Against a Defense of Fictional Realism

Those are my principles. If you don't like them, I have others.

(Attributed to Groucho Marx in Duck Soup)

I. INTRODUCTION

In The Golden Compass, the first novel in Philip Pullman's His Dark Materials trilogy, Lyra Belacqua, a young girl, travels to Svalbard with Iorek Byrnison, an armored bear.¹ In the sequel, The Subtle Knife, Lyra meets Will Parry, a young boy from another world who is looking for his father, John.² According to fictional realism, the realm of objects includes fictional characters such as Lyra, Iorek, Will, and his father.³ One reason for accepting fictional realism is that it accounts for the truth of some sentences that contain names that seem to refer to, or quantifier expressions that seem to quantify over, fictional characters: for example, ‘Lyra and Iorek are fictional characters’ and ‘Some fictional characters that come from The Golden Compass reappear in The Subtle Knife’.⁴

In ‘Against Fictional Realism’, Anthony Everett argues that fictional realism is false, in part because it entails that some fictional characters are indeterminately identical.⁵ Consider the following (very) short story.

‘A Curious New Shop’: A curious man came into the shop. A curious man left the shop. It's indeterminate whether the curious man who came into the shop is the curious man who left the shop and whether the curious man who left the shop is the curious man who came into the shop. It's even indeterminate whether the curious man who came into the shop is the curious man who came into the shop and whether the curious man who left the shop is the curious man who left the shop.⁶

If fictional realism is true, there is a fictional character that, according to ‘A Curious New Shop’, is the curious man who came into the shop. Let's call that fictional character ‘Fred’. If fictional realism is true, there is a fictional character that, according to ‘A Curious New Shop’, is the curious man who exited the shop. Let's call that fictional character ‘George’. According to ‘A Curious New Shop’, it's indeterminate whether Fred and George are identical. The worry is that this indeterminacy within the fiction will carry over into reality and that it will be indeterminate, outside of the fiction, whether Fred and George are identical.

One response to this worry is to accept that some fictional characters are indeterminately identical.⁷ Another is to reject fictional realism.⁸ And a third is to deny that fictional realism entails that some fictional characters are indeterminately identical. Benjamin Schnieder and Tatjana von Solodkoff pursue this third response in ‘In Defense of Fictional Realism’.⁹ But, we argue in this paper, their view is arbitrary, since there is no reason to prefer their principles to alternative ones.¹⁰ We don't take this to show that fictional realism should be rejected. But we do take this to show that fictional realists who deny that fictional characters can be indeterminately identical need another response to Everett's argument.¹¹

In Sections 2–4, we consider three principles that Schnieder and von Solodkoff endorse, and we introduce three alternative principles that, we argue, are just as plausible. In Sections 5 and 6, we consider and reply to further arguments in favour of two of Schnieder and von Solodkoff's principles. We conclude that the choice between their principles and the alternative principles remains arbitrary.

II. IDENTITY

Schnieder and von Solodkoff adopt the following principle (p. 143).

Identity: For any fictional characters x and y that both come from a fiction T, x = y if and only if, according to T, x = y.

In effect, Identity exports identity claims.¹² In the case of fictional characters that come from the same fiction, if an identity claim is true according to the fiction, then by Identity that claim really is true outside of the fiction; otherwise, we can infer distinctness. Identity gets the right results in uncontroversial cases. (Warning: the rest of this paper contains spoilers.) For example, according to Identity, Dr Stanislaus Grumman and Colonel John Parry are identical, because they come from the same fiction—The Subtle Knife—and, according to that fiction, they’re identical.¹³ By contrast, according to Identity, Lyra is distinct from Iorek, because they come from the same fiction—The Golden Compass—and it's not the case that, according to that fiction, they’re identical.

In the case of ‘A Curious New Shop’, Identity entails that Fred and George are distinct. Fred and George come from ‘A Curious New Shop’, and it's not the case that, according to ‘A Curious New Shop’, Fred and George are identical. (After all, according to ‘A Curious New Shop’, it's indeterminate whether the curious man who came into the shop is the curious man who left the shop.) According to Identity, Fred and George are distinct for the same sort of reason that Lyra and Iorek are: they come from the same fiction, and it's not the case that, according to that fiction, they’re identical.¹⁴

But, instead of Identity, we could adopt the following principle.

Identity*: For any fictional characters x and y that both come from a fiction T, x ≠ y if and only if, according to T, x ≠ y.¹⁵

In effect, Identity* exports distinctness claims. In the case of fictional characters that come from the same fiction, if a distinctness claim is true according to the fiction, then by Identity* that claim really is true outside of the fiction; otherwise, we can infer identity.

Identity and Identity* agree about the uncontroversial cases. According to Identity*, Dr Grumman and Colonel Parry are identical, because they come from the same fiction and it's not the case that, according to that fiction, they’re distinct. By contrast, according to Identity*, Lyra is distinct from Iorek, because they come from the same fiction and, according to that fiction, they’re distinct.

But Identity and Identity* disagree about Fred and George: according to Identity, Fred and George are distinct, whereas, according to Identity*, Fred and George are identical. Fred and George come from ‘A Curious New Shop’, and it's not the case that, according to ‘A Curious New Shop’, Fred and George are distinct. (After all, according to ‘A Curious New Shop’, it's indeterminate whether the curious man who came into the shop is the curious man who left the shop.) According to Identity*, Fred and George are identical for the same sort of reason that Dr Grumman and Colonel Parry are: they come from the same fiction, and it's not the case that, according to that fiction, they’re distinct.

Since Identity and Identity* disagree about Fred and George, they can't both be right. So, if one of them is right, which one is it, and why?

III. GROUNDING

Robert Howell describes a principle like Identity as ‘highly arbitrary and ad hoc’.¹⁶ Schnieder and von Solodkoff disagree. They say that Identity is ‘not just ad hoc’ and that there is in fact ‘some principled basis’ for—‘some principled reasoning’ behind—it (p. 143, italics in original). In particular, Identity is supposed to be supported by two further principles (p. 143). We introduce the second of these in the next section. The first is the following.

Grounding: ‘The nature (and identity) of fictional entities must be grounded in facts about their stories; unless the story provides sufficient grounds for the identity of an entity x and an entity y, no such identity is constituted’ (p. 143, italics in original).

But, instead of Grounding, we could adopt the following principle.

Grounding*: The nature (and distinctness) of fictional entities must be grounded in facts about their stories; unless the story provides sufficient grounds for the distinctness of an entity x and an entity y, no such distinctness is constituted.

And Grounding isn't more plausible than Grounding*.

Grounding begins by noting the need to ground the nature and identity of fictional characters, whereas Grounding* begins by noting the need to ground the nature and distinctness of fictional characters. For example, according to Grounding, we need to ground the identity of Dr Grumman and Colonel Parry, whereas, according to Grounding*, we need to ground the distinctness of Lyra and Iorek. But the need to ground the identity of Dr Grumman and Colonel Parry isn't any greater than the need to ground the distinctness of Lyra and Iorek. Grounding says that, in the absence of sufficient grounding in the story, fictional characters aren't identical, so we have ungrounded distinctness (for example, of Lyra and Iorek), whereas Grounding* says that, in the absence of sufficient grounding in the story, fictional characters aren't distinct, so we have ungrounded identity (for example, of Dr Grumman and Colonel Parry). But there seems to be no reason to prefer ungrounded distinctness to ungrounded identity.

IV. INTERPRETATION

The second principle that is supposed to support Identity is the following.

Interpretation: ‘Since stories seldom explicitly state the non-identity of an entity x and an independently mentioned entity y, their non-identity is the (warranted but defeasible) default assumption in interpreting a story’ (p. 143, italics in original).

But, instead of Interpretation, we could adopt the following principle.

Interpretation*: Since stories seldom explicitly state the identity of an entity x and an independently mentioned entity y, their identity is the (warranted but defeasible) default assumption in interpreting a story.

We admit that, at first blush, Interpretation might seem more plausible than Interpretation*. But we think that, in the end, Interpretation isn't more plausible than Interpretation*.

Interpretation begins by noting that stories seldom explicitly state distinctness. So, for example, when Lyra meets Iorek in The Golden Compass, Pullman writes, ‘This strange hulking presence gnawing its meat was like nothing she had ever imagined, and she felt a profound admiration and pity for the lonely creature’.¹⁷ The text does not explicitly state that ‘the lonely creature’ is distinct from Lyra.¹⁸ By contrast, Interpretation* begins by noting that stories seldom explicitly state identity. So, for example, when Lyra meets Iorek in The Golden Compass, the text doesn't explicitly state that ‘the lonely creature’ is distinct from Lyra, but it doesn't explicitly state that ‘the lonely creature’ is identical with Iorek (or the ‘strange hulking presence’) either. Or consider page 498 of The Subtle Knife. At the top of the page we find this passage.

Ruta Skadi walked a little way upstream, and Serafina Pekkala went with her.

‘Ah, Serafina Pekkala, you should see Lord Asriel’, said the Latvian queen quietly.

The text does not explicitly state that ‘the Latvian queen’ is identical with Ruta Skadi. Towards the bottom of the page we find this passage.

The two queens sat quietly by the stream. Time went past; stars set, and other stars rose; a little cry came from the sleepers, but it was only Lyra dreaming. The witches heard the rumbling of a storm, and they saw the lightning play over the sea and the foothills, but it was a long way off.

The text doesn't explicitly state that ‘the two queens’ are identical with Ruta Skadi and Serafina Pekkala; nor does it explicitly state that ‘the sleepers’ are identical with Lyra and Will; nor does it explicitly state that ‘the witches’ are identical with the two queens.

Interpretation is supposed to support Identity, which allows us to draw conclusions about the identity of fictional characters outside of the fiction rather than about the identity of fictional characters within the fiction. So, if it is to support Identity, Interpretation should be read, not as a claim about a default exegetical assumption about what's true in the fiction, but rather as a claim about a default metaphysical assumption about what's true outside of the fiction. On this reading, Interpretation says that the distinctness of fictional characters is the default metaphysical assumption. On a parallel reading, Interpretation* says that the identity of fictional characters is the default metaphysical assumption.

Sometimes Interpretation seems more plausible than Interpretation*. For example, it seems plausible that the default metaphysical assumption is that Lyra is distinct from the fictional character that, according to The Golden Compass, is ‘the lonely creature’; it doesn't seem plausible that the default metaphysical assumption is that Lyra is identical with the fictional character that, according to The Golden Compass, is ‘the lonely creature’.

But sometimes Interpretation* seems more plausible than Interpretation. For example, it seems plausible that the default metaphysical assumption is that Iorek is identical with the fictional character that, according to The Golden Compass, is ‘the lonely creature’; it doesn't seem plausible that the default metaphysical assumption is that Iorek is distinct from the fictional character that, according to The Golden Compass, is ‘the lonely creature’. Similarly, it seems plausible that the default metaphysical assumption is that Lyra and Will are identical with the fictional characters that, according to The Subtle Knife, are ‘the sleepers’; it doesn't seem plausible that the default metaphysical assumption is that Lyra and Will are distinct from the fictional characters that, according to The Subtle Knife, are ‘the sleepers’.

We’re not arguing that Interpretation* is true. We’re not even arguing that it's plausible. Indeed, we think it's almost certainly false. But then we think that Interpretation is almost certainly false, too. Rather, what we’re arguing is that Interpretation* is ultimately no worse off than Interpretation is, just as Grounding* is no worse off than Grounding is.

V. ARGUMENTS FOR INTERPRETATION

V.I Cases

Schnieder and von Solodkoff might appeal to cases to support Interpretation. For example, they might concede that some cases support Interpretation* rather than Interpretation but insist that most cases support Interpretation.¹⁹ But figuring out whether Interpretation or Interpretation* is better supported by the textual evidence turns out to be complicated. As Schnieder and von Solodkoff state it, Interpretation is a claim about fictional characters that are ‘independently mentioned’ in a fiction (p. 143). To figure out whether Interpretation or Interpretation* is better supported by the textual evidence, we'd need a sense of when one or more fictional characters are independently mentioned (and when they’re not). But it can get complicated quickly, and Schnieder and von Solodkoff don't say anything further to guide us here.

To bring one issue to the fore, consider the opening and closing sentences of The Golden Compass. It might seem that one fictional character is independently mentioned twice, once by each of the occurrences of the name in boldface in the following sentences.

Lyra and her dæmon moved through the darkening hall, taking care to keep to one side, out of sight of the kitchen.²⁰

So Lyra and her dæmon turned away from the world they were born in, and looked toward the sun, and walked into the sky.²¹

But the default metaphysical assumption isn't that the fictional character that, according to The Golden Compass, ‘moved through the darkening hall’ is distinct from the fictional character that, according to The Golden Compass, ‘walked into the sky’. So, if Interpretation is true, then using ‘Lyra’ twice in those passages must not count as independently mentioning one or more fictional characters.

Schnieder and von Solodkoff might say that using the same name twice doesn't count as independently mentioning a fictional character. But there are stories in which more than one character has the same name, and using that name twice should sometimes count as independently mentioning two fictional characters. For example, in Daniel Handler's Adverbs, the narrator addresses the reader:

there's no sense in keeping track of what everyone is doing … you’re likely to confuse them, as so many people in this book have the same names. You can't follow all the Joes, or all the Davids and Andreas. You can't follow Adam or Allison or Keith, up to Seattle or down to San Francisco or across—three thousand miles, as the bird flies—to New York City, and anyway they don't matter.²²

Drawing on Kaplan's distinction between generic and common currency names, Schnieder and von Solodkoff might say that, although there is only one generic name ‘Joe’, in this case there are really two common currency names both spelled ‘Joe’, and using the same common currency name twice doesn't count as independently mentioning a fictional character.²³ But then we would need a way of telling when we have two occurrences of the same common currency name and when we merely have two occurrences of the same generic name, and that way of telling had better not rely on our prior sense of when it's the same fictional character (and when it's not).²⁴

But suppose that, these complications aside, most cases in fact support Interpretation. What then? Interpretation might be closer to the truth than Interpretation* is; but Interpretation still wouldn't be true. The point here is not that Interpretation is a claim about what happens as a default and that default is sometimes defeated; rather, the point is that Interpretation is sometimes wrong about what the default is. To return to an earlier example, we don't think that Iorek and the fictional character that, according to The Golden Compass, is ‘the lonely creature’ are distinct; and we don't think that their distinctness is a default metaphysical assumption that subsequently gets defeated either.

A related justification for Interpretation is that certain special cases—namely, postmodern fictions—support it, in the sense that it's required to make sense of them.²⁵ (Interpretation would then be the rule that certain postmodern exceptions prove, as it were.) We don't think ‘A Curious New Shop’ is one of those postmodern fictions, but Everett's ‘Frackworld’ might be.²⁶ But, if Interpretation is justified because certain postmodern fictions exploit default assumptions about the distinctness of fictional characters, then Interpretation* would be equally justified because other postmodern fictions exploit default assumptions about the identity of fictional characters. Perhaps Handler's Adverbs, mentioned above, is such a fiction.²⁷ So, in the end, we think that appeals to cases don't support Interpretation rather than Interpretation*.

V.II Conversations

Alternatively, Schnieder and von Solodkoff might appeal to features of conversations to support Interpretation. For example, one might think that H.P. Grice's conversational maxims support Interpretation rather than Interpretation*.²⁸ The maxims are, roughly, (1) be just as informative as necessary, (2) make your contribution true, (3) be relevant, and (4) avoid obscurity and be brief. Suppose someone introduces two referring expressions into a conversation. And suppose that the two expressions are in fact coreferential, although this is not made evident. It seems that this would violate several maxims. The relevant utterances would be less informative than necessary and could even be obscure. (If one of the expressions is particularly complicated, the relevant utterances might also be less brief than they could have been.) If the listener is entitled to assume that the speaker is following the maxims, then the listener is entitled to assume that the expressions are not coreferential, which, it might seem, is precisely the assumption that Interpretation captures.

But there are two problems with this line of reasoning. First, it is not clear that, in the case of fiction, the reader is entitled to assume that the author is following the maxims. To begin with, the reader shouldn't assume that the author is telling the truth.²⁹ Nor is the reader entitled to assume that the author is being just as informative as necessary. Some authors are more informative than necessary, as anyone who has read G.R.R. Martin's descriptions of banquets and the banners of minor houses in attendance might come to suspect. And of course the whole point of a mystery is to withhold certain key information for as long as possible.³⁰ The reader should also not assume that the author is being relevant. An author might introduce certain loose ends and never tie them up, and red herrings are common in mysteries. And the reader should probably not assume that the author is invariably going to be brief or lacking in obscurity either. (Pick your favourite literary examples here.) So, even if a listener is entitled to assume that two referring expressions encountered in conversation are not coreferential, it doesn't follow that a reader is entitled to assume the same about two referring expressions encountered in fiction. (For example, we don't think that the passages from Pullman quoted in Section 4, in which two referring expressions are coreferential, violate any maxims.)

Second, even if a reader is entitled to assume that two referring expressions encountered in fiction are not coreferential, it is not clear that this would support Interpretation on the required reading. Recall that, if it is to support Identity, Interpretation should be read as a metaphysical claim about what's true outside of the fiction rather than as an exegetical or semantic claim about what's true in the fiction. And the assumption that two referring expressions encountered in fiction are not coreferential might be more relevant to the exegetical or semantic claim than to the metaphysical one. It's just not clear that a metaphysical claim about the identity and distinctness of fictional characters is the sort of claim that would fall out of norms governing conversations.

One might think that other features of conversations support Interpretation. For example, perhaps Interpretation captures some facts about the psychological processing of discourse.³¹ But what we said about appealing to Gricean maxims applies equally well to appealing to principles of discourse processing: it is not clear that such appeals would support Interpretation on the required reading, on which it is a metaphysical claim about what's true outside of the fiction rather than as an exegetical or semantic claim about what's true in the fiction.

But suppose we’re wrong about all of this; suppose that Interpretation is in fact true. What then? It doesn't follow that Identity is true. Interpretation says what happens as a default. But Identity says what happens in all cases. And this mismatch creates a potential problem. According to Interpretation, the warranted but defeasible default assumption is that Fred and George, for example, are distinct. But, since this assumption is defeasible, it can be defeated; and, if it's defeated even in cases in which the story doesn't explicitly state the identity, then—contrary to Identity—we can't conclude that Fred and George are distinct simply because ‘A Curious New Shop’ doesn't explicitly state their identity. And, one might think, the assumption is defeated precisely in cases, like ‘A Curious New Shop’, in which something funny about identity is true in the fiction, even though the identity isn't explicitly stated. So, we think, the choice between Identity and Identity* is still arbitrary.

VI. ARGUMENTS FOR IDENTITY

On the one hand, we have Identity, which entails that Fred and George are distinct. On the other hand, we have Identity*, which entails that Fred and George are identical. We’ve been arguing that there's no reason to prefer either principle to the other. But perhaps Fred and George really are distinct; and, if so, then we would have a reason to prefer Identity: it would get the right result.

We can think of three quick arguments for the distinctness of Fred and George. First, perhaps authorial intentions settle the matter.³² Perhaps the author intended Fred and George to be distinct; and perhaps as a result they are in fact distinct.

This sort of reasoning might work in other cases. For example, perhaps Lyra and Iorek are distinct because Pullman intended them to be distinct.³³ But this sort of reasoning doesn't work in the case of ‘A Curious New Shop’. We are the authors of ‘A Curious New Shop’, and we didn't intend Fred and George to be distinct. We didn't intend Fred and George to be identical either. We might have intended Fred and George to be indeterminately identical, or we might have avoided forming intentions about their identity or distinctness. (We’re not saying!)

Second, perhaps the number of referential devices used settles the matter. Perhaps the fact that we used both ‘the curious man who came into the shop’ and ‘the curious man who left the shop’ in ‘A Curious New Shop’ is enough for us to have created two fictional characters, whether we intended to or not. But the sorts of cases discussed in Section 4—for example, in which Pullman used both ‘Iorek’ and ‘the lonely creature’, or ‘Ruta Skadi’ and ‘the Latvian queen’—show that using two referential devices isn't always enough for the author to have created two fictional characters. Perhaps it is enough in the right circumstances; but then the question becomes whether we were in such circumstances when we wrote ‘A Curious New Shop’. Without knowing what the right circumstances are, it's hard to know whether we were.³⁴

Third, perhaps Fred and George have different properties. Perhaps, according to ‘A Curious New Shop’, Fred is identical with Fred. So perhaps Fred has the property being an x such that, according to ‘A Curious New Shop’, x = Fred. But it isn't the case that, according to ‘A Curious New Shop’, George is identical with Fred. So George doesn't have the property being an x such that, according to ‘A Curious New Shop’, x = Fred. So Fred and George are distinct.³⁵

But this argument can be resisted. It needn't be the case that, according to ‘A Curious New Shop’, Fred is identical with Fred.³⁶ After all, Fred is the fictional character that, according to ‘A Curious New Shop’, is the curious man who came into the shop; and, according to ‘A Curious New Shop’, it is indeterminate whether the curious man who came into the shop is the curious man who came into the shop. In that case, Fred wouldn't have the property being an x such that, according to ‘A Curious New Shop’, x = Fred.

And there's a deeper problem with the argument, one that can't be circumvented by insisting that, according to ‘A Curious New Shop’, Fred = Fred.³⁷ The argument assumes that Fred has, but George lacks, the property being an x such that, according to ‘A Curious New Shop’, x = Fred. That might be true if Fred and George are distinct, but it might not be true if they’re identical. If Fred and George are identical, then the property being an x such that, according to ‘A Curious New Shop’, x = Fred just is the property being an x such that, according to ‘A Curious New Shop’, x = George. And, if it is the case that, according to ‘A Curious New Shop’, George = George (as it would be if, according to ‘A Curious New Shop’, Fred = Fred), then George has the property being an x such that, according to ‘A Curious New Shop’, x = George, in which case George has the property being an x such that, according to ‘A Curious New Shop’, x = Fred after all. The claim that Fred has, but George lacks, that property thus presupposes that they’re distinct and hence can't be used to provide a justification for their distinctness, just as the claim that Hesperus has, but Phosphorus lacks, the property being an x such that Sam said that x = Hesperus (or being an x such that, according to Sam, x = Hesperus) can't be used to provide a justification for the claim that they’re distinct either.

So we don't think that any of these arguments gives us reason to prefer Identity to Identity*. The choice remains arbitrary.

For comments and discussion, thanks to students at Ohio State; to participants at talks at Copenhagen, the CSPA, Dubrovnik, Kenyon, NTNU, Oslo, SPAWN, Stockholm, Wayne State, and the WCPA; and to Elizabeth Barnes, Stuart Brock, Scott Brown, Einar Duenger Bøhn, Don Caplan, Sam Cowling, Wesley Cray, Gregory Currie, Eva Della Lana, Kit Fine, Cody Gilmore, Olav Gjelsvik, Lars Bo Gunderson, Bob Howell, Carrie Ichikawa Jenkins, Tim Kenyon, Greg Lavers, Hans Lottenbach, Michael McKinsey, Brendan Murday, Larry Powers, Tor Sandqvist, David Sanson, Amie Thomasson, and two anonymous referees.