Talking about what does not exist
A Socratic walk-through of talking about what does not exist — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #How can a sentence about Sherlock Holmes be true when there is nothing for the name to pick out?
"Holmes lived at 221B Baker Street" strikes almost everyone as true, and "Holmes lived in Blackpool" as false. We would correct a stranger who got it wrong. Yet a sentence of that form is supposed to say something about whoever the subject term picks out, and this one picks out nobody, so on the face of it there is no man for the predicate to be true or false of.
The confidence has to be explained rather than dismissed. Either we are systematically mistaken about a judgement we make effortlessly, or truth here is fixed by something other than a man in a flat.
Reasoning it through
REASONING #Before choosing a theory, sort the sentences, because they are not all doing the same job. "Holmes lived at Baker Street" is a claim from inside the stories. "Holmes does not exist" is a claim about whether there is any such man — and it is true, which is awkward, since denying a subject's existence seems to need that subject to talk about. "Holmes is more famous than any real detective" and "Holmes was created by Conan Doyle" are true too, and neither is true in the stories: inside them Holmes exists and nobody wrote him. Three kinds of sentence, and any theory owes us all three.
Now the tempting first move: if the name must pick out something for the sentence to be true, let it pick out a non-existent something. Meinong's theory of objects takes exactly this line — there are objects that do not exist, and Holmes is one, with the properties the stories give him. It handles the internal sentences beautifully and at an obvious price. Russell pressed the price: admit an object for every description and you have admitted the round square, which has contradictory properties.
Russell's own answer, in "On Denoting" (1905), refuses the object. A sentence like "the present King of France is bald" is not a subject-predicate claim at all but a compressed existential one — there is exactly one such king and he is bald — which is simply false, with no ghost required. Treat "Holmes" as a disguised description and "Holmes does not exist" comes out true, cleanly. Notice what it costs, though. On this analysis "Holmes lived at Baker Street" is false as well, and so is "Holmes lived in Blackpool" — the very distinction we started from is flattened.
So try to recover it. What actually settles the Baker Street question? The stories do. That suggests a prefix: read the internal sentences as saying in the Holmes stories, such-and-such — David Lewis's proposal in "Truth in Fiction" (1978, recalled), which sets the standard by asking what would be true if the stories were told as known fact. Baker Street comes out true, Blackpool false, and no object is posited.
Does the prefix do everything? Try it on fame and authorship. "In the stories, Holmes is more famous than any real detective" is not what we meant, and "in the stories, Conan Doyle created Holmes" is flatly false there. The external sentences resist the operator, which is why Kripke, in lectures given in the 1970s and published much later (recalled), argued that fictional characters are abstract artefacts — things genuinely brought into being by an author's storytelling, of which "was created by Conan Doyle" and "is famous" are literally true. The internal sentences then still need the prefix, since an abstract artefact does not live anywhere, so the theory buys the external cases by carrying two devices instead of one.
The opposite strategy drops objects altogether. Kendall Walton's account (recalled from Mimesis as Make-Believe) treats a work as a prop in a game of make-believe: uttering "Holmes lived at Baker Street" is not asserting anything but participating, and it is correct in the game rather than true of a man. That is economical, and it strains where Kripke is comfortable, because a critic comparing Holmes to a real detective seems to be speaking in earnest, not playing.
One more fact any of these must explain. Did Holmes have a mole on his left shoulder? There is no answer — not an unknown answer, no answer. Real people are complete in a way fictional characters are not, and incompleteness is what you would expect if the truths are generated by a finite text plus conventions of extrapolation, and not what you would expect of an object with properties of its own.
The analogy
THE ANALOGY #Think of a chess position written down in notation. "The bishop on c4 is pinned" is true and "the bishop on g8" is false, and you would correct someone who got it wrong — yet there is no wooden bishop anywhere, and never was. What settles it is the record plus the rules everyone shares for reading consequences out of it. Ask whether the bishop is chipped and there is no answer, because the record does not reach that far. And "this position is from a famous game" is a claim about the record, not about any piece.
chess rules fix every consequence mechanically, whereas working out what is true in a story is defeasible and relies on background assumptions readers import without noticing — and a chess record makes no assertion, while a novel at least mimics one.
Clarifying the model
THE MODEL #The most useful correction is the one made at the start: the puzzle looks intractable mainly because three different jobs run through one grammatical form. Once internal, external and existential sentences are separated, each theory can be scored on which it handles without special pleading — and none handles all three with a single device. That is the state of play, not a gap waiting on a better argument.
Two honest qualifications. This is a live dispute, not a settled result: artefact and pretense theories both have serious current defenders, and choosing between them turns partly on how much weight you give to the earnestness of critical talk. And the account offered here leans on fictional truth being fixed by a public work plus shared conventions of extrapolation — which explains incompleteness and explains why we can be corrected, but "shared conventions" is doing real work and is not itself well specified, which is precisely where the disagreement sits.
A picture of it
THE PICTURE #How to readThe placements are schematic rather than measured — read the relative positions, not the numbers. Left to right is how much of an object each account commits to; bottom to top is how comfortably it handles sentences made from outside the fiction, such as fame and authorship. The lower-left pair refuse an object and pay for it there. Meinong's objects sit far right and low: they secure the internal sentences and make "Holmes does not exist" hardest of all. Only the top-right position tries for both, and the cost the picture cannot show is that it needs a second device — the story operator — to say where Holmes lived.
What became clearer
WHAT CLEARED #The confidence about Baker Street is not a mistake, but it is not a report about a man either. What makes it correct is a public text plus shared habits of reading consequences out of it — which is why fictional characters are incomplete, why we can be corrected, and why sentences made from inside the story and sentences made about the story behave so differently. The remaining disagreement is about whether anything at all is being referred to, and it is genuinely unresolved.
Where to go next
ONWARD #- How readers agree on what a story implies but never states, and what governs that extrapolation.
Key terms
TERMS #| Term | What it means |
|---|---|
| Empty name | a name with no bearer, such as "Holmes" or "Vulcan". |
| Negative existential | a sentence denying that something exists, which seems to need the thing in order to deny it. |
| Theory of descriptions | Russell's analysis recasting apparent subject-predicate claims as existential ones. |
| Abstract artefact | a fictional character understood as a real object created by an author's activity. |
Every term the collection defines is gathered in the glossary.