Content area
Full Text
Journal of Philosophical Logic (2006) 35: 9Y40DOI: 10.1007/s10992-005-9005-8# Springer 2005DIANE PROUDFOOTPOSSIBLE WORLDS SEMANTICS AND FICTIONReceived on 29 June 2005ABSTRACT. The canonical version of possible worlds semantics for storyprefixes is due to David Lewis. This paper reassesses Lewiss theory and drawsattention to some novel problems for his account.KEY WORDS: contradiction, counterfactuals, fiction, impossible worlds, incompletenarrative, inconsistent narrative, Lewis, modal fictionalism, possible worlds, truth infiction1. INTRODUCTIONThe phenomenon of apparent reference to fictional nonexistents poses aseemingly intractable problem. Contemporary Meinongians have soughtto avoid Russells objections to Meinongs Fimpossible objects_ by introducing an unrestricted Fexistential_ quantifier and non-standardaccounts of negation, and by making distinctions either in the sorts ofproperties attributed to fictional nonexistents or in the modes of predication involved in such attributions (Linsky and Zalta, 1991; Zalta,1995; Parsons, 1995). Other neo-Meinongians have championed the possibility of true contradictions (Routley, 1980). Some modern realists haveconstrued fictional nonexistents not as nonexistent and concrete but asabstract and existent (Salmon, 1998; Schiffer, 1996; Searle, 1979; VanInwagen, 1983; Kripke, 1973). Some modern proponents of the Russellianstrategy have proposed fictionalist theories of fictional discourse (Evans,1982; Walton, 1990). Some employ the Kaplanesque notion of Funfilledpropositions_ (Braun, 1993). Others apply contemporary work in psychology (Currie and Ravenscroft, 1997; Walton, 1997) or Artificial Intelligence(Rapaport, 1991). All these accounts face serious objections.Possible worlds semantics has seemed to many to offer hope with thisdifficult problem. The canonical account, of course, is Lewiss semanticsfor story prefixes. This paper provides a reassessment of Lewiss attractive theory. Possible worlds semantics appears to offer a means of takingapparent reference to the nonexistent seriously while steering a middlecourse between ontological extravagance and counter-intuitive reductionism. The fact that outside philosophical circles we in any case talk aboutthe Fworld_ of a fiction points to a troublesome characteristic of fiction that10DIANE PROUDFOOTpossible worlds semantics is particularly well suited to address, namely thatwhat is true in any given fiction is not limited to what is explicitly stated inthe fiction or to the logical implications of explicit statements.The debate between modal realism and modal fictionalism provides afurther reason to revisit a possible worlds semantics for story prefixes.The modal fictionalist (Rosen, 1990) aims to analyse modal discourse bymeans of a fictional operator, the rationale being to obtain the advantagesof a possible worlds analysis of modal discourse while avoidingontological commitment. (For the modal fictionalist, possibly (necessarily) P iff according to the fiction of a plurality of worlds, at some (every)world P.) Critics of modal fictionalism have weighed the costs andbenefits of a prefixing strategy (e.g., Divers, 1995, 1999; Hale, 1995a;Brock, 1993), but this has been in the absence of any detailed analysis ofthe Faccording to the possible worlds fiction_ prefix.1 The canonicalmodal fictionalist account relies on Fan understanding of how storyprefixes generally work_ (Rosen, 1990, p. 343) and critics typicallyreiterate, without questioning, this claim. This is unfortunate, given theabsence of any agreed-upon semantics for such prefixes. (The samedifficulty arises for those mathematical fictionalists and other semanticfictionalists who use story-prefixing strategies.) In response to criticisms(Hale, 1995a) of informal glossings of his prefix, the modal fictionalisthas declared, FOfficially, the prefix is primitive_ (Rosen, 1995, p. 70) andsome have accepted this declaration (e.g., Nolan, 1997, p. 267). But thiswill not do, since then we have little reason to opt for modal fictionalism rather than to take the logicians standard modal operators asprimitive.2 At worst, such a strategy reduces modal fictionalism to a merepaper semantics, of negligible explanatory value in comparison withmodal realism. Lewis himself suggests as much (Lewis, 1992, p. 222).Can the modal realist turn the tables on the fictionalist? In the absenceof any adequate paraphrase of the fictionalists operator, the modalrealist can argue that it is modal realism which offers the best analysis ofstory prefixes.3 There is reason, moreover, to think that certain criticismsof the fictionalists glossings of his operator could be forestalled byproviding a possible worlds semantics for that operator.4 However, if the