Fiction as a defeater

Abstract

This paper argues that no instances of acquiring knowledge from works of literary fiction are instances of the way we ordinarily learn from the testimony of others. The paper argues that the fictional status of a work is a defeater for the justification of beliefs formed on the basis of statements within that work, which must itself be defeated for such beliefs based on fiction to amount to knowledge. This marks a fundamental difference with learning from testimony, since regardless of one's views on testimony and testimonial knowledge, the fact that your belief that p was based on someone's testimony that p is not in and of itself a defeater for your justification for believing that p.

I. Two ways of learning from fiction

We often form beliefs as a result of engaging with works of literary fiction. Undeniably, some of the beliefs we acquire in this way are actually true. Adherents of the so-called literary cognitivism maintain that at least sometimes true beliefs based on literary fiction amount to knowledge.¹

Among literary cognitivists, it is common to distinguish broadly between two ways in which works of literary fiction can sometimes furnish us with knowledge of the actual world. On the one hand, fiction is said to be able to provide knowledge in the way we ordinarily gain knowledge from the testimony of others. On the other hand, fiction is said to be able to provide knowledge by analogy.

Examples of the first of these categories include learning from Hilary Mantel's Wolf Hall that Thomas Cromwell's father's name was Walter or learning from Dostoyevsky's Crime and Punishment that it was customary for common people to drink tea from a glass rather than a cup in nineteenth-century Russia. Examples of the second include life lessons like learning from Dickens's A Christmas Carol that greed leads to unhappiness or learning from Proust's In Search of Lost Time that desire is always mediated through phantasies.

Davies (2016) writes:

One seemingly uncontroversial, if relatively uninteresting, sense in which fictions can inform us about the actual world is by providing knowledge of particular facts. Authors, after all, often embed their fictions in a ‘real setting’ […]. More interesting, but more controversial, is the claim that literary fictions can provide readers with knowledge of general principles that govern the unfolding of events in the real world. (377)

Similarly, Gendler (2000) distinguishes between what she calls narrative as clearinghouse and narrative as factory:

narrative as clearinghouse: I export things from the story that you the story teller have intentionally and consciously imported, adding them to my stock in the way that I add knowledge gained by testimony. […]

narrative as factory: I export things from the story whose truth becomes apparent as a result of thinking about the story itself. These I add to my stock the way I add knowledge gained by modelling. (76)

Others describe the difference slightly differently while the central idea remains the same. Green (2010: 351), for instance, distinguishes between gaining knowledge from cases in which an author will intersperse her novel with observations […] that she puts forth as straightforward assertions and others in which something might happen in a work of fiction that suggests an implicit claim about how things are.

As seen from these passages, to a first approximation, the difference is between learning that p from a work of fiction that includes one or more sentences that communicate that p vs learning that p from a work of fiction as a result of thinking about the story itself, to borrow Gendler's phrase. The latter category of learning from fiction by analogy is often thought to be more worthy of notice.² This is due to the assumption that instances of the other, testimony-like category are just instances of how we ordinarily acquire knowledge from what other people tell us. Consequently, this testimony-like way of learning from fiction is typically considered seemingly uncontroversial and relatively uninteresting, as Davies (loc. cit.) makes explicit.

I want to argue here that no instances of learning from literary fiction are instances of how we ordinarily learn from the testimony of others. The reason is that the fictional status of a work is a defeater for the justification of beliefs acquired on the basis of statements within that work. If you form the belief that p on the basis of a statement in a fictional work, the fact that your source was fictional—or, alternatively, believing that the source was fictional—downgrades your justification for believing that p. Consequently, defeating this default fiction defeater, as I will call it, is a necessary condition for acquiring knowledge from statements within fiction.³

If you come to believe that Cromwell's father's name was Walter on the basis of reading Wolf Hall, the fictional status of the work is a defeater for your justification. Hence, in the absence of further evidence to act as a defeater-defeater, you are not justified in believing that Cromwell's father's name was Walter on the basis of Wolf Hall. Yet, as we will see, very often one does have such evidence to defeat the default fiction defeater, and hence very often one does come to know such things on the basis of fictional works.

That the fictionality of a work acts as a default defeater for the justification of beliefs formed on its basis shows that learning from literary fiction is fundamentally different from learning from testimony. Whatever one's views on testimony and testimonial knowledge, the fact that your belief that p was based on someone's testimony that p is not in and of itself a defeater for your justification for believing that p—nor is believing that your source was someone's testimony. Hence, while works of literary fiction often communicate propositional information in ways that, depending on one's views, can be seen as testimony, acquiring knowledge of such information is unlike acquiring knowledge of what other people tell us in ordinary circumstances.

• I begin by considering the question of whether fictional works can be said to include testimony.

II. Can works of fiction include testimony?

If some instances of learning from works of literary fiction are nothing more than instances of learning from testimony, such works must include testimony. There are two ways of thinking about the nature of testimony in the literature. The first is by far the most widely accepted. This is the view that testimony is a species of assertion. I spell this out simply as follows:⁴

Assertion View of Testimony

A testifies that p only if A asserts that p.

Audi (2011) gives a representative formulation:

To give testimony that p, to attest to it, in my terminology, is—in an assertive as opposed to a sarcastic or theatrical way—to say that p. (507)

The alternative view is that we should accept that testimony can be given in the form of communication beyond full-blooded assertions. In a general form, then, this Broad View of testimony is just the negation of the Assertion View.⁵

Proponents of the Broad View are motivated by the fundamental idea that there is no special category of testimony, but rather we should be focused on communication more generally, and how we can acquire knowledge of what other people communicate to us. Here I follow Lackey's (2008) formulation of the Broad View:

S testifies that p by making an act of communication a if and only if (in part) in virtue of a’s communicable content, (1) S reasonably intends to convey the information that p or (2) a is reasonably taken as conveying the information that p. (35)

While this is a significantly weaker account of testimony, this version of the Broad View is constrained by the inclusion of the clause in virtue of a’s communicative content.

Imagine that a salesperson calls you. Being reliable in detecting accents, from hearing them speak you quickly realize that the person calling is Australian. Given that there is nothing unusual going on, most will agree that you come to know that the caller is Australian. But the salesperson did not testify that they are Australian, and your knowledge that they are Australian is not testimonial. The Broad View of testimony stops short of counting this as a case of testimony, since the information that the caller is Australian was not conveyed in virtue of the content of what the salesperson said. Neither (1) nor (2) is satisfied: the content of their utterance is entirely irrelevant to your belief. Rather, your belief is based on perception: your hearing of their voice.⁶

Still the Broad View allows, for instance, conversational implicatures to be included in the category of testimony.⁷ If you ask me what I think about a colleague's new book and I reply with, I like the cover, I am reasonably taken to convey that I did not like the book, and I reasonably intended to convey that information; both of these rely on the content of my utterance—that I like the cover. (We return to this in Section VI.)

Given the Assertion View of testimony, works of literary fiction can be sources of testimony only if they include assertions. By contrast, at least on Lackey's version of the Broad View, all such works include testimony, since they all include sentences that satisfy both (1) and (2) in her characterisation of testimony. Unquestionably, many, if not all, declarative sentences in works of literary fiction like Wolf Hall, Crime and Punishment, A Christmas Carol, or In Search of Lost Time were reasonably intended to convey some proposition in virtue of their contents (usually that very content itself), and are reasonably taken to do so.

Many philosophers have thought that fictional works often do include assertions. Some, like Searle (1975), Currie (1990), Lamarque & Olsen (1996), and Deutsch (2000), hold that many fictional works are patchworks of pure assertions and fictive statements.⁸ On this view, while fictional works consist mainly of statements with non-assertoric, fictive force, sometimes authors include statements with assertoric force among these fictive statements. Others, like Stock (2011, 2017a, b), hold that, while all statements within fictional works are fictive, some are hybrid statements that are both fictive and assertoric. By contrast, Mahon (2019), Marsili (2022), and Stokke (2023a) have defended the view that there are no assertions within works of fiction.

In other words, most positions in this region of logical space allow that fictional works can include testimony. If one holds the Assertion View of testimony and the Patchwork View of fictional works—the two mainstream positions—fictional works arguably regularly feature instances of testimony. (We do not have to assume that asserting that p is sufficient for testifying that p for this to be a reasonable expectation.) And as we said above, regardless of one's view of fictional works, the Broad View of testimony implies that testimony is abundant in such works.

The table below summarises these options concerning whether fictional works can include testimony:

As seen from this, it is only the combination of the No-Assertion View of fictional works and the Assertion View of testimony that rules out by default that learning from fiction is sometimes just learning from testimony. However, even if one is in this camp, one can still think that some instances of learning from literary fiction are testimony-like in an interesting way. Even if you think that to give testimony requires making an assertion, and that no works of literary fiction include assertions, it might still be the case that fiction can give us knowledge in a way that is closely analogous to how we ordinarily learn from testimony.

However, as I will argue in what follows, regardless of which combination of views on testimony and fictional works you accept, you should nevertheless reject the idea that cases of learning propositions communicated by statements in works of literary fiction are analogous to, let alone instances of, the way we ordinarily learn from the testimony of others.⁹

III. Testimonial justification and defeaters

Both proponents of the Assertion View and the Broad View of testimony typically think that acquiring knowledge from testimony requires the absence of undefeated defeaters.¹⁰ In other words, independently of whether one thinks that testimonial knowledge can only be acquired from assertions, few will deny that the presence of undefeated defeaters for one's testimonial beliefs prevents such beliefs from amounting to knowledge by preventing them from being justified.

As is usual, we can distinguish between propositional and mental defeaters.¹¹ As Moretti & Piazza (2018) summarise this difference, the former is a proposition that would negatively impact on a subject S’s current justification, were S to become aware of its truth, while a mental defeater is a mental state of S that actually negatively impacts on S’s current justification (Moretti and Piazza 2018: 2846).

Along with the distinction between propositional and mental defeaters, it is standard to distinguish between rebutting and undercutting defeaters.¹² A rebutting defeater is a reason of S for believing the negation of P or for believing some proposition Q incompatible with P, while an undercutting defeater is a reason of S that attacks the connection between S’s ground for believing P and P (Moretti and Piazza 2018: 2847).

Here is a simple example. You were taught in high school that Julius Caesar was assassinated in 44 bce, and you still believe that. Yet, a recent New York Times article has reported that new evidence shows that the assassination in fact took place in 43 bce. Those who accept a strong notion of propositional defeat will think that you are no longer justified in believing the traditional view. Others will think that so long as you are unaware of the article, you are still justified. But if you do read it—or perhaps simply by being made aware of its existence—you now have a mental defeater, and your belief does not amount to knowledge, even if the original date was in fact right. Either way, in this case, the article is a rebutting defeater for your belief, since it is a reason for you to believe that Caesar was not assassinated in 44 bce. By contrast, had your high school teacher's competence been brought into question, this would constitute a (propositional or mental) undercutting defeater.

To have a simple and neutral formulation, let us state the requirement of no undefeated defeaters on testimonial knowledge as follows:

No Defeat

B knows that p on the basis of A’s testimony that p only if B has no undefeated defeaters for believing that p.

One might interpret No Defeat in terms of propositional defeaters, or one might interpret the condition in terms of mental defeaters, or one might think that each kind can defeat testimonial justification in different circumstances. At the same time, most will agree that both undercutting and rebutting defeaters can prevent a testimonially based belief from amounting to knowledge, even if true.

No Defeat, or some condition like it, is standardly accepted by both reductionists (Humeans) and anti-reductionists (Reidians) about testimonial knowledge. The disagreement between these two traditions is not over whether you can know something for which you have undefeated defeaters—few would think you can. The disagreement is over whether one needs positive, non-testimonial evidence for testimonially based beliefs. Anti-reductionists think not. Reductionists maintain that beliefs based on testimony require further, positive non-testimonial grounds—in the form of either local evidence bearing directly on the proposition in question, or global evidence supporting the reliability of the testifier, or testimony in general. Yet, both camps will accept No Defeat. (I will comment on this again briefly in Section VI.)

As I go on to argue next, fiction is a defeater for justification.

IV. Fiction as a defeater

Consider the following sentence from the 2009 novel The Children's Book by A. S. Byatt:

• (1) In 1884, the Fabian Society branched out of the Fellowship of the New Life. (Byatt 2009: 37)

I assume without further argument that (1), as it appears in the novel, means the same as what it would mean had the sentence appeared, say, in a work of history or on Wikipedia. I take this assumption to be uncontroversial.¹³ So, since the Fabians actually did break out of the Fellowship of the New Life in 1884, (1) is actually true.

But, moreover, (1) is what Currie (1990) and Stock (2011, 2017b) have called non-accidentally true. By this, they have in mind a certain kind of counterfactual dependence on the facts. As Stock puts it:

had different events occurred, the content of the utterance would have been correspondingly different; [and] if the same events had occurred in otherwise different circumstances, the utterance would still have described them […]. (Stock 2011: 156)

Byatt did not just happen to write (1) and thereby accidentally say something actually true. She was guided by the historical evidence, the latter being a reliable way of gathering information about the past. This is reflected by (1) being non-accidentally true, in Currie's and Stock's sense.

For example, all other things being equal, had the Fabian breakout taken place in 1885 rather than 1884, Byatt would have written 1885 instead of 1884, since in that case, her evidence would have been different. And had the Fabian breakout taken place in 1884, while, say, Queen Victoria was slightly taller than she actually was, Byatt would still have written 1884, since a slightly taller Victoria would not have altered the evidence concerning the Fabian breakout.

For theorists, like Currie and Stock, who argue that fictional works sometimes include assertions, non-accidentally true statements like (1) are prime candidates for assertions within fiction. Correspondingly, I take it that cases like (1) are prime candidates for things one can learn from fiction in the way one usually learns things from the testimony of others. If we can sometimes come to know things from fiction in the same way that we ordinarily learn from testimony—or at least in a way that is sufficiently akin to testimonial knowledge acquisition—statements like (1) are among such things.

Are you justified in believing (1) on the basis of reading that sentence in The Children's Book?¹⁴ One might think the answer is obviously yes. Indeed, I agree that, in many situations, one is justified in believing such things, and one comes to know them. However, suppose you had no evidence to support the truth of (1). You simply read the sentence in the novel. In that case, it is clear that you are not justified in believing it, despite its being non-accidentally true.

This suggests that the very fact that (1) occurs within a work of fiction defeats justification for believing it. At the same time, in most of these cases, this default fiction defeater is itself immediately defeated. For instance, you might know that the Fabian Society actually existed, and so since you know that the novel is a historical novel, and along with other evidence you have, the default fiction defeater for (1) is defeated.

Yet, all of these are situations in which you do have further evidence for the truth of (1). That is, you have additional defeaters for the default fiction defeater. What I am claiming is that if you have no such further evidence, you are not justified in believing (1) on the basis of reading it in the novel. In other words, believing that p solely on the basis of a statement within a fictional work can never be justified.

Consider a slightly different kind of example. When you read the beginning of Wolf Hall, you realize that the father of the main character is called Walter. Concretely, take this paragraph from the opening page of Wolf Hall, the first time we are given the information that the man named Walter is the father of the main character:

• (2) ‘So now get up!’ Walter is roaring down at him, working out where to kick him next. He lifts his head an inch or two, and moves forward, on his belly, trying to do it without exposing his hands, on which Walter enjoys stamping. ‘What are you, an eel?’ his parent asks. (Mantel 2009: 3)

Let us assume that you know that the novel is about Thomas Cromwell. So, you form the belief that Thomas Cromwell's father's name was Walter as a result of reading (2). Again, this is an example of non-accidentally true information. Accordingly, most will think that your belief is justified, and that this is a fact that you can learn from reading Wolf Hall.

Yet, again, the reason is that you have evidence to defeat the default fiction defeater. Suppose all you know is that the book is about Thomas Cromwell, but you have no reason to believe that the novel is realistic, historical, and so on—indeed, you have only the vaguest idea who Cromwell was. In such cases, you are not justified in believing that Cromwell's father's name was Walter on the basis of (2). After all, there are plenty of novels about historical figures that depart wildly from the facts, such as Seth Grahame-Smith's Abraham Lincoln, Vampire Hunter and works like it. Indeed, there are novels in which historical figures are portrayed as having different parents from they actually had. To be sure, one can arguably come to know things about the actual world from such novels, just as one can learn things about Abraham Lincoln from Grahame-Smith's novel. But again, in each case, the reason is that one has grounds that defeat the default fiction defeater.

There are no cases in which you are justified in believing that p on the basis of a statement within a fictional work without having further evidence to support your belief. Consequently, you cannot know that p solely on the basis of a statement within a fictional work.

We can formulate this suggestion schematically as follows:

Fiction as a Defeater

For any a statement S within a fictional work w such that S communicates that p: S’s occurring in w is a defeater for justifiably believing that p on the basis of S.

I use statement within a fictional work to include stand-alone declaratives like (1) but also larger chunks of discourse like (2). It is undeniable that (2) communicates that the father of the main character (he) is called Walter. It is not the task of this paper to give a theory of how this is achieved.¹⁵ I take it to be clear enough that cases like (2) fall under Fiction as a Defeater, alongside simpler examples like (1).

Further, it is worth underlining that we are not merely claiming that if a statement is fictive, this fact defeats justification. Rather, Fiction as a Defeater applies to any statement within fiction regardless of its force. So, even if one thinks that fictional works sometimes include non-fictive, purely assertoric statements—as on the Patchwork View—the fact that such a statement occurs within a fictional work is a defeater for believing its content. Even if you think (1) is a pure assertion appearing amongst the many fictive statements within the novel, the fact that it appeared in this environment is a defeater for believing it.

As with the general No Defeat principle, one can construe Fiction as a Defeater in either propositional or mental terms. Is it enough for a default fiction defeater to be in place that the work on the basis of which you formed a particular belief is fictional, regardless of whether you are aware of this or not? Or is your justification jeopardised only if you believe, perhaps falsely, that the work is fictional? As for other cases of defeaters, there can be arguments on either side.

Whether one is partial to propositional or mental defeaters, default fiction defeaters are clear cases of undercutting defeaters. The fact that your belief that Cromwell's father's name was Walter was based on a fictional work is not a reason to believe its negation. After all, there are many true statements within fiction. Rather, the provenance of your belief undercuts your grounds for believing it, requiring you to marshal further grounds to be justified.

As this suggests, there are many different ways in which a default fiction defeater can itself be defeated. Defeater-defeaters for fiction defeaters can take many different forms. As before, the space of possibilities depends on, among other things, the distinction between propositional and mental defeaters. Suppose you read that p in a work that you initially believe to be fictional, before discovering later on that it was in fact a work of non-fiction. At least crudely, proponents of propositional defeaters will think default fiction defeaters were absent throughout, whereas proponents of mental defeat may argue that an initial, mental default fiction defeater was itself defeated by your subsequent discovery.

Arguably, though, the most common cases of defeaters for default fiction defeaters are simple cases of there being (propositionally or mentally) more evidence to support the relevant belief. Again, suppose you know that the Fabian Society actually existed, and you have some general awareness of the fact that a range of socialist political organisations were emerging in the late-nineteenth century. In this case, it is plausible to think that the default fiction defeater for (1) is defeated. How much is needed—what will count as a defeater-defeater—will depend on the case.

Before moving on, it is worth commenting briefly on Fiction as a Defeater in relation to cases of learning by analogy on the factory model. Suppose you think one can learn that greed leads to unhappiness from reading A Christmas Carol. Suppose you also agree that there is no statement within A Christmas Carol that communicates that greed leads to unhappiness.¹⁶ (One can choose another example if one likes.) Given this, Fiction as a Defeater does not apply. This does not mean that you are not justified—or that you are for that matter. Rather, the principle I have formulated is silent on such cases. Similarly, assuming that there is no statement in the work that communicates this information, the view says nothing about learning from In Search of Lost Time that desire is always mediated through phantasies, and so on for similar cases of learning by analogy.

One might think, not implausibly, that A Christmas Carol as a whole communicates that greed leads to unhappiness. Arguably, it was Dickens's intention to do so, and it was his intention that we should recognise this intention, and so on in the usual Gricean manner. Indeed, even if one accepts a liberal approach on which, for instance, (2) can be seen as a statement that communicates that the protagonist's father's name is Walter, many fictions undeniably convey life lessons, even though these are not communicated by statements within the work. Correspondingly, some, like Voltolini (2021), have suggested that we think of such cases on the model of conversational implicature. Given a view of this kind, clearly the Broad View of testimony is likely to count such cases as straightforward instances of testimony. As we noted earlier—while the view excludes purely perceptual information from the category of testimony—it does allow implicatures.¹⁷

Along such lines, one might think that the fictionality of a work is a defeater even for such cases. Or one might have some reason against such a suggestion—perhaps because one thinks the beliefs in question are formed in a relevantly different way from when learning from particular statements.¹⁸ Yet, I refrain from taking a stand on this here.

What I am arguing is that, in cases where some statement within a fictional work can be said to communicate that p, its appearing within fiction is ipso facto a defeater for believing that p on the basis of this statement. If you come to believe that the Fabians broke out in 1884 from reading (1) in The Children's Book, the fact that you read this in a novel undercuts your justification for believing it in the absence of defeater–defeaters. The same applies to believing that Cromwell's father's name was Walter based on (2).

V. Fabrication, unreliability, and defeat

Why does fiction defeat justification? I suggest that at least the main reason is that many (arguably, most) statements within fiction are fabricated: they are simply made up.¹⁹ Further, as argued by Stokke (2018), if a statement is fabricated, it is at most accidentally true; or, to put it differently, fabricated statements are unreliable even if actually true. To illustrate, consider the following story:²⁰

Slander

Sue hates her boss and wants to give Bob a bad impression of her. So, while Sue in fact believes that her boss would never do anything of this sort, she makes up the story that her boss moved funds illegally, and she tells Bob that. Bob has no reason to think that Sue is lying, and so as a result of her testimony comes to believe that her boss moved funds illegally. As it turns out, the story is true. Sue's boss did move funds illegally.

Clearly, in this case, Bob does not end up knowing that Sue's boss moved funds illegally, even though he does end up with a true belief that she did. But, moreover, few would deny that Bob is justified in believing what Sue tells him, since he has no reason to distrust her.

Perhaps some adherents of propositional defeat will argue that the very fact that Sue lied is a defeater for Bob. Yet, I am merely using this example to illustrate the point that, while merely being told that p by someone one has no reason to distrust is not itself a defeater, fabrication often prevents knowledge in virtue of unreliability. Along the same lines, a reductionist may add to the example that Bob has positive evidence for believing Sue, if needed.

Even though Bob's belief was justified (modify the example as required), it was merely luckily true, and hence does not amount to knowledge.²¹ In other words, the reason Bob's belief falls short of knowledge is that Sue's statement was unreliable, or merely accidentally true. There are nearby scenarios in which Sue's boss did not move funds illegally but where Sue still says that, and there are nearby scenarios in which Sue says that her boss moved funds illegally, even though she did not. Making things up is not a reliable guide to the facts, even though one might happen to get things right.

Fiction is rife with fabricated statements, this being one of the things we enjoy about literary fiction: its free use of the imagination. One obvious consequence of this is that fictional works include many statements that are actually false, and hence a  fortiori cannot be known. Yet, equally for statements within fiction that are true, even non-accidentally so, their occurring within fiction defeats justification. The reason for this is that many (perhaps most) statements in fiction are fabricated. Correspondingly, as suggested earlier, to have grounds for thinking that a particular statement within a fiction is not fabricated—that it is a factual, reliable statement within the work—is eo ipso to have a defeater-defeater. Yet, if one has no such grounds, one is not justified in believing what one reads within a fictional work.

One might be worried that this view overgenerates. For instance, one might ask: if occurring within fiction is a defeater even for cases of non-accidentally true statements in fictional works, why is occurring on the BBC News not a defeater, given that the BBC also shows many fictional programmes?²² After all, just as we said that many statements within fiction are fabricated, it is equally true that many statements on the BBC are fabricated, given its wide range of fictional content. Similarly, one might ask whether our view can accept that one is justified in believing what one reads in a non-fiction book published by Penguin Press, in light of the large number of fictional works published by Penguin.

Fiction as a Defeater distinguishes between particular statements and the work in which they occur. Generalizing slightly, we can distinguish between a particular statement and the source for it. The source for (1) is The Children's Book, the source for (2) is Wolf Hall, and so on. Given this, it is natural to say that, in the BBC case, your source of information is not the BBC tout court but the BBC News. One central reason why responsible media broadcasters mark the beginning and end of their programmes—with animated vignettes, music, credits, and so on—is precisely so that it is clear which programmes make which claims. The distinction is likewise upheld by conventional features: presenters sitting behind a desk-like stage set wearing a certain style of clothes and speaking in a certain tone of voice, reporters holding microphones with signs on them, absence of background music, and so on.

Correspondingly, when coming to believe a statement in a non-fictional work published by Penguin, such as Stephen Hawking's Black Holes and Baby Universes, your source was that work, not Penguin Press. The latter would be your source if you were to believe, say, a statement in a press release by Penguin, or the like. To be sure, this presupposes a way of distinguishing particular works (or sources) from others. This, however, is a general issue involving, at least, one's views on the ontology of literary works, fiction or non-fiction—and correspondingly for non-literary media like TV.²³

In particular, note that we do not need to rely on a specific view of what makes Black Holes and Baby Universes a different work from, say, Iris Murdoch's The Black Prince, also published by Penguin. What we are assuming is just that the former is a different work from the latter. Correspondingly, we should be permitted to assume that the BBC News, or a particular BBC News programme, is a different work (or source) from some BBC Drama series, or one of its episodes. But, moreover, the fact that there arguably can be borderline cases—both concerning the delineation of works and concerning fictionality—presumably means that there are borderline cases of whether one's belief is subject to Fiction as a Defeater. This I take to be the right result.

Default fiction defeaters arise because of the prevalence of fabricated statements within fictional works. The reason there is no analogous defeater for believing the BBC News, or a Penguin non-fiction work, is the scarcity of fabricated statements within such works (or sources). By contrast, sources such as The Onion, or perhaps even some broadcasters posing as serious, non-fictional outlets, do feature a large number of fabricated statements. In some such cases, they may be said to have crossed the line to being fictional works, or sources. Hence, according to my view, default fiction defeaters do apply in such cases.

Yet, in cases such as non-fictional works by Penguin or the BBC News, one's justification is not subject to default fiction defeaters, since one's source is clearly non-fictional. The fact that there are other works published, or produced, by the same organization—or even the same author in many cases—which do feature many fabricated statements, does not threaten your justification in the sense under discussion here. The kind of default defeat we are highlighting arises within, not across, works and sources.

VI. Fiction and testimonial knowledge

The fact that fiction is a default defeater means that learning from literary fiction is fundamentally different from learning from the testimony of others. The reason is that, regardless of what else one thinks about testimony and testimonial knowledge, the fact that someone testified that p is not in itself a defeater for believing that p.

Even the reductionist who requires that we have positive evidence for believing what others tell us does not do so because she thinks that the fact that someone testified that p is a defeater that needs to be defeated. Instead, roughly, reductionists think that you need positive evidence because the fact that someone testified that p is not evidence that p, as on the anti-reductionist view. By contrast, on the view I am arguing for, being communicated by a statement within fiction in and of itself downgrades justification. Yet, the reductionist about testimony does not hold that simply being told something by someone else defeats justification. Such a view is not a reductionist but a sceptical view of testimony.

It is consistent to hold that works of literary fiction often include testimony—perhaps in the form of assertions—while recognizing that fiction acts as a default defeater. One can hold that some of the statements made in the course of producing a fictional work are instances of testimony, and still agree that acquiring knowledge of their contents is unlike ordinary cases of learning from testimony.

Moreover, as we said earlier, there are positions according to which fictional works do not strictly speaking include testimony at all, in particular the conjunction of the Assertion View of testimony and the No-Assertion View of fictional works. Yet, as we noted, even so one might think that we can learn from literary fiction in ways that are nevertheless sufficiently like learning from testimony to make it fruitless to insist on the difference. And, moreover, those who think that fiction often does include assertions, as well as any adherent of the Broad View of testimony, should accept that fictional works often do include testimony.

However, learning from fictional works is fundamentally unlike learning from ordinary, non-fictional testimony, regardless of whether one thinks that fictional works can include testimony or not. Fiction is a default defeater, testimony is not. Hence, we can acquire knowledge from fictional works only if we have evidence that acts as defeater-defeaters for the relevant default fiction defeaters .²⁴

Footnotes

References