“People Always Think Something’s All True”. How to ascertain fictional truth from unreliable narration
In this paper I argue that none of the most influential theories of truth in fiction can adequately handle the generation of fictional truths via the common literary device of the unreliable narrator. After providing a philosophically useful definition of what an unreliable narrator is, I examine the view of David Lewis and argue that, even with the modification of his theory laid out in his Postscript to ‘Truth in Fiction’, Lewis’ theory is either false or redundant. I argue that Currie’s fictional author fares a little better – his account can arguably handle cases of exaggeration such as Lewis's 'Flash Stockman'. However, in instances of first person unreliable narration I argue that Currie has no principled reason to posit a further reliable and dispassionate fictional author when an existing fictional entity (the unreliable narrator) fulfills all the requirements for that role. I go on to refute Bonomi’s and Zucchi’s claim that Currie’s account can be made to accommodate unreliable narrators by introducing a ‘presumption of reliability’, defeasible when reasons ‘intrinsic to the fiction’ point us to the narrator’s unreliability. Finally, I argue that Walton’s principles of generation also struggle to cope with various types of unreliable narration. I conclude by suggesting that an illocutionary-act account of fiction making, of the type recently suggested by Manuel Garcia-Carpintero, would be uniquely well placed to not only ‘handle’ the generation of fictional truths by unreliable narrators, but also to give full philosophical expression to the notion that sometimes the unreliability of a narrator is the key fictional truth that must be apprehended if a literary work is to be properly understood. I defend this account by showing that it can handle a variety of literary examples (as well as some from film and other media) that cause problems for rival accounts.