Fictional and Intentional Objects

Sherlock Holmes lived at 221B Baker Street. Sherlock Holmes is a fictional character created by Conan Doyle. Both sentences seem true; only the second is true of the actual world, and only the second seems to be about anything.

That asymmetry is the whole problem, and it is a better test case than Pegasus, because fiction is a practice we engage in deliberately, with conventions of its own, and because literary criticism makes claims that appear to quantify over characters in earnest.

The argument from criticism

The strongest case for realism about fictional characters is not from the fiction but from the discourse about it. Consider:

There are three characters in the novel who never meet. Some nineteenth-century characters were modelled on real people. Holmes is more famous than any real detective.

These are made outside any pretence, are straightforwardly true or false, are argued about with evidence, and quantify over characters. By the criterion of ontological commitment, that commits us to characters — unless a paraphrase is available.

The comparison-across-fictions cases are the hardest for the anti-realist: "Holmes is cleverer than Poirot" relates characters from different works by different authors, so no single pretence encompasses it.

Abstract artifacts

The leading realist account (Thomasson, van Inwagen, Salmon) holds that fictional characters are abstract artifacts: abstract objects that are nonetheless created, dependent on the author's activity and on the literary work.

Their profile:

• Created, at the moment of composition. Holmes did not exist in 1800.
• Dependent — rigidly on the author's creative acts, and generically on there being copies, memories, or records to sustain the work.
• Capable of ceasing to exist, if every trace of the work is lost.
• Abstract: not located, no causal powers of their own.

The account explains the asymmetry the page opened with. "Holmes lived in Baker Street" is not true of the artifact — an abstract object lives nowhere — but is true in the story; "Holmes is a fictional character" is true of the artifact simpliciter. The predicates divide into those the character has and those the character is ascribed in the fiction, which is Zalta's encoding/exemplifying distinction in a different key.

The distinctive cost is that this makes fictional characters created abstracta — and, as the propositions page noted about musical works, created abstracta break the standard abstract/concrete criterion, on which abstract objects are eternal and independent. Either the criterion is wrong or these objects are not abstract.

Identity conditions are the other pressure point. When are the Holmes of two stories the same character? When is a character in an adaptation the same as the original? The realist's answer appeals to authorial intention and historical continuity — which are the same materials that individuate artifacts generally, so the difficulty is not special to fiction.

Pretense theories

The main anti-realist alternative (Walton, Everett) denies there are characters at all. Fiction is a game of make-believe: the text is a prop that prescribes imaginings, and to say "Holmes lived in Baker Street" is to make a move within the game, not to describe anything.

The apparently earnest external claims are then handled by extending the pretence. "There are three characters who never meet" is true because within an extended game — one that includes pretending that our talk of characters refers — it is correct to say so. Walton's machinery of prescribed imaginings and unofficial games is designed to make this systematic rather than ad hoc.

Two objections:

• The extended pretence looks like a device to avoid a commitment we have already recognised. Literary criticism does not feel like play, and its claims are supported by evidence and corrected by argument.
• Cross-fiction and cross-medium comparisons require a game encompassing works never intended to be combined, which strains the account.

Everett presses a different line: fictions are frequently inconsistent or incomplete, and a realist ontology of characters must then say what the character is like where the text does not say, or where it says contradictory things. Is Holmes's blood type determinate? The pretense theorist has no problem — the game simply does not prescribe an answer — while the realist must either accept indeterminate objects or supply an arbitrary completion.

The realist's reply is that incompleteness is exactly what one expects of an artifact constituted by a text: the character has only the properties the text ascribes, and questions the text does not settle have no answers because there is nothing to settle them. That is a virtue rather than a defect, since it is what distinguishes a character from a person.

Intentional objects generally

Fiction is the tractable case of a wider problem. Thought is about things, and we can think about what does not exist: the fountain of youth, phlogiston, a planet between Mercury and the Sun. If aboutness is a relation, it needs a relatum.

The options mirror those above: relata that do not exist (Meinong); abstract representations as relata; or intentionality as not genuinely relational — a state's being as of an object without there being an object it is of. The last is now the standard view in philosophy of mind, and its ontological attraction is that it removes the pressure to populate the world with objects of failed reference. Its cost is that aboutness must then be explained without the relational model that made it seem intelligible.

Where fictional objects sit

An inventory question whose interest for the section is that both sides have a fully worked-out theory and the dispute turns entirely on the cost of paraphrase — exactly the situation the commitment page predicted would be typical, and exactly the situation the holes case exhibited in miniature.

The case also supplies a useful corrective. Fictional characters are the stock example of things that obviously do not exist, and yet the serious discourse about them is as disciplined as any. Whatever one concludes, the example shows that "obviously there is no such thing" is not by itself an ontological argument.