The chapter appraises David Lewis’s seminal work on truth in fiction. This will allow us to make an important distinction between three uses of fictive discourse, including the one that Lewis’s work focuses on: discourse characterizing the content of fictions. The chapter examines variations of standard criticisms of Lewis’s account, aiming to show that, if developed as Lewis suggests in his 1983 “Postscript A,” his proposals on the topic are – as Hanley puts it – as good as it gets. Thus elaborated, Lewis’s account can resist these objections, and it offers a better picture of fictional discourse than recent resurrections of other classic works of the 1970s by Kripke, van Inwagen, and Searle. The turn that Lewis suggests, and which the chapter recommends, draws on the remaining outstanding contribution from that time, that of Walton, which is to be examined in Chapter 3.

Imaginings play a crucial role in accounting for fictionality, but what are they? Focusing on those invited by fictions, this chapter argues for the deflationary view that imaginings are just entertainings, I=E. This view was standard in early analytic philosophy, but few current writers appear to hold it. The chapter critically addresses an argument by Walton against I=E that may contribute to explaining this turn; some who espouse views that are otherwise close to I=E endorse this argument against it. In response to Walton’s argument, the chapter invokes a point suggested by Walton himself: Many imaginings – i.e., entertainings, on the view defended here – are mental episodes that agents launch for a purpose. The chapter also appeals to this fact to dispose of a miscellany of other contemporary considerations against I=E. In addition to answering objections, the chapter offers a positive consideration in favor of I=E: to wit, that it may help to establish the imagination as a fundamental, irreducible mental attitude – a view that many philosophers do endorse.

Fictional discourse is, primarily, discourse that is used to produce literary fictions; but there is also the ‘metafictional’ discourse used to talk about fictions, i.e., to report their contents or other features. On a traditional view articulated by Searle, primary fictional discourse doesn’t have any specific semantics; sentences there just have the semantics that they would have in their standard uses. Fiction-makers convey their fictions by pretending to use sentences in their standard ways without doing so, and without giving them a specific, dedicated representational point with a semantics of its own. This Mere Pretense view is less popular nowadays than it used to be. A question the now more popular alternative Dedicated Representation view raises is: What is the contribution of intuitively empty names to such a dedicated semantics? Many supporters of the traditional Mere Pretense view, including Searle, argue that, in the metafictional uses to which they grant a semantics, apparently empty names are not in fact empty; rather, they each refer to some more or less exotic entity. Some of those who grant a dedicated semantics to primary fictional discourse, like Salmon and others, extend this realist view to it. This chapter aims to uphold considerations that have already been raised by irrealists against these proposals, by highlighting their counterintuitive features and explaining in our theoretical setting why they are bad.

Some fictions have explicit narrators, like Marcel in À la recherche du temps perdu, Watson in the Sherlock Holmes stories, or the unnamed first-person teller in Don Quixote. Explicit narrators are less common in fiction films, but there are some – the late Joe Gillis in Wilder’s 1950 film Sunset Boulevard, and Addison DeWitt in Mankiewicz’s 1952 film All About Eve. This chapter addresses the debate on whether there are covert fictional narrators in most or all fictions, which is assumed, for instance, in David Lewis’s account of truth in fiction. The chapter argues that many fictions, in literature, theater, and film, do have covert narrators, although they may well “fade into the background and have little or no significance for criticism or appreciation,” as Kendall Walton put it. Nevertheless, like Walton (and George Wilson), I reject their ubiquity. To that effect, the chapter relies on the constitutive-rules speech-act account of fictionality that was defended in Chapters 2 and 3 to elaborate on two distinctions suggested by Wilson, and to defend on that basis effaced fictional narrators, by developing his ‘silly question’ reply to skeptics’ arguments against covert narrators.

The recreative view of the imagination sees it as a ‘mirror’ of basic mental attitudes: There are imaginative (pretend) variants of beliefs, i-beliefs (ordinarily called imaginings), i-seeings (visualizings), i-desires, i-emotions … the imagination is ‘half of psychic life’, as Meinong put it. The single attitude rival view sees the imagination instead as a sui generis nonderivative mental attitude, with distinctive traits: distinctive functional roles, distinctive norms to which it is beholden, and a distinctive phenomenology. This chapter confronts a recent argument by analogy for i-desires, due to Greg Currie, which is based on an alleged parallel between beliefs and desires. The chapter argues in response that this argument fails because the parallel on which it relies fails to obtain on different influential accounts of desires. The discussion strengthens responses to earlier arguments for i-desires.