«DAVID SANSON Abstract. We want to say both that Sherlock Holmes does not exist, and that he is a ﬁctional character. But how can we say these ...»
Abstract. We want to say both that Sherlock Holmes does not exist, and that
he is a ﬁctional character. But how can we say these things without committing ourselves to the existence of Sherlock Holmes? Here I develop and defend
a non-commital paraphrase of quantiﬁcation over ﬁctional characters, modeled
upon the non-commital paraphrase Kit Fine provides for quantiﬁcation over possibilia. I also develop and defend the view that names for ﬁctional characters are weakly non-referring, in Nathan Salmon’s sense, so and so provide us with a non-commital means to express singular propositions. The resulting position allows us to reap the beneﬁts of Fictional Realism without paying the associated ontological cost.
1. Introduction Many of us want to say that ﬁctional characters don’t exist. We want to say this both in general, as I just did, and in each case: for example, that Sherlock Holmes does not exist, that Harry Potter does not exist, and so on.
Many of us also want to make apparently substantive non-ﬁctional claims about ﬁctional characters, both in general and in particular cases: for example, that some ﬁctional characters are more well-developed than others, that Sherlock Holmes was created by Conan Doyle, and that Sherlock Holmes is a ﬁctional character. But such claims seem to commit us to the existence of ﬁctional characters.
Popular positions on the ontology of ﬁctional characters force us to make a choice.
Fictional Realists choose to give up nonexistence in order to save the truth of the substantive non-ﬁctional claims.1 Pretense Theorists choose to give up on the substantive non-ﬁctional claims, in order to save nonexistence.2 In this paper, I develop a position that allows us to have it both ways, and so satisﬁes what Walton (1990, p. 386) calls “the urge to stand with feet on both sides of the fence” on this issue.
Noneists (or Meinongians) have it both ways, because they make a distinction between being and being something.3 On their view, some things don’t exist, but Date: August 18, 2015. Forthcoming Res Philosophica.
Thanks to Ben Caplan, Joshua Spencer, Wes Cray, David Braun, Heidi Savage, and all the participants in the 2013 Ohio State-Maribor-Rijeka Conference on Art and Reality at the Inter University Centre in Dubrovnik for helpful comments on drafts of this paper.
See van Inwagen (1977); Kripke (2013); Thomasson (1999); Salmon (1998); Braun (2005);
Zalta (1983); Schnieder and Solodkoﬀ (2009) See Walton (1990); Everett (2013).
See Parsons (1980; 2011); Routley (1980); Priest (2005); Berto (2011).
2 DAVID SANSONthere are still substantive truths about those things. In order to say this, Noneists need to reject the simplest deﬁnitions of existence and being, namely,
I am sympathetic to Noneism, but the position I will develop here is not Noneist. I will assume throughout that to exist is to be something, and that ﬁctional characters do not exist in this sense.
The position is drawn from the metaphysics of modality, and modeled closely after Fine’s (1977; 1985; 2003) attempt to provide noncommittal paraphrase for possibilist discourse, as augmented by Salmon’s (1987) attempt to provide a noncommittal account of singular reference to merely possible objects. It is similar in many respects to the “de dicto, quantifying out” approach discussed but not endorsed by Howell (1979, §5.2).
My goal here is to develop the position, and, to the extent possible, defend it. I think the position has some signiﬁcant advantages, and I think more can be said in its defense than one might have thought. It is also my hope that the position will provide a useful foil for those who would reject it, and, perhaps, a useful foil for those who wish to reject the corresponding views in the metaphysics of modality, defended by Fine and Salmon.
2. The Non-Fictional Truth about Fictional Characters Consider the sentence,
1. Sherlock Holmes is a ﬁctional character.
Taken at face value, this expresses a true singular proposition about Sherlock Holmes, and so seems to commit us to the existence of Sherlock Holmes. Or consider,
2. Sherlock Holmes is a detective.
Again, taken at face value, this expresses a singular proposition about Sherlock Holmes, and so seems to commit us to there being such a thing as Holmes. But there is an important diﬀerence between (1) and (2). (1) is a piece of non-ﬁction while (2) is a piece of ﬁction. That is, (1) tells us something about what Holmes is really like, outside of the story, while (2) tells us something about what Holmes is like according to the story.
Any theory of ﬁctional characters will need to make this distinction. A popular strategy among Realists is to distinguish two kinds of predication. For example, van Inwagen (1977, p. 305) draws a distinction between having a property and being ascribed a property: Holmes has but is not ascribed the property of being a ﬁctional character; he is ascribed but does not have the property of being a detective. Zalta
FRIVOLOUS FICTIONS 3(1983) says that characters encode but do not exemplify the properties they have according to the story.4 Let ‘isN F ’ express the relation between a character and the properties it has outside the ﬁction, and ‘isF ’ express the relation between a character and the properties it has according to the ﬁction. Then a sentence like (2) is ambiguous, and could express either,
which is true.
Alternatively, we can introduce to a sentential operator to mark the distinction between ﬁctional and non-ﬁctional truths. Here we have a choice. We can introduce an operator that takes two arguments, a ﬁctional work and a sentence—e.g., ‘According to A Study in Scarlet, Holmes is a detective’. Or we can introduce an operator that takes just a sentence as an argument—e.g., ‘According to a ﬁction, Holmes is a detective’–and worry about how to identify and distinguish diﬀerent ﬁctions later. For simplicity, I adopt the latter approach here, and I write the operator as ‘F’.
So, according to the operator view, (2) is false, but (3) is true:
3. F(Sherlock Holmes is a detective),
I prefer the operator approach because it is more general. For example, it applies to the sentence, ‘It is always raining,’ which is true in Ray Bradbury’s short story, “The Long Rain”, but is not non-ﬁctionally true (thank goodness). And it applies to the sentence, ‘There are exactly ﬁve wizards’, which is true according to the Lord of the Rings. Perhaps there is a way to capture what is going on in these cases by judicious paraphrase and disambiguation of two kinds of predication, but it is not obvious. These examples suggest that the fundamental distinction is a distinction between ﬁctional and non-ﬁctional circumstances, and not just between ﬁctional and non-ﬁctional property ascriptions to the objects that show up in those circumstances.
However we mark the distinction, we are left with two sorts of non-ﬁctionally true sentences about ﬁctional characters. One sort, like (1), describes what a ﬁctional character is like outside the ﬁction. The other, like (3), describes what a ﬁctional character is like inside the ﬁction. Both appear to commit us to the existence ﬁctional characters, for, absent some further story, both appear to express singular propositions about Holmes.
Apparent commitment to ﬁctional characters also shows up in quantiﬁcational claims that make no use of proper names from ﬁction. For example,
4. There are characters in some 19th-century novels who are presented with a greater wealth of physical detail than is any character in any 18th-century novel. (van Inwagen, 1977, p. 302) (4) entails that there are ﬁctional characters in 19th-century novels, and so that there are ﬁctional characters. And, as Kroon (2003) emphasizes, we are happy to make such quantiﬁcational claims and, in the same breath, deny the existence of the characters we appear to be quantifying over, e.g.,
5. There are numerous creatures mentioned in Lord of the Rings that don’t really exist (Kroon, 2003, p. 151).
(5) looks both to commit us to there being ﬁctional creatures, and commit us to their nonexistence. And, of course, we are equally happy to make such claims in the singular case, using proper names, as with,
But it is hard to see how negative existentials of this sort could possibly be true, since they seem to entail a contradiction.
So this is our problem: what should see say about such sentences, and their apparent ontological commitments?
Two popular responses are Fictional Realism and Pretense Theory. According to the Fictional Realist, ﬁctional characters are existing
objects, and when authors tell stories, they are telling stories about these objects. So, on this view, Holmes exists, is an abstract object, and a ﬁctional character. But he (it?) is not a human, nor a detective, nor does he smoke a pipe. Rather, he is all those things according to the ﬁction, but not in reality.
Fictional Realism provides a straightforward account of the truth of (1), (3), and (4), and it entails that (5) and (6) are, strictly speaking, false. So Realists need to say something more complicated to explain why we are inclined to say things like (5) and (6).5 Pretense Theorists, by contrast, argue that none of these sentences are in fact committing, because they are all just part of a game of pretense. Here the key idea is that we can pretend that there is a detective that smokes a pipe without there being some thing that we are pretending to be a detective that smokes a pipe.
Similarly, we might pretend that there is something named ‘Holmes’, and pretend that it is a detective.
Pretense Theory would seem to oﬀer a nice account of sentences like (2), which are not true, but are treated as true (or “authorized”) within the appropriate game of pretense. But the view has trouble with sentences like (1), (3), and (4), which seem, at ﬁrst blush, to express straightforward non-ﬁctional truths about ﬁctional characters. Here Pretense Theorists, following Walton (1990, ch. 10), argue that even these sentences are not literally true, but are instead only true (or “authorized”) within an appropriate pretense. So, for example, Everett (2013, p. 68ﬀ.) argues 5For critical overview of what they might say, see Everett (2013, §7.1-2).
FRIVOLOUS FICTIONS 5that an utterance of (3) counts as true within a “complex extended pretense”, that “extends the domain of make-believe to include further real objects.” Brock (2002) oﬀers a nice suggestion: let the relevant “extended pretense” be the pretense that Fictional Realism is true. Then we can say, of sentences like (1), (3), and (4), that they are false, but they are true according to the ﬁction of Fictional Realism. He calls his view “Fictionalism about Fictional Characters”. So, on Brock’s view, all sentences that appear to commit us to ﬁctional characters are not true, but only true according to the ﬁction of Fictional Realism.
Pretense Theorists run into trouble with sentences like (5) and (6). No doubt we can play a game of pretense in which we pretend that these sentences are true. But the whole point of sentences like (5) and (6), it seems, is to make a claim about what, as a matter of non-ﬁctional fact, does not exist. So, like the Realists, the Pretense Theorists must say something more complicated about what is going on in these cases.6 So, putting aside the hard problem of negative existentials, we appear to face a choice: following the Realists, we can pay the ontological cost, and purchase robust non-ﬁctional truths about ﬁctional characters; or we can, following the Pretense Theorists, refuse the ontological cost, and seek a way of doing without the robust non-ﬁctional truths. But it would be better if we could get what we want without paying for it.
3. Possibilia and Paraphrase
The solution I want to consider aims to provide non-committal paraphrases of everything we want to say about ﬁctional characters, including negative existential claims like (5) and (6).7 The paraphrases make use of two correlated intentional operators, ‘F’ (‘in a ﬁction’) and ‘O’ (‘outside of all ﬁction’), and are closely modeled after paraphrases oﬀered by Fine (1977; 1985; 2003) for quantiﬁcation over possibilia. But ﬁnding non-committal paraphrases for quantiﬁcational claims, like (4) and (5), is not enough. We also need to deal with sentences involving proper names, like (1). Here again we can borrow from the literature on possibilia, and adapt Salmon (1987)’s account of “weakly non-referring” names to the case of ﬁction.
Consider the sentence, There is a possible purple cow: ∃x(x is a possible purple cow) Taken at face value, this commits us to the existence of a possible purple cow. But
we can avoid this by placing the quantiﬁer within the scope of a modal operator:
6For some options, see Walton (1990, § 11.2), Brock (2002, § 4), and Everett (2013, § 3.4).
7There are at least two rather diﬀerent ways of understanding the claim that some sentence, S2, is a non-committal paraphrase of some other sentence, S1. On the ﬁrst, the idea is that S1 and S2 both express the same proposition, but S2 is a better guide to the quantiﬁcational form of that proposition, and so allows us to see that the proposition is non-committal. On the second— which I prefer—the idea is that S1 and S2 express diﬀerent propositions, that the proposition expressed by S1 is committal, but that we can use S1 as a sloppy way of expressing S2, which is non-committal. I don’t think anything I say in this paper hinges on which way you prefer to understand the proposed paraphrases.