash with our intuitions about fiction that deceives about the fictional world (as noted in §4), and with linguistic data about fictions designed to deceived about the actual world (i.e. the test for the lying/misleading distinction just discussed).Footnote 28 The results of the test put pressure on Weak Assertionism too, for also this view classifies statements like (3-F) as genuinely mendacious, rather than misleading. Still, the test does not provide conclusive evidence; in the next section, I will review some further arguments for classifying some utterances in fiction as lies.

5.3 Assertions in fiction?

Despite the results of the test, it’s undeniable that many fictions talk, in some important sense, about the actual world. In many realistic works (including Les Miserables), this connection is so obvious that denying utterances in fiction the status of assertions might seem to go against common sense. Let’s consider the case for and against classifying them as assertions.

Dixon (2022, cf. also 2020, §4) offers an excellent reconstruction of the disagreement between Weak and Strong Assertionists on these matters.Footnote 29 Dixon (2022, 117) begins by presenting three conditions for asserting which, according to Mahon’s (2019) influential defence of Strong Assertionism, cannot be satisfied in a work of fiction.Footnote 30 On this view, a statement isn’t asserted unless:

  1. (i)

    (Affirm) It affirms that the depicted persons and events occurred or existed prior to the existence of the text;

  2. (ii)

    (Liability) It’s subject to a norm of accuracy (it may be faulted for being false);

  3. (iii)

    (Explicit) It’s explicit, i.e., forms ‘what is said’, not implied, by the uttered/written sentence.

This characterisation of assertoric statements could certainly be refined. For instance, Affirm does not cover assertions about a priori truths, like mathematical assertions (e.g. “three is a prime number”). But we can easily loosen Affirm to fix this problem, by interpreting Affirm as requiring that assertions must present their content as true (as opposed to something that “occurred or existed”).Footnote 31 Read in this way, the proposed characterisation of assertion is more plausible: it holds that assertions (i) present their content as true, (ii) can appropriately be criticised for being false, and (iii) are explicit, rather than implied.Footnote 32

This characterisation correctly rules out “implicit thematic statements”: statements that aren’t included verbatim in the text, but that a sufficiently attentive reader can easily extrapolate (Lamarque & Olsen, 1994, 324; Mikkonen, 2009). To illustrate, a thematic statement in Le Avventure di Pinocchio is that lying is bad (or that lies don’t get you far in life, or some periphrasis of the sort). This message is presented as true by Carlo Collodi (ii), but is not asserted, because it doesn’t satisfy condition (iii): it’s not explicitly stated in the story.

What shall we make of claims that are instead explicitly included in the fiction, like (3-F)? Dixon (2022) notes that, even by the Strong Assertionist’s strict standards (i-iii), these statements should be classified as genuine assertions. (3-F) is explicit in the sense required by (iii): it states, word by word, that Napoleon had never been trained to be an artillery officer. This statement is also (i) presented as true by their authors, and (ii) it can be faulted for its falsity (cf. §5.2). The idea that assertions must satisfy (i)-(iii), which appeared to support Strong Assertionism, now lends support to the opposite view (Weak Assertionism).

In reply, the Strong Assertionist might insist that (i-iii) cannot be jointly satisfied in fiction. Or they might counter that satisfying (i-iii) is not sufficient for an utterance to count as an assertion. In what follows, I consider a combination of both strategies.

I’ll start with Affirm, the idea that assertions must present their content as something that occurred in the actual world. Dixon rightly points out that authors can successfully communicate that salient passages of their work are true in the actual world (and not only in the fiction)—we already considered examples of this, like the excerpt (3) from Les Miserables.

However, recognising that authors can covey information about the actual world falls short of proving that utterances in fiction can present their content as true in the same way as assertions do. To see this, it can be helpful to compare works of fiction with works that are paradigmatically assertoric, like works of non-fiction. The default in non-fiction is that its assertions actually present their content as true. Lacking reasons to think the opposite, we are entitled to assume that any declarative statement occurring in a work of non-fiction is presented as true. Works of fiction, even when realistic, are more ambiguous. The content of the story is not straightforwardly presented as something that actually happened. Readers can surely recover information about the actual world—but they can only do it by guessing, on the basis of context and genre-conventions, that the author meant to communicate that a particular utterance (or passage, chapter, etc.) is true in the actual world, and not only in the fiction.

It is from content that is presented as true in the fiction, then, that readers of fiction draw inferences about what they are meant to recognise as being true in the actual world. This speaks against the idea that fictions can present their content as true in the same way as assertions do: when readers of fiction learn something about the actual world, they always have to infer it from content that is presented as true in the fiction in the first place. If asserting rather requires presenting content as true without relying on such inferences, it follows that utterances in fiction cannot be asserted.

The Strong Assertionist can therefore suggest that (a) a genuine assertion must unambiguously and straightforwardly be presented as true, and (b) utterances in fiction cannot satisfy (a). If both (a) and (b) hold, utterances in fiction cannot be assertions, and Strong Assertionism is vindicated.

The second criterion, Liability, admits a similar criticism. As we saw in Sect. 5.3, it’s undeniable that authors can appropriately be faulted for the falsities contained in their fictions. But the Strong Assertionist can counter that, so understood, Liability is too cheap a requirement. Non-assertoric speech also renders the speaker liable to criticism in this broad sense. For instance, I can appropriately be faulted for deceptively implying something false without asserting it, as illustrated by example (4 M). If our goal is to capture a property that is distinctive of assertions, Liability is too broad to be a plausible candidate.

Luckily, apt conceptions of assertoric liability are available in the literature. Various authors have suggested that assertions generate distinctive commitmentsFootnote 33—commitments that go beyond what Liability covers. Assertions, some note, guarantee or warrant the truth of the asserted proposition (Carson, 2006, 2010; Moran, 2005, 11). As Peirce puts it, asserting is a bit like signing a contract stating that the asserted proposition is true (MS[R] 454:5). Once you assert a proposition, you are expected to back it up with adequate evidence if challenged, or else retract it (Brandom, 1994; MacFarlane, 2005). While authors disagree about the details, there is wide enough consensus that asserting a proposition generates a distinctive set of responsibilities that goes beyond Liability (being liable to criticism if the proposition turns out to be false).

If we interpret Liability as requiring this set of stronger commitments, it’s easy to argue utterances in fiction cannot be asserted. Authors of fiction, by definition, do not guarantee that any specific utterance contained in their work is true. This is, one might argue, the key difference between a work of fiction and a work of non-fiction. The Strong Assertionist can hence offer another argument: (a) asserting requires overtly committing yourself to the truth of a proposition, and (b) this sort of commitment is unachievable in fiction. If both (a) and (b) hold, Dixon’s argument fails.

What about the third condition, Explicit? Explicit establishes that assertions have to be explicitly articulated, word-by-word, in the text. There is no denying that utterances in fiction can be explicit in this sense. But the Strong Assertionist might once again reply that explicitness is not enough. Arguably, what is distinctive about assertions is that they make direct claims about the actual world, as opposed to indirect claims (like claims that are advanced by means of implicatures, indirect speech acts, or presuppositions). If this is right, utterances in fiction should not only be explicit to count as asserted, but also advance direct claims about the actual world.

Various authors (Alcaraz León, 2016; García-Carpintero, 2016; Marsili, 2023; Ohmann, 1971, 15) have suggested that utterances in fiction cannot be assertions because they cannot be direct in this sense. This point is often established with linguistic tests similar to the one introduced in Sect. 5.2. The underlying idea is that direct claims aren’t typically cancellable. By contrast, we saw that utterances in fiction invariably are: authors can cancel any putative claim about the actual world contained in their fictions with constructions like (M) and (C).

This suggests that utterances in fiction, unlike assertions, cannot advance direct claims about the actual world.Footnote 34 The Strong Assertionist has a third argument available, then. If (a) assertions must be direct, in the sense of not being cancellable with expressions like (M) and (C), and (b) utterances in fiction cannot satisfy (a), then utterances in fiction cannot be asserted.

Recapitulating, Dixon’s argument for Weak Assertionism admits three objections. Each one is motivated by a fairly orthodox understanding of what assertion requires. Crucially, to establish the desired conclusion, it’s enough that one of the three arguments succeeds. This is because all three arguments independently establish that utterances in fiction cannot ever qualify as assertions. If just one of the three counterarguments flies, Strong Assertionism is vindicated.

6 Strong assertionism

The pleasure that [novels] occasionally offer is far too heavily paid for: they undermine the finest characters. They teach us to think ourselves into other men’s places. […] The reader learns to understand every point of view. Willingly he yields himself to the pursuit of other people's goals and loses sight of his own. Novels are so many wedges which the novelist […] inserts into the closed personality of the reader. The better he calculates the size of the wedge and the strength of the resistance, so much the more completely does he crack open the personality of the victim. Novels should be prohibited by the State.

Peter Kien, in Elias Canetti’s Auto da FéSydney argued that the fiction-maker “nothing affirms”, so that he “never lieth”. On a similar note, John L. Austin (1975) suggested that we should distinguish between ‘serious’, assertive discourse (for the truth of which a speaker takes full responsibility) and fictional, ‘etiolated’ discourse (for the truth of which the speaker cannot be held fully responsible). The version of Strong Assertionism I defended incorporates both insights and goes one step further. Although authors of fictions cannot make genuine assertions, they can still indirectly suggest, or otherwise convey, that salient propositions that are presented as “true-in-fiction” are also true in the actual world. Thus, utterances in fiction can mislead, but they can’t be genuine lies.

One potential worry about Strong Assertionism is that it seems to presuppose a sharp distinction between fiction and non-fiction. Clearly, not every work lends itself to a straightforward classification within this dichotomy. Borderline cases abound, ranging from essayistic fictions (non-fiction essays tainted by fictional elements, like Enrique Vila-Matas’s A brief history of portable literature) to fictions that are entirely true (like Natalia Ginzburg’s Family LexiconFootnote 35). Borderline works of this sort have prompted many authors to challenge the idea that fiction and non-fiction are separated by sharp boundaries. Friend (2008, 2011, 2014), for instance, argues that the opposition between fiction and non-fiction is more like a distinction between genres: loose, rather than sharp; dependent on ever-evolving conventions, rather than modally strict; and allowing for some overlap between the two categories.

I am sympathetic to Friend’s view. But Strong Assertionism needs not presuppose that every work clearly sits on one side or another of the fiction/nonfiction distinction. Strong Assertionism is a thesis about utterances in fiction: it says that if a work is classified as a work of fiction, then it cannot contain assertions or lies. When a work is classified as neither fictional nor non-fictional (or unclassifiable as either), Strong Assertionism simply makes no predictions about its status. Borderline cases, then, don’t challenge this view, which is a view about works that are classified as fictions, and takes no stance about how works should be classified.Footnote 36

We are now ready to offer a conclusive answer to our initial questions. First, contrary to what is claimed by Weak Deceptionism and Weak Assertionism, fictions cannot lie—FNL is vindicated. Second, if fictions never lie it’s because lying requires asserting, and fictions never assert, just like the Strong Assertionist claims. Sydney was right. However, this doesn’t yet mean that authors like Plato were wrong in highlighting the deceptive potential of fictional works.

Far from redeeming deceptive fictions, Strong Assertionism highlights their potential to deceive in ways that are sneakier than non-fiction. Works of non-fiction are bound by the expectations of accuracy of assertoric discourse. The same isn’t true of works of fiction. Authors of fiction can easily deflect accusations of mendacity: if pressed, they can always deny that any particular statement included in their work was meant to be read as a claim about the actual world (even if they indeed included it purposefully, in order to deceive). The net result is that although we can learn a lot from fiction, this path to knowledge is often a perilous one to tread (cf. Friend, 2014; Ichino & Currie, 2016; O’Brien, 2017), because fiction authors don’t overtly take responsibility for the truth of what they communicate.

This doesn’t mean that we should endorse the ramblings of Peter Kien, the protagonist of Canetti’s Auto da Fé cited in the opening of this section, who thinks (like a modern Plato) that “novels should be banished by the state”. But Kien is right in pointing out that fiction can deceive in pernicious and subtle ways. Fictional works can sustain racist stereotypes and narratives, and incorrect accounts of historical events (as illustrated in Sect. 3). Crucially, while authors of non-fiction assume full responsibility for the inaccuracies contained in their text, the same isn’t true of authors of fiction, who can take advantage of some degree of deniability to pass on a false message without being held fully accountable for its falsity. This means that fiction can deliver the same epistemic damage of non-fiction, while freeing authors from full accountability for the falsehoods that they convey. Although authors of fiction may protest that they only ever talk about fictional affairs, their works can be an effective and insidious tool for deceiving about the actual world. Caveat lector!