Factual and fictional narrative are generally defined as a pair of opposites. However, there is no consensus as to the rationale of this opposition. Three major competing definitions have been proposed: (a) semantic definition: factual narrative is referential whereas fictional narrative has no reference (at least not in “our” world); (b) syntactic definition: factual narrative and fictional narrative can be distinguished by their logico-linguistic syntax; (c) pragmatic definition: factual narrative advances claims of referential truthfulness whereas fictional narrative advances no such claims. One could add a fourth definition, narratological in nature: in factual narrative author and narrator are the same person whereas in fictional narrative the narrator (who is part of the fictional world) differs from the author (who is part of the world we are living in) (Genette  1993: 78–88). But this fourth definition is better seen as a consequence of the pragmatic definition of fiction.
Poststructuralist philosophers, anthropologists and literary critics have questioned the validity of the fact/fiction distinction as such, sometimes contending, in a Nietzschean vein, that fact itself is a mode of fiction (a fictio in the sense of a ‘making up’). Applied to the domain of narrative, this approach insists on the ‘fictionalizing’ nature of narrative because every narrative constructs a world. But at least in real-life situations, the distinction between factual and fictional narrative seems to be unavoidable, since mistaking a fictional narrative for a factual one (or vice versa) can have dramatic consequences.
One could object to this common-sense assertion that not all societies produce fictional narratives and that often the socially most important narratives, notably myths, cannot be accounted for in terms of the dichotomy between fact and fiction. But even if it may be true that fictional narrative as a socially recognized practice is not an interculturally universal fact, all human communities seem to distinguish between actions and discourses that are meant to be taken “seriously” and others whose status is different: they are recognized as “playful pretense” or as “make-believe.” Furthermore, developmental psychology and comparative ethnology have shown that the distinction between representations having truth claims and ‘make-believe’ representations is crucial in the ontogenetic development of the cognitive structure of the infant psyche and that this phenomenon is transcultural (see Goldman & Emmison 1995; Goldman 1998). Finally, as far as myth is concerned, it is clearly considered a type of factual discourse: people adhere to it as serious discourse referring to something real (this is also the case of the Bible; see Sternberg 1985, 1990). As shown by Veyne ( 1988), the social construction of “truthful discourse” posits an array of “truth programs” linked to various ontological domains (e.g. the profane as distinct from the sacred). Thus “myth” can be “true” (i.e. treated as serious and referring to some reality), even if believing in its truth enters into conflict with what in another ontological domain is accepted as truthful. For example, in myth and its corresponding reality, people can be endowed with powers nobody would imagine them having in everyday life. This does not imply that there is no distinction between fact and fiction, but that what counts as a fact may be relative to a specific “truth program.”
The poststructuralist criticism of the fact/fiction dichotomy has pointed out that every (narrative) representation is a human construction, and more precisely that it is a model projected onto reality. But the fact that discourse in general, and narrative discourse in particular, are constructions does not by itself disqualify ontological realism or the distinction between fact and fiction. To rule out ontological realism, it would be necessary to show independently that the constructive nature of discourse in general or of narrative in particular makes them fictional or at least implies a “fictionalizing” dynamics. This proof has never been delivered, and so the common-sense hypothesis remains the default option.
The relationship between narratology (Meister → Narratology) and theory of fiction long remained non-existent, in part because classical narratology rarely addressed the question of the fact/fiction difference. The theory was intended to be valid for all narratives, although in reality the classical narratologists drew only on fictional texts. The classical models by Genette ( 1980,  1988) and Stanzel (1964,  1984), for example, were general narratologies whose sole input was fictional texts. It was only at a later stage that narratologists explicitly investigated the relationship between narrative techniques and the fictionality/factuality distinction (Genette  1993; Cohn 1999).
It is important, therefore, that the problem of the distinction between factual and fictional narrative be placed in its wider context. First, not every verbal utterance is narrative, nor is every referential utterance narrative. Thus discursive reference cannot be reduced to narrative reference. More generally, reference is not necessarily verbal: it can also be visual (e.g. a photograph makes reference claims without being of a discursive nature). The same holds for fiction. Not every fiction is verbal (paintings can be, and very often are, fictional), and not every fiction, or even every verbal fiction, is narrative: both a painted portrait of a unicorn and a verbal description of a unicorn are fictions without being narrations. Factual narrative is a species of referential representation, just as fictional narrative is a species of non-factual representation. And of course not every verbal utterance without factual content is a fiction: erroneous assertions and plain lies are also utterances without factual content. Indeed, fiction, and its species narrative fiction, are best understood as a specific way of producing and using mental representations and semiotic devices, be they verbal or not. This means that narrative and fiction are intersecting categories and must be studied as such (see Martínez & Scheffel 2003).
The difficulty of getting a clear picture of the distinction between factual and fictional narrative results in part from a long history of shifting uses of the term “fiction.” The sense which is most current today—that of a representation portraying an imaginary/invented universe or world—is not its original nor its historically most prominent domain of reference. In Latin, fictio had at least two different meanings: on the one hand, it referred to the act of modeling something, of giving it a form (as in the art of the sculptor); on the other hand, it designated acts of pretending, supposing, or hypothesizing. Interestingly, the second sense of the Latin term fictio did not put emphasis on the playful dimension of the act of pretending. On the contrary, during most of its long history, “fiction,” stemming from the second sense of the Latin meaning, was used in reference to serious ways of pretending, postulating, or hypothesizing. Hence the term has usually been linked to questions of existence and non-existence, true and false belief, error and lie.
In classical philosophy, “fiction” was often used to designate what we today would call a cognitive illusion (Wolf → Illusion (Aesthetic)). Hume used the term in this sense when he spoke about causality or about a unified self, calling them “fictions” (Hume  1992: Bk I, Pt IV, Sec VI). Now, this type of fiction, as Hume himself explicitly stated, is quite different from fiction in the artistic field. It is part of the definition of a cognitive fiction that it is not experienced as a fiction. An artistic fiction, by contrast, is experienced as a fiction. This means that artistic fictions, contrary to cognitive fictions, should not produce real-world beliefs (even if in fact they sometimes do: fiction has its own pathologies).
The term fiction has also often been used to designate willful acts of deception intended to be misleading or to produce false beliefs. In this sense, deceptive fiction resembles cognitive fiction. But in the case of willful deception, the production of a false belief depends at least partly on the existence of true beliefs entertained by the person engaged in deceiving others: to induce willfully false beliefs, one must hold at least some correct beliefs concerning the state of affairs about which false beliefs are to be produced, for otherwise the result of willful deception will be haphazard. Willful deception (lies and manipulations) is, once again, quite different from artistic fiction, which implies that at some level pretense is experienced as pretense.
In science, the term is sometimes applied to theoretical entities postulated to account for observational regularities which otherwise would be unexplainable. Electrons and other elementary particles have been called “fictions” in this sense. “Fiction,” used this way, does not designate something known to be non-existent, but is rather the hypothetical postulation of an operative entity whose ontological status remains indeterminate. Theoretical fictions are postulated entities whose ontological status remains unclear but which operate in real-world cognitive commitments. Here again, the situation is quite different from fictional entities in the context of artistic fiction: such entities do not operate in real-world commitments. On the other hand, and contrary to theoretical entities, artistic fictional entities are entities which, if they existed, or if their existence were asserted, would have a canonical ontological status—part of the real stuff of reality. So the difference is the following: in the case of theoretical fictions, fictionality is due to the fact that the ontological status (theoretical terms/real entities) of the entities is indeterminate; in the case of artistic fictions, fictionality is due to the fact that the entities are not inferentially linked to real-world existential propositions (although they are of course in general inferentially linked to real-world beliefs and evaluations).
Finally, the term is also used to designate thought experiments. Searle’s “Chinese Room” thought experiment and Putnam’s “Brain in a Vat or Twin Earth” thought experiments are fictions in this sense of the word. Thought experiments are generally counterfactual deductive devices giving rise to valid conclusions which are integrated into the real-world belief system. Superficially, this may seem to be a situation which resembles that of artistic fiction, but in fact, a narrative fiction cannot be a thought experiment in the technical sense. The principal reason why this assimilation is impossible is that the mental experience induced by an artistic fiction and its validation are very different from those of a thought experiment, for the attitude adopted when creating or reading a thought experiment is an attitude of logical discrimination: we have to verify its formal validity, determine whether or not it is conclusive, think about how its relevance could be increased or refuted, etc. Validating (or rejecting) a thought experiment is achieved through technical controversies between specialists who accept it or not, reformulate or modify it using criteria of logical consistency and necessity. An artistic fiction, by contrast, is activated in an immersive way: it is “lived” and stored in the reader’s or spectator’s memory as a universe closed on itself. As far as validating it is concerned, this is also quite different from validating a thought experiment, since one would not say of an artistic fiction that it is conclusive or faulty, but rather that it is successful or unsuccessful in terms of its “effectiveness” as a vector of immersion, its richness as a universe, etc. In other words, its “felicity conditions” are tied primarily to its immersion-inducing effectiveness and to its capacity for producing an aesthetically and hermeneutically satisfying experience of its mimetic and artifactual properties. Admittedly, artistic fictions can be evaluated in terms of the consistency of the fictional universe or in those of their plausibility in relation to supposed real-world situations or in terms of the desirable character or not of their explicit or implicit standards. But all this has nothing to do with validating a thought experiment. To state the difference more bluntly: a thought experiment is an experimental device of a logical nature, a suppositional or counterfactual propositional universe intended to help resolve a philosophical problem; an artistic fiction, by contrast, invites mental or perceptual immersion in an invented universe, engaging the reader or the spectator on an affective level with the persons and events that are depicted or described.
Historically (at least in Western culture), the key concept for analyzing and describing fiction in the sense of artistic and, more specifically, narrative fiction has not been the Latin concept of fictio, but the Greek concept of mimesis. Unfortunately, mimesis, like fictio, is far from being a unified notion. In fact, the first two important discussions of mimesis, in Plato’s Republic (1974: chap. III and X) and a little later in Aristotle’s Poetics, develop two quite divergent conceptions which have structured Western attitudes toward fiction up to this day. Plato’s theory of representation is founded on a strong opposition between imitation of ideas and imitation of appearances (the empirical world): representation of events as such, contrary to rational argument, is an imitation of appearances, which means that it is cut off from truth. He further posits a strong opposition between mimesis and diegesis. Speaking about stories and myths, he distinguishes between: (a) a pure story (haple diegesis), in which the poet speaks in his own name (as in dithyrambs) without pretending to be someone else; (b) a story by mimesis (imitation), in which the poet speaks through his characters (as in tragedy and comedy), meaning that he pretends to be someone else; (c) a mixed form combining the two previous forms (as in epic poetry, where pure narration is mixed with characters’ discourse). Plato’s preference goes to pure narration, for he disapproves of representation by mimesis (in Book X of The Republic, he goes so far as to exclude mimetic artists from the “ideal city”). Mimesis is a simulacrum, an “as if,” and as such it is opposed to truth: mimesis can never be more than a “make-believe” (for the concept of “make-believe,” see Walton 1990).
The concept of mimesis developed by Aristotle in his Poetics diverges from Plato in several important regards. For the fact/fiction problem, only one is of interest: according to Aristotle, mimesis is a specific form of cognition. Mimetic representation is even considered by Aristotle to be superior to history because poetry expresses the general (i.e. the verisimilar or necessary relations between events), while history only expresses the particular (that which has happened): history relates the life of the individual Alcibiades, while poetry is a mimetic rendering of the typical actions that an Alcibiades-like individual would probably or by necessity carry out (1996: chap. 9, 1451b). This means not only that, according to Aristotle, mimesis triggers cognitive powers of a different kind from those of history, but also that these powers are of a higher order than those of factual discourse. Most classical literary theories which assert that fiction possesses its own truth value do so by reactivating some form or another of the Aristotelian distinction between “mere” factual truth representing contingent actualities and a more “general” type of truth, that of verisimilitude or of necessity, representing ontological possibilities.
The Aristotelian conception must be distinguished from “possible worlds” theories of fiction (Pavel 1986; Ryan 1991; Ronen 1994; Doležel 1998, 1999), inspired by the possible worlds logics of Kripke (1963, 1980) or Lewis (1973, 1978). In terms of possible worlds theories, a fictional world is a counterfactual world, but this counterfactual world is as individual as the world we live in: the counterfactual world is not of a superior kind to our actual world (whereas in Aristotle mimetic reference attains a higher order of truth than factual reference), but simply an alternative world. In fact, the real world is also a possible world. According to modal fictionalism, it differs from other possible worlds because it is the only one which is also actual, whereas according to the modal realism defended by Lewis, it differs from other possible worlds (which are as real as “our” world) only by the contingent fact that we happen to live in it. Possible worlds theories of fiction therefore do not claim that fictional truth is more general than factual truth: it is simply true in another world or universe.
The semantic definition of the distinction between factual and fictional narrative is the most classical one. It was defended by Frege in his famous “On Sense and Reference” ( 1960) and by Russell in the no less famous “On Denoting” ( 2005), two seminal papers of 20th-century philosophical theories of reference. It emphasizes the ontological status of represented entities and/or the truth value status of the proposition or the sequence of propositions which assert these entities. The ontological status of entities and the truth value status of propositions are related, since an assertion which states something about an entity that is non-existent is ipso facto referentially void. But it is important to bear in mind, firstly, that some types of fiction assign “fictive” properties and actions to proper names that refer to existing entities. This is the case for example of the subgenre of counterfactual novels which, like counterfactual history (see Ferguson ed. 1997), ascribe fictional actions to historical persons (e.g. Hitler winning World War II). Autofiction can be seen as a special case of such counterfactual fictions. Secondly, historical persons and descriptions of their real historical actions figure prominently in fictional texts, as in historical novels that often contain a fair amount of factual information.
These mixed situations are difficult to integrate into a semantic definition of the fact/fiction distinction (see e.g. Zipfel 2001), since semantic definitions (with the exception of possible worlds semantic definitions: see Doležel 1999) are by necessity “segregationist” (Pavel 1986: 11–7). Counterfactual fictions seem on the face of it easy to manage, at least in terms of possible worlds semantic models. These models being ontologically holistic, it can be said, for example, that a narrative in which Napoleon wins the battle of Waterloo is not an example of outright falsehood, but refers to a possible world in which Napoleon wins the battle of Waterloo. But is it the same Napoleon? The principle of “minimal departure” (Lewis 1973; Ryan 1991) suggests a positive answer, but the holism of the possible worlds approach (each possible world being complete) suggests a negative answer. Whatever the answer, it is difficult to distinguish counterfactual fiction from counterfactual history on these grounds. Other mixed situations are even more difficult to handle. For example, the sentence “Napoleon lost the battle of Waterloo” seems to express a plain simple truth. Does its status change when it is read in a historical novel as compared to when it is read in a biography of, say, Chateaubriand or Stendhal? Does it lose its truth value when it is integrated into a novel? Most advocates of semantic definitions of the fact/fiction dichotomy give a positive answer to this question: the proper name “Napoleon,” when used in a novel, does not refer to the real Napoleon but to some fictional counterpart (e.g. Ryan 1991; Ronen 1994). However, this seems counterintuitive, for in a historical novel it is important for the reader that the proper names referring to historical persons really do refer to the historical persons as he knows them outside of fiction, and not to some fictional homonym of those real persons (see Searle  1979). Counterfactual fictions give rise to an analogous problem: it seems counterintuitive to say that in an autofiction, for example, proper names lose their referential power, since the point of autofiction is precisely the idea that fictional assertions apply to an existing person (the author himself).
This does not amount to saying that semantic criteria are irrelevant, for the idea that there is a semantic difference between fact and fiction certainly is part of our conception of fiction. Thus a narrative in which every sentence is true (referentially) and which nevertheless pretends to be a fiction would not be easily accepted as a fiction. Invented entities and actions are the common stuff of fiction, and for this reason the idea of the non-referential status of the universe portrayed is part of our standard understanding of fictional narrative. Even so, this does not necessarily mean that a semantic definition of fiction is workable.
Syntactic definitions of the distinction between factual and fictional narrative commend themselves by their promise of economy: if it were possible to distinguish factual and fictional narrative on purely syntactic grounds, there would be no need to take a position as far as semantic problems are concerned, be they epistemological or ontological. It would then be possible to arrive at a purely “formal” definition of the two domains.
The best-known theories that seek to define fiction on a syntactic level have been elaborated by Hamburger ( 1973) and Banfield (1982). Both theories define fictional narrative by syntactic traits which, in theory, are excluded from factual narrative. Hamburger famously stated that the domain of what is usually regarded as fiction divides into two radically disjoined fields: “pretense,” which is a simulation of real utterances and defines the status of first-person non-factual narrative; and “fiction proper,” which is a simulation of imaginary universes indexed to perspectively organized mental states and which defines non-factual third-person narrative. In other words, according to Hamburger, in the narrative realm only third-person narrative is fictional, non-factual first-person narrative belonging to another logical field, that of pretended utterances. Hamburger, at least in the first edition of her book ( 1973), contends that, contrary to pretense, fiction is narratorless, a view sharply opposed to mainstream narratology according to which the narrator (not necessarily personified) is a structural element of any narration, be it factual or fictional, first-person or third-person. Banfield, although her theory is formulated in a much more technical way (based on Chomskyan generative grammar), defends a position similar to that of the German critic. She develops a “grammatical definition” (Banfield 1982, 2002) of the genre “novel,” which in fact is a definition of internally focalized heterodiegetic fiction. Among the anomalies defining the novel understood this way, Banfield puts particular emphasis on the specific use of deictics and free indirect discourse. According to her theory, the specific grammar of the novel consists in a double phenomenon: elimination of the first person except in inner direct speech coinciding with the construction of a special third-person pronoun (called “the E-level shifter” by Banfield). This special shifter suspends the “one text / one speaker” rule that governs discourse outside of fiction and which is grounded in the principle that deictics shift referents with each new E (each new speaker). In a novel, a new point of view need not correspond to a new referent of the first person and hence to a new text. This situation is of course impossible in real-life communication, where each point of view is tied to a specific person. Therefore, fictional sentences are “unspeakable.” In fact, Banfield’s “E-level shifter” is functionally equivalent to Hamburger’s floating “narrative function” which can move freely between different “I-origins.”
Hamburger and Banfield have clearly identified linguistic processes which are typical of internally focalized heterodiegetic fiction (Niederhoff → Focalization) and which cannot be easily accounted for in terms of pretense in third-person factual narrative. This is especially true of free indirect discourse and grammatical anomalies of spatial and temporal deictics. All of these phenomena are tied to what Banfield aptly calls a “special” third-person pronoun which is able to shift freely between different Egos. They invite an analysis of fictional narrative in terms of direct simulation of imaginary universes presented perspectively and (on the side of the reader) in terms of immersion (see Ryan 2001: 89–171). The symptoms of fictionality (see Schmid 2010: 21–33) analyzed by Hamburger and Banfield all share the same characteristic: they use a third-person grammatical perspective to present a first-person mental (perceptual, etc.) perspective (Schaeffer 1998: 148–66;  2010: chap. 3.4). On the side of the writer, these deviating practices are in fact the grammatical third-person transcription of the imaginative simulation of “fictive I-origins” (Jannidis → Character). On the side of the reader, they activate an immersive dynamics: the reader “slips into” the characters, experiencing the fictional world as it is seen perspectively by the characters from within or sometimes, as Banfield suggests, from a point of view that remains “empty” (in terms of a specific “I”).
Contra Hamburger and Banfield, however, it is no less true that the majority of heterodiegetic fictions also contain elements that are best described as simulations of factual narrative statements (Schaeffer  2010: cap. 2, 41–108). The textual passages which Banfield calls “pure narration,” and which correspond to Plato’s haple diegesis, are a case in point. Furthermore, if we look at the history of narrative fiction, the systematic use of internal (variable) focalization is fairly recent (as Banfield and Hamburger acknowledge). If we take a broad historical and intercultural outlook, it appears that heterodiegetic fictions without any element of formal mimesis of third-person factual narrative are relatively rare except in some 19th-century fiction and, more frequently, in the 20th-century fiction. So instead of interpreting the symptoms of fictionality in an essentialist way and trying to use them as definitional criteria of fiction, as Hamburger and Banfield do, we should study them in a historical, cultural, and cognitive perspective: why did verbal fiction in the course of its evolution develop devices aimed at neutralizing the enunciative structure of language in favor of a purely “presentational” use? To our best knowledge, the answer to this question has to do with the processes of immersive simulation induced by narrative and maximized by fictional narrative.
Whatever the importance of the insights gained by syntactic definitions of the fact/fiction distinction, as definitions they have severe shortcomings: to accept them, it would be necessary either to exclude first-person narration from the realm of fiction (Hamburger) or to distinguish between a grammar of epic narration and a grammar of the novel (Banfield). More generally, it would be necessary to accept the counterintuitive conclusion that most fictional texts fall short of the definition of fiction. If semantic definitions of fiction are generally too weak (they fail to distinguish between a fiction and a lie), syntactic definitions are generally too strong (many texts must be excluded which common sense considers to be fictional).
The pragmatic definition of fiction is generally linked to the name of Searle, who is certainly its most important proponent, even though the idea of defining fiction pragmatically is much older than Searle. A pragmatic theory of narrative fiction was implicitly defended by Hume. It could be argued, more generally, that wherever and whenever public representations function as fictions, people link them to their pragmatic specificity because it is only by treating representations in this particular way that they become fictional representations (instead of false statements or lies). Even so, Searle’s definition of verbal fiction in terms of pretended speech acts ( 1979: 58–75) is certainly one of the most important and influential contemporary pragmatic analyses of the fact/fiction distinction in the domain of verbal narrative.
Walton, whose contribution to a pragmatics of fiction is as important as Searle’s, objected to the latter’s definition that the notion of a pretended speech act cannot yield a general definition of fiction because it has no application in, among other things, the domain of pictorial depiction: paintings cannot be described in terms of pretended speech acts because pictorial depiction is not a speech act (1990: Part I, 2.6). It could be argued, however, that Searle’s theory operates at two levels: a definition of verbal narrative fiction in terms of pretended speech acts, and a general definition of fiction in terms of intended playful pretense. It has also been objected to Searle that his definition of fiction as intended playful pretense is unable to explain the fact that many texts intended to be factual end up being read as fictions. Walton argues that fictional intention cannot be a defining property of fiction: a fiction is any object which serves as a prop in a game of make-believe, meaning that a fiction is a fiction because it functions as such independently of the question of whether or not somebody intended it to function in that way. Walton is surely right, but Searle’s interest lies primarily in the canonical public status of narrative fiction, and most of the time narrative texts which publicly function as props in a game of make-believe or as playful pretenses are intended to function in this way and, more importantly, have been specifically designed to do so. So if it is true that fictional intention cannot define fiction as a pragmatic stance, it is nevertheless the existence of a shared intention which explains the fact that the emergence of fictional devices has the cultural and technical history it has.
It is important to distinguish the question of the structural function of intentionality from that of the communication of that intentionality. According to Searle, public representations only possess derived intentionality, which implies that mental intentionality is not transparent across minds: it has to be communicated by conventional means, i.e. using verbal or other signals. This is true also for the intention of fictionality: as shown by Koselleck (1979), the intention to create a factual or a fictional text has to be communicated by signals to be effective. These signals are often paratextual, but for the competent reader there also exist many textual “signposts” (Cohn 1990) signaling fictionality or factuality (see Iser 1983: 121–52).
The pragmatic definition of fiction also highlights the difference between narrative fiction qua playful or artistic fiction and the types of fiction which are tied to the question of truth value and belief. Narrative fiction qua artistic fiction is not opposed to truth in the way cognitive illusion, error, and manipulation are opposed to truth, nor is it constrained by real-world truth conditions in the way the suppositional and counterfactual fictions of thought experiments are. As propounded by Searle, it is best characterized by the irrelevance of real-world truth conditions. In the light of this pragmatic definition, what distinguishes fictional narrative from factual narrative is not that the former is referentially void and the latter referentially full. What distinguishes them is the fact that in the case of fictional narrative the question of referentiality is irrelevant, whereas in non-fictional narrative contexts it is important to know whether the narrative propositions are referentially void or not.
Searle has been criticized for excluding the possibility of any syntactical criterion of fictionality (Cohn 1990). In fact, he only claims that syntactical markers of fictionality are neither necessary (a fictional text can be textually indistinguishable from a factual counterpart) nor sufficient (a factual text may use fictional techniques). The same fact was pointed out long ago by Hume: one and the same text may be read both as fiction and non-fiction. The text (in its syntactic and semantic dimensions) remains the same whatever the type of pragmatic attitude, but the use to which it is put will differ according to the pragmatic attitude (see Hume  1992: Bk I, Pt III, Sec VII). So Searle’s thesis is compatible with the fact that fictional texts and factual texts generally differ syntactically.
A more important criticism is that Searle’s pragmatic definition is only negative: it tells us what fiction is not, but not what fiction is. Genette ( 1993: chap. 2), while accepting Searle’s definition of fiction as a series of non-serious utterances, proposed to amend it by distinguishing two levels of illocution: a literal level—the level of the pretended speech acts—concealing a figural or indirect level that transmits a serious speech act (a declaration or a demand) which declares fictionally that such and such an event occurred, or, alternatively, invites the reader to imagine the content transmitted by the pretended speech acts (see Crittenden 1991: 45–52; Zipfel 2001: 185–95).
In conclusion, the pragmatic definition claims that the syntactic status of fiction depends on its formal make-up, its semantic status on its relationship to reality, but that its status as fiction (or not) depends on the way the representations implemented by the text are processed or used. This would imply that the pair fact/fiction is logically heterogeneous. The conditions for satisfying the criteria of factual narrative are semantic: a factual narrative is either true or false. Even if it is willfully false (as is the case if it is a lie), what determines its truth or its untruth is not its (hidden) pragmatic intention, but that which is in fact the case. The conditions for satisfying the criteria of fictional narrative are pragmatic: the truth claims a text would make if it (the same text, from the syntactic point of view) were a factual text (be these claims true or false) must be bracketed out.
In recent years, theories of fiction and narratology have been renewed by cognitive science (Herman → Cognitive Narratology). The notion of simulation and its cognate immersion seem especially fruitful and may well lead to a better understanding of both the distinction between fact and fiction in narrative and their interplay.
Simulation and playful pretense are basic human capacities whose roots are situated in mental simulation, a partly sub-personal process (Dokic & Proust 2002: intro., vii). Could it be that the mental specificity of fictional narrative is to be found in mental simulation? Actually, simulation is a very broad concept which encompasses much more than fiction. Theories of mental simulation were originally developed in order to account for “mind reading,” i.e. the ability to explain and predict the intentional behaviors and reactions of others. The assumption of simulation theories is that the competence of mind reading makes it possible to put oneself imaginatively “into someone else’s shoes.” It is true that mind reading has a strong narrative component, as the “mind reader” immerses himself in scenarios and scripts. But, of course, not every narrative is fictional.
Basically it can be said that if every fiction results from a process of mental simulation, the opposite is not the case, i.e. that every simulation produces a fiction. Mind reading has a strong epistemic component: (a) it simulates the mental states of a really existing person; (b) simulation must reproduce that person’s intentional states in a reliable way, i.e. it is constrained by the necessity of correctly identifying and assessing the real properties of the person whose mental states are being simulated as well as by the context in which that person is found. In the case of fictional simulation, however, the agents and actions are invented in and through the process of simulation. This process is not referentially constrained and cannot be validated or invalidated in a direct way (e.g. by a comparison between behaviors predicted by the simulation and an actually occurring behavior). This means that, contrary to the results of mind reading, the results of a fictional narrative simulation are not directly fed into ongoing real-world interactions. Fictional (narrative) simulation is not only off-line representational activity (as is every simulation), but also a pragmatically encapsulated activity of simulation. Except for pathological cases, the postulated entities of fictional representations are not fed into our belief system concerning the trappings of the real world. Among other things, mental representations triggered by fictional simulation are not fed into real-world feedback loops. This does not mean that make-believe beliefs do not play into the inferential processes concerning real-world situations, but that this “playing into” is pretty much indirect.
Cognitive science also has shown that simulation and immersive processes are not limited to fictional narratives. Every narrative induces varying degrees of immersive experience. As Ryan has convincingly shown, both fictional and non-fictional narrative texts invite readers to imagine a world (2001: 93): this “recreative” imagination (Currie & Ravenscroft 2002) is a process of immersive simulation. Of course, contrary to referentially oriented representing devices, fictional devices are generally (but not always and not necessarily) constructed so as to maximize their immersion-inducing power. Nevertheless, narrative immersion is not limited to fiction.
Another point where simulation theories could be illuminating concerns the ongoing debate in narrative studies as to whether, as is the case in factual narrative, narrative (heterodiegetic) fiction implies the existence of a narrator or not (Margolin → Narrator). What is at stake here is in fact the question of the target domain of narrative immersion: does the reader or spectator immerge into a (fictional) world, or into a narrative act depicting a world? Does narrative fiction induce immersion through mimetic primers feigning descriptive utterances, or simply through a perspectively organized mentally centered and phenomenologically saturated presentation of a universe? As Currie and Ravenscroft (2002) have shown, both options are open, depending on the structure of the text.
Finally, simulation theories may also help to achieve a better understanding of the grammatical deviations or anomalies of internal focalization in heterodiegetic fictional narrative as studied by Hamburger and Banfield. These “deviations” are not the result of conscious stipulations or decisions, but rather they have arisen slowly out of the practice of writing fiction. At the same time, they are not random, but on the contrary structurally coherent and functionally pertinent. It could therefore be hypothesized that they are the result of deep-level linguistic rearrangements due to cognitive-representational pressures stemming from the immersive process of mental simulation. If such were the case, and if these linguistic anomalies were to be read as a co-optation of language by fictional simulation, this would imply that at some deep level the immersion induced by verbal narrative is never only propositional, but also phenomenological and imaginative. The fact that the evolution of third-person fiction has given rise to techniques for neutralizing the enunciative anchoring of sentences could be interpreted as a symptom of the fact that narration as such induces this type of phenomenological immersion. The difference between factual and fictional narrative as far as simulation is concerned could thus be explained by the fact that once narrative is liberated from the epistemic constraints of truth value, the real aim of the immersive process becomes how to maximize it. This in turn would serve to account for the development of the anomalies studied by Hamburger and Banfield.
The interplay of the semantic, syntactic and pragmatic aspects of fictionality need to be further clarified. Historical and comparative studies of the way they co-evolved differently in different historical and cultural contexts are still too rare.
The problem of the inferences we draw from the fictional world to the world in which we live is still very poorly understood, partly because these inferences are difficult to document by non-introspective methodologies.
(a) Studying the “pathologies of fiction”—the different ways fictions can “go wrong”—would shed considerable light on the conditions under which fictions function “normally.” Some psychological studies suggest that these pathologies, operating on a sub-personal level, might be more common than a fiction-friendly attitude would have it.
(b) Comparative work on various fictional “devices”—mental, verbal, visual, “actantial”—is necessary, because fiction is still too often identified with verbal fiction, and verbal fiction with fiction incarnated in a narrative act (oral or written). This is a “reductionist” move which underestimates the importance of theater, i.e. embodied verbal fictions being acted out in front of a public.