Walton, Truth in Fiction, and Video Games: A Rejoinder to Willis: Discussion

In her thought‐provoking article, Willis (2019) argues that video games differ from other media—with regard to fictional truth—in ways which cannot be accounted for by Kendall Walton's theories on truth in fiction. To prove this she uses two examples with which she attempts to show, first, that playthroughs of the same game can differ in terms of fictional truth, and, second, that there can be fictional truths that obtain in all playthroughs of a given game, even when not represented. However, both of her examples are problematic: not only can both be accounted for by Walton's model of fictional truth, Willis's own proposed model introduces issues absent from Walton's model.

Willis (44) begins by asserting that video games are unique since a video game's fictional truth can differ in each playthrough. She rejects the possibility that this is a purely interpretative matter, since video games in their capacity of interactive media react to player input by representing different fictional actions. As an example, she mentions how the playable main character Commander Shepard in Mass Effect can be either male or female, and it is up to the player to decide. Willis compares video games with theater, where a performance may include truths absent in the script, such as the height of one character in relation to another. With this in mind, Willis (45) draws the conclusion that there are two kinds of fictional truths in theater, play‐truths and performance‐truths, of which the play contains only the former whereas a performance contains both. She applies this taxonomy to video games, and says that there are truths that are fictionally true in a game and in a playthrough of the game, “video game‐truths,” and those that are true only in the playthrough, “playthrough‐truths” (46).

It should be noted that Willis is not the first to draw a parallel between video games and Walton's conception of fiction; while this does not detract from the value of her arguments, it would have been interesting to see how she approaches claims made by others on this topic.¹

Willis's distinction between different kinds of fictional truths may be unnecessary, since Walton already provides us with at least two ways of accounting for Commander Shepard's gender. First of all, according to Walton (1990, 64–66), a fiction may contain separate prescriptions to imagine a fictional proposition and its negation without prescribing that the interpreter imagines their conjunction. That is, one and the same work may prescribe imagining both p and not‐p, but not the conjunction (p and ∼p). Hence, it poses no problem that Mass Effect states both that Shepard is male and that Shepard is female insofar as the conjunction is never stated. Such paradoxes are by no means unique to games, and require no novel solution in terms of creating a separate set of truths unique to each playthrough.

The second solution would be to appeal to fictional indeterminacy, which Walton explains by referring to differences between various stagings of a given play. Walton (181) argues that if actors playing ancient personages wear modern clothes, we would not take it to be fictionally true that the clothes worn by the characters for this reason are modern too. Instead, he suggests that there is no specific fictional truth about what the characters wear, it is a case of “fictional incompleteness,” it is indeterminate what they wear. Walton (66) also mentions a literary example: in Metamorphosis, nothing is said about the occupation of Gregor Samsa's great‐grandfather, so it is not fictionally true that the great‐grandfather had such‐and‐such occupation, nor is it fictional that he did not have it. The concept of fictional incompleteness has been adopted by several other philosophers (Currie 1990; Lamarque and Haugom Olsen 1994; Phillips 1999; Kania 2005; Stock 2017; Friend 2017; Badura and Berto 2018) as well as some game scholars (Wesp 2014; Ryan 2015). From this it follows that it is not necessarily fictionally true even in a performance that the characters have the same height in relation to one another as the actors playing them. A better approach would be that of Walton, to say that the fiction is incomplete or indeterminate on this matter. Thus, the second solution is that Commander Shepard's gender is indeterminate: Shepard is either male or female, but it is not fictional that Shepard is male, nor that Shepard is female.

Willis's second example is taken from Dragon Age: Origins. She (46) explains that the character Alistair has a sister, but this fact is so peripheral to the plot that one can complete the game without finding it out, so Willis did not discover it in her first playthrough. She says that if video game‐truths must be portrayed in all playthroughs—meaning that Alistair would not have a sister in Willis's first playthrough—it would “severely limit the amount of information which can be counted as video game‐truths” (47). Representation of a fictional fact, Willis (46) argues, is therefore not necessary for it to be fictionally true. Further, she also argues that representation is insufficient: it is not fictional in the play that Laertes is shorter than Hamlet only because this is true in all performances. However, if we assume that representation in a playthrough is both unnecessary and insufficient for a playthrough‐truth to count as a video game‐truth, it raises the question of what purpose playthrough‐truths serve, and what relation these two concepts have to one another. Willis has perhaps created a dilemma. The first horn is that playthrough‐truths are somehow related to video game truths, but that raises several questions: how can a given proposition be a video game‐truth if it is not necessary that it is ever conveyed in an actual playthrough? Conversely, why is a given proposition not necessarily a video game‐truth in spite of its appearance in all actual playthroughs? The relation between playthrough‐truths and video game‐truths seems arbitrary. All actual playthroughs of a given game could potentially contain no video game‐truths whatsoever, which raises the question of how the latter are ascertained. This would in turn undermine Willis's claim that a game has to be played several times to find all video game‐truths (the fact that nothing represented in an actual playthrough is necessarily fictionally true is not altered by the sheer amount of playthroughs). She does say (53n4) that every faithful performance of a play contains all play‐truths, but this sheds little light on which playthrough‐truths that correspond to video game‐truths, and risks begging the question: playthrough X is faithful because it contains fictional truths p₁ … p_n, and p₁ … p_n are all fictionally true because they appear in faithful playthrough X. The second horn is that playthrough‐truths and video game‐truths have no necessary correlation, but that yet again raises the question of how we ever access the latter, and, more importantly, why we should not regard each playthrough as its own, self‐contained fiction.

Willis's (46–47) own solution to the problem of how to account for fictional truths absent in some representations of a given fiction is that a video game‐truth must be present in at least one playthrough without being contradicted in another. However, Willis neglects Walton's theory on unportrayed fictional truths, which provides a more economical solution: Walton (60) asserts that what is fictionally true must be true in any game where the work functions as a fictional prop. This relates to Walton's assertion that props “generate fictional truths independently of what anyone does or does not imagine” (38; emphasis added). Walton further adds: “Fictional propositions are propositions that are to be imagined—whether or not they are in fact imagined” (39; original emphasis). As an example he mentions how children may make it fictional in their game of make‐believe that all stumps in the forest are bears, and that all stumps therefore function as props, even the ones not discovered. Hence, to not imagine what is prescribed in the game of make‐believe would be to play the game improperly, and Willis does not explain how Alistair's sister differs from this principle. The sister can be regarded as a fictional prop present but not discovered in all playthroughs of the game, always prescribing imagining a certain proposition regardless of whether a given player chooses to do so. This seems more intuitive, especially when compared to how we would assess similar practices in relation to other media: if interpreters skip a scene in a movie or rip out pages in a book, we would not claim that they provide counterexamples to Walton's thoughts on fictionality, but would accuse them of using the props improperly in a way with no bearing on fictional truth.

Furthermore, Willis's criterion of noncontradiction—that a video game‐truth must be portrayed in at least one playthrough but never contradicted in others—suffers from at least two problems. The first problem is that we risk ending up with an infinite amount of video game‐truths. If no two playthroughs are identical, every action or event present in a playthrough would count as an addition to that game's set of video game‐truths, insofar as said action or event does not consist of a binary choice where one playthrough contradicts another. Thus, all playthroughs would add video game‐truths ad infinitum. This cannot be countered by distinguishing between actual and possible playthroughs as Willis does herself when she claims that an actual playthrough could contain no video game‐truths, since actual playthroughs are also possible playthroughs (entailing that all playthrough‐truths must, contra Willis, also be video game‐truths). Moreover, some playthrough‐truths should intuitively not be fictional: cheats, bugs, and glitches appear in some possible playthroughs, but we would hardly say that it is fictional, for instance, that the protagonist clips through a wall only because it is not contradicted in other possible playthroughs.

However, the criterion of noncontradiction could also be too strict, leading to the second potential problem: if no two possible playthroughs are identical, there will be no playthrough‐truths that are not contradicted by other playthroughs, from which it follows that there will be no video game‐truths. For instance, in The Legend of Zelda: Ocarina of Time, the protagonist Link can speak to townsfolk, but doing so is optional. Hence, in some playthroughs it is fictional that Link speaks to townsfolk, in others that he ignores them and walks straight to the tasks vital for the main plot. If we apply Willis's criterion of noncontradiction, it would neither be true that Link speaks to people, nor that he does not, because both propositions are contradicted by other playthroughs. Games with multiple endings would have no ending at all, since they contradict one another. A playthrough lasting only a minute would not contain the same video game‐truths as other playthroughs, as Willis (47) argues, because if the hero dies after a minute it would contradict playthroughs where the hero survives. Not even cut‐scenes would necessarily be fictionally true if we take into account those possible playthroughs in which cut‐scenes are skipped. If all possible playthroughs must contain all cut‐scenes, it would nevertheless mean that only noninteractive parts of games are fictionally true—since only these are identical in all possible playthroughs—which further undermines Willis's claim that video games are unique in terms fictional truth because of their interactivity.² Moreover, we could never say that a given proposition p is fictional since there is an infinite amount of possible playthroughs for every game, so there could always be a possible playthrough yet to be actualized where p is contradicted.

The underlying fallacy in both of Willis's examples is that she presupposes that games are fundamentally different and then tries to construct a new model of fictional truth. Her two examples can easily be accounted for by Walton's theories, so to not regard video games as Waltonian fictions would be ad hoc. My suggestion is that we instead reject the concept of playthrough‐truths and endorse Walton's idea of prescribed imagination and fictional truths that obtain regardless of what any actual interpreter imagines. Note that this model obviates Willis's objection mentioned above concerning the severe limitation it would pose if video game‐truths must be portrayed in all playthroughs, since Walton's model does not decrease the quantity of propositions that can be counted as video game‐truths. It will be true in all playthroughs that Alistair has a sister, not because this fact is not contradicted in some playthroughs (as Willis would have it), but because players are arguably always prescribed to imagine it, even if some of them choose not to. That is, all video game‐truths are prescribed to be conveyed in all playthroughs. This would unfortunately introduce the same problem mentioned above, of how to ascertain whether a given representation is faithful, but in principle my model would not be too different from how we usually approach narratives, and does not necessarily introduce problems not prevalent in the interpretation of narratives in other media; as already mentioned, we would not say that fictional truth in a novel changes only because a particular reader refuses to read certain pages, and although debates concerning matters such as the existence of ghosts in The Turn of the Screw may continue indefinitely, people can nevertheless—at least theoretically—agree that a truth to the matter exists (even if that truth could be that we cannot be certain).

Furthermore, in many linear games the issue of assessing faithfulness to the story could be resolved easily: it is intuitive that a playthrough of Ocarina of Time is faithful if it ends with Link defeating the evil Ganondorf and rescuing princess Zelda, but unfaithful if Ganondorf defeats Link. Exactly how many steps Link takes before reaching Ganondorf varies between playthroughs, but insofar as there is no prescription pertaining to this matter we can label this truth as indeterminate.

It could be objected that this solution works for linear games but not games with mutually exclusive choices, such as RPGs where intricate customization of the protagonist and optional quests result in possible playthroughs with little in common.³ However, this objection neglects the second problem with the criterion of noncontradiction: all choices in games are in some sense mutually exclusive, and many of these—clothes worn, spells and weapons used—are not fundamentally different from aforementioned examples of fictional indeterminacy. The player may be prescribed to play either as a wizard or as a fighter, making it fictionally true in all playthroughs that the protagonist has either of these vocations. My model is similar to Willis's in the respect that each disjunct will be true in at least one possible playthrough, but without the problematic criterion of noncontradiction, so it allows for the same fictional truths to obtain in all possible playthroughs; my model is not undermined by mutually exclusive disjuncts, so (p v q) is always true. Moreover, my model has the added benefit that appearance in an actual playthrough is not sufficient for a proposition to be true (but it is according to Willis's model, since actual playthroughs are also possible playthroughs). Therefore, cheats, bugs, and glitches are not fictional insofar as they are not prescribed, and a one‐minute‐long playthrough has no impact on fictional truth, since one is not prescribed to die and stop playing after one minute. Moreover, possible playthroughs that seem counterintuitive have no bearing on fictional truth in my model: for instance, a playthrough in which the player spends hours making the avatar jump into walls and run around in circles for no reason does not make it fictionally true that the protagonist does the same.

Should one nevertheless insist that some choices are somehow more significant—for example, that choosing a class in an RPG is fundamentally different from choosing the hero's clothes—one would have to explain what it means for a choice to be more significant, why this cannot be resolved with reference to fictional indeterminacy, and why a game with significantly different playthroughs should still be regarded as representations of the same story and not a collection of separate but somewhat similar stories. If we nevertheless grant this point for the sake of argument, one could perhaps argue that such games are more akin to thought experiments offering a set of counterfactuals from which to choose.⁴ This, however, would not necessarily make truth in video game fiction fundamentally different from other media since many games do not prescribe their players to make choices with a significant impact on the story. Moreover, no inherent properties of other media preclude the representation of different sets of counterfactuals, good examples of this in film being Sliding Doors and Run Lola Run, both of which show different possible versions of their respective stories, not to mention fictions in which the protagonist(s) witnesses or visits other possible worlds and alternate timelines, such as It's a Wonderful Life, Rick and Morty, and the Back to the Future—trilogy. Not even games allowing the player to choose between multiple endings pose a serious challenge to my model, since this phenomenon too exists in other media.⁵ Hence, even if it is appropriate to regard certain games as thought‐experiments—and this is a big if—it still does not set these video games apart in terms of fictional truth.

Although I have only presented a tentative, simplified model that has to be developed further, it does resolve many initial problems of assessing fictional truth in video games and has the added benefit of being more economical a solution than invoking concepts of video game‐truths and playthrough‐truths. Furthermore, it is more similar to interpretative practices in relation to narratives in other media; perhaps fictional truth in video games is not so different after all?

REFERENCES