Abstract
Fiction tells us about things (or people) that don't exist. How is this possible? Doesn't talking or thinking about them give them a minimal existence? This is the problem of referring to the non-existent. A second problem is that of existential generalization. Since Superman is, as everyone knows, endowed with superpowers and can fly through the air, doesn't it follow that there is an individual who can fly through the air? But what does "there is" mean, given that Superman doesn't exist?
Faced with the problem of existential generalization, logicians have developed several strategies. One is to deny that "there is" implies existence, and thus to reinterpret existential quantification in such a way as to make it independent of any ontological commitment. Another option is to restrict the principle of existential generalization: to validate the inference from the premise " a is F" to the conclusion "there is an x who is F" (interpreted in the traditional way), we now need a second premise establishing the existence of the individual a who is said to be F.
There are other ways of solving the problem of existential generalization, without touching logic. In the philosophy of language, we distinguish between the grammatical form of a statement and its logical form (its true structure, corresponding to the information it conveys). What a statement like "Sherlock Holmes is a private detective who occasionally works for Scotland Yard" really says is that, in Conan Doyle's fictions, Sherlock Holmes is a private detective who occasionally works for Scotland Yard. The prefix "in Conan Doyle's fiction" is left implicit, but it plays an essential role. Because of its presence, all we can conclude by existential generalization is that in Conan Doyle's fictions there is an individual who is a private detective and occasionally works for Scotland Yard.
The distinction between superficial grammatical form and logical form can also be invoked to solve the problem of reference to non-existent people: thus Russell analyzes "Sherlock Holmes" as a definite description in disguise, not as a real proper noun. This allows us to deny that, when we use a fictional name like "Sherlock Holmes", we are really referring to a particular individual to a particular individual, and thus saves the thesis that we can only refer to what exists. Another way of saving this thesis is to argue that, when we talk about Sherlock Holmes or something we know doesn't exist, we're not referring to a particular individual, but pretending - we're pretending that there is such an individual and that we're referring to him.
A third type of solution to the problem of referring to the non-existent is metaphysics. Unlike the first two solutions, it maintains that, when we talk about Sherlock Holmes, we are referring to something. The example of numbers and abstract objects shows that there is room for a certain ontological variety. Similarly, fictional characters do not exist as concrete individuals in reality, but they do have a mode of existence, distinct from the mode of existence that characterizes real objects.
