During the past several months, With a Terrible Fate has been working on more theoretically intensive work, focused on analyzing what I take to be one the most important factors that render video games a unique storytelling medium: possibility. Though it does not focus on any particular game, the piece that I now present to you, readers, weaves through “BioShock Infinite,” “Tales of Symphonia,” and “Dishonored,” among others. First and foremost, it is my most recent and involved defense of the uniqueness of video games as an art form, which demands academic attention and a new aesthetics. (In the market for shorter pieces on this topic to get your feet wet? Read this first, or perhaps this.)
Looking for a granular examination of what makes video games the new novel? Read on, and find out.
Many thanks to Ned Hall, who advised me closely on the entirety of this project.
Possibility in Video Game Aesthetics
Aaron Suduiko, Harvard ‘17
Summer 2015
Introduction
It is tempting to think about the fictional “worlds” of representational pieces of art and their propositional content as ‘possible worlds’ in the Lewisian sense of the term, as David Lewis himself does in “Truth in Fiction” (1978). However, as Kendall Walton notes in Mimesis as Make-Believe (1990), there are significant reasons to avoid this temptation: in particular, the worlds of fiction are often either incomplete, impossible, or both. Yet, as I will show in this paper, the narratives of video games (where, by ‘video games’, I mean ‘single-player, role-playing video games’, except where otherwise noted) are uniquely poised to avoid these issues, and to reward a possible-worlds semantics analysis with surprising aesthetic insights.
I begin in Part I by showing that a particular type of propositional content exists in video games which lacks analogue in traditional narrative media: what I call ‘verifiable disjunctions’, the truth values of which can be “checked” by examining the representational work of art in question. In Part II, I rehearse the reasons why fictional worlds typically ought not to be defined as possible worlds; I argue that these reasons are the result of indexing fictional worlds as possible worlds with respect to the appreciator’s actual world, whereas video games use the concept of possible worlds without reference to the appreciator’s actual world. I close in Part III by exploring some of the potential areas of aesthetics that this feature of video games has the potential to impact.
Part I: Verifiable Disjunctions
You decide to enjoy a work of fiction. Perhaps it is a novel; perhaps it is a film; perhaps it is a video game. Regardless of medium, your ability to follow the fiction’s continuity, draw inferences about the fiction, and, generally speaking, to reason about the fiction, requires that you have some way of determining what is true within that particular fiction. We might rightly wonder, then, what necessary and sufficient conditions are for some proposition P to be true within some fiction F – and, supposing whatever it means for a story to be ‘fictional’ to be consistent across all representational media, we would expect these conditions obtain in a film just as much as in a book. However, I contend that these conditions do vary across media, and that we therefore cannot suppose truth in fiction to have the same necessary and sufficient conditions in all fictional stories. I show this by focusing on a particular class of propositions, whose necessary and sufficient conditions for truth are unique to the representational medium of video games.
I begin by outlining a general framework of explicit and implicit propositions in works of fiction, the former of which are propositions that constitute the text of a given work, and the latter of which are propositions that, while not expressed by the text of a given work, must obtain in order for the events described by the narrative’s explicit propositions to be adequately explained. I then show that the design of video games as a medium supports ‘real possibility’, which uniquely bridges the gap between explicit and implicit propositions in fictions: that is to say, real possibilities in video games are implicit truths that can be made to be explicit truths. Finally, I show that real possibility underpins what I call ’verifiable disjunction’, sets of potentially explicit, mutually incompatible propositions, which allow video game narratives to represent distinct possible worlds in a way unavailable to most other media.
Definitions of Terms
I coopt a few peculiar terms and word usages in this paper in order to distinguish between various kinds and elements of story. I use ‘literature’ to refer to any representational art form that contains a narrative. I use ‘narrative’ to refer to the set of explicit and implicit truths supported by a fictional series of events, where ‘explicit propositions’ refer to propositions constituting the text of a given work, and ‘implicit propositions’ refer to propositions that, while not expressed by the text of a given work, must obtain in order for the events described by the narrative’s explicit propositions to be adequately explained. I use ‘fabula’ to refer to the mere events within a narrative, sans interpretation or inference on the part of the appreciator. I use ‘appreciator’ to refer to someone actively engaging a work of literature – e.g., the person reading a novel or playing a video game. I use ‘text’ to refer to the representational elements of a work of literature – e.g., the literal text of a novel or the audiovisual output of a film.
With respect to video games in particular: I use ‘playthrough’ in reference to a sequence of events from the beginning to end of a game’s fabula, as determined by the choices of the relevant appreciator (i.e., the appreciator playing the game at the time). I define a video game’s narrative as ‘the set of all possible playthroughs of the video game’. The ‘text’ of a video game refers to all the audiovisual output of its narrative. A ‘player’ is the appreciator of a video game. An ‘avatar’ is the character(s) in the video game that the player controls.
Explicit and Implicit Propositions in Fiction
I prescind in this paper from a precise analysis of necessary and sufficient conditions for proposition P being true in fiction F; rather, my argument will be that, given any plausible analysis of these conditions, there will be different necessary and sufficient conditions for truth in the case of a particular type of propositions in video games. I will be satisfied with talk of a particular work of fiction as ‘a set of propositions S that holds across some set of worlds W, distinct from the appreciator’s actual world’. These worlds need not be as restrictive as Lewisian possible worlds.[1] Rather, it is sufficient for our analysis that the narratives described by S are generally intelligible to the appreciator.
We distinguish between two major classes of truths within the fiction: explicit and implicit truths. ‘Explicit propositions’ refer to propositions that are affirmed or denied by the text of the literary work in question. This is most straightforward in linguistic works, the text of which comprises explicit propositions; while other forms of text may have the potential to be less clear with respect to propositional content, I take it to be the case that such forms are also typically clear enough to establish explicit propositions. For example, when we see a Jason Bourne portrayed by Matt Damon running across rooftops in “The Bourne Supremacy,” I take it to be the case that the proposition ‘Jason Bourne is running across rooftop’ is affirmed by the text as unambiguously as a book might affirm the same proposition.
More ambiguous are the ‘implicit propositions’ supported by a literary work. I use this term to refer to propositions that are not explicated by the text of a work of literature, but which must obtain in order for the events described by the narrative’s explicit propositions to be adequately explained. By definition, these do not exist within the text of the work, and so the appreciator is left to derive these truths by inference. Some such propositions are noncontroversial to the point that they do not bear mentioning: unless given evidence to the contrary in the form of an explicit truth, I take it to be an implicit truth in most works of fiction that the characters that are alive are regularly respiring – this explains the persistent existence of characters through the course of the narrative. However, there are many cases in which the truth-values of implicit propositions are less obvious than this, and, in some such cases, multiple, mutually incompatible inferences seem plausible for the reader to make. I will focus on two cases that exemplify this: implicit propositions with seemingly indeterminate truth-value, and post hoc counterfactuals.
In some literary works, there are times at which it seems that a constituent of the fabula is conspicuously absent – that is, there exists some proposition P with an explanatory relation to the explicit truths of the narrative, and yet P is not itself an explicit truth within the narrative, and its truth-value, unlike the example of respiration, is not readily apparent. Take, for example, the final scene of Christopher Nolan’s “Inception” (2010): a central plot device of the film is that each character has a “totem” by which they are able to determine whether they are currently conscious in reality or a dream. Cobb, played by Leonardo DiCaprio, carries a spinning top as his totem: if he spins it and it ultimately stops spinning, then he is in reality; if it spins ad infinitum, then he is in a dream. In the last scene of the film, Cobb’s goal of the movie is fulfilled: he is reunited with his family. He sets his top spinning on the table, and then goes to greet his family. The top, left spinning on the table, begins to wobble, as though it were about to fall; at this moment, the film ends and the credits roll. It seems that there must be a fact-of-the-matter about whether or not Cobb is in a dream during the final scene: the fact as to whether the top continues spinning or falls has an explanatory relation to the explicit propositions of the film because it makes clear whether the final events of the fabula are merely dreamed, or obtain in reality. Yet this fact is withheld from the viewer of the film; in fact, it seems that the aesthetic experience of that moment aims to frustrate the appreciator by withholding the fact-of-the-matter about the top from them, where it seems at least plausible that it might stop or keep spinning. Accordingly, appreciators argue about whether or not the last scene was in fact a dream, drawing on explicit truths from the work, together with other implicit truths. Importantly, because the debate is about the truth-value of an implicit proposition – ‘the top ultimately stopped spinning’ – it cannot be resolved by drawing on the text of the work itself.
Similar dynamics apply with respect to a common type of counterfactual analysis in literary works. It is natural to suppose that there are different ways in which the fabula of narratives could have evolved, though these ways in fact did not obtain in the narrative – so, for instance, one could say it is possible that Harry and Hermione could have ended up together in the Harry Potter series, although they in fact did not. In this way, narrative possibility can serve to generate implicit propositions that establish other “possible worlds” of a given narrative, with the explicit propositions of the narrative referring to the relevant “actual world.” I call such counterfactual analysis ‘post hoc counterfactual analysis’ because it is performed after the appreciator already knows the relevant explicit propositions in the narrative, which are being negated by the counterfactual. As in the “Inception” case, appreciators can plausibly argue about the implications of post hoc counterfactual analyses in a way that cannot be resolved by the text of the literary work, because these analyses are concerned with implicit propositions.
To see this, consider an example of two appreciators, John and Susan, who have both read Jack and the Bean Stalk and are perfectly knowledgeable about all of its explicit propositions. They discuss the work together, and wonder aloud about what would have happened had Jack, contra fabula, been discovered by the Giant on his first infiltration of the giant’s home. John supposes that the giant would have instantly killed Jack; Jack’s mother would think him missing, organize a search party, ultimately presume him dead, weep, and so forth. Susan disagrees with John: she thinks instead that the giant’s wife would have protected Jack from the giant’s ire, and that Jack would have sent home, living out the rest of his days as a quiet peasant, terrified of the giant’s wrath. Both John and Susan’s inferences seem plausible, and the appreciators are clearly not confused about the explicit propositions of the story in question; yet, as in the case of Cobb’s top, they are able to rationally argue about the narrative, because the literary work’s text lacks the resources to provide determinate truth-values for the implicit propositions at stake in the analysis; therefore, the text cannot explicitly provide an answer to their debate.[2]
Real Possibility
Video games contain a dimension of possibility that gives rise to a class of propositions that are sometimes explicit and sometimes implicit – this type of possibility is what I term ‘real possibility,’ the fact that the text of a video game (that is, the total output of all possible playthroughs) supports more than one distinct fabula.[3] Video games contain choices for a player to make, which lead to distinct outcomes; this can be “choice” in the microcosm of a player moving her avatar in different directions locally, or “choice” in the macrocosm of pushing the entire fabula in drastically different directions globally. The capacity for a player to choose in a game is underpinned by the potential for a game’s fabula to vary in accordance with those choices.
Real possibility is a feature of video games because the set of explicit propositions described by the text of a video game is larger than the set of propositions to which the player has access when she engages that work of literature from the beginning of its fabula to the end of its fabula – something that verges on nonsensical in more traditional forms of literature. If someone reads a novel attentively from its beginning to its end, then we assume by definition that the reader has ascertained all the explicit propositions of the novel; likewise in the case of a film. However, a video game, by virtue of being a program that supports the player making various, mutually incompatible choices, only presents some of the explicit propositions of its overall text, in accordance with which choices the player of it makes. This is why multiple playthroughs of the same game can be substantively different, whereas we assume that the explicit content of a novel or film is held fixed across multiple “read-throughs” or “watch-throughs.”
By way of example, consider the narrative of the video game “Dishonored” (Bethesda, 2012), which centers on the revenge/justice quest of the avatar and protagonist, Corvo, who has been framed for the murder of an Empress. At one point in the game, Corvo must confront and neutralize the Lord Regent, who was the mastermind behind the murder and who subsequently usurped the throne. There are two possible courses of action with respect to how Corvo can “defeat” the Lord Regent and progress in the narrative: he can either assassinate him or publicly expose him as a fraudulent criminal. In order for the narrative to continue, the player must choose one of these courses of action to undertake. When we consider the potential courses of action, then, we see that one of them will become explicit within the narrative – that is, whichever course of action the player chooses to undertake. This course of action will obtain within the playthrough, and be taken as explicit fact within the world represented by the playthrough. The other course of action, in contrast, becomes non-actual – it is something that could have happened, but did not, because the player chose not to make it so. It becomes implied propositional content.
However, the player could just as easily play through the game a second time and choose to undertake the other course of action, in which case the exact same event that was previously an implicit proposition becomes an explicit proposition, obtaining within the fabula of this second playthrough. This potential for the player to make implicit propositional content explicit – that is, real possibility – is characteristic feature of video games as a medium, and it is what affords their stories a domain of counterfactual analyzability that is textually grounded, and thereby more concrete, than the post hoc counterfactual analysis discussed above.
Verifiable Disjunctions
With respect to this special type of counterfactual analysis, one can actually determine the fact-of-the-matter about implicit propositions in precisely the way that is unavailable to post hoc counterfactual analysis. That is to say, when two players of the same game are debating what would have happened in a game’s narrative had a certain type of event not obtained, the players can actually “go and check” what actually would have happened in the game. I call mutually incompatible propositions that belong to this set, such as ‘Corvo killed the Lord Regent’ and ‘Corvo exposed the Lord Regent’, ‘verifiable disjunctions’ of propositions, because the piece of literature in question, by virtue of real possibility, provides the explicit content necessary to determine what would actually happen in the narrative in the case of counterfactuals concerning these disjunctions. Appreciators can check the truth or falsity of their own counterfactual reasoning in a way that John and Susan cannot in their Jack and the Bean Stalk debate.
A work of literature contains verifiable disjunctions just in case its narrative consists of branching series of events that are mutually incompatible – that is, for at least one event p, the narrative describes two or more series of events proceeding from the time at which that event would occur, at least one of which evolves from p and at least one of which evolves from q, where p and q are mutually incompatible. Real possibility makes this possible in the case of video games: so, one player may choose to take some action phi at a particular point in a game’s narrative, although they also had the option to take some different action psi at that same point. This means that the mere events a particular player experiences when playing through the entirety of a game’s narrative – i.e., a ‘playthrough’ of the game – will not necessarily be the same as the mere events experienced by a different player playing the same game.
Part II: Constructing Possible Worlds from Verifiable Counterfactuals
There is a danger when analyzing fictional worlds to jump to an analysis of them as possible worlds in a way that is ultimately misleading and unproductive. In order to distance the analysis at hand from this danger, as well as to pick out precisely which features of video games provide the basis for a possible-worlds semantics, it will be useful to arrive at these semantics by way of a survey of the traditional usage of possible worlds in analyzing fiction. Therefore I begin this section with a review of Lewis’ own proposal for treating fictional worlds as possible worlds; I then rehearse the broadest criticisms of this proposal, as noted by Walton; finally, finally, I show that the verifiable disjunctions in video games motivate a type of possible-worlds semantics that lacks the features that make Lewis’ proposal unappealing.
Lewis takes the seeming fact of truths within a given fiction being closed under implication as grounds for describing fictional worlds as certain sets of possible worlds (39). He proceeds to offer three analyses of truth in fiction in terms of possible worlds, of which I focus on what Lewis names ‘Analysis 1’: “A sentence of the form ‘In the fiction f, phi’ is non-vacuously true iff some world where f is told as known fact and phi is true differs less from our actual world, on balance, than does any world where f is told as known fact and phi is not true. It is vacuously true iff there are no possible worlds where f is told as known fact.”[4] This analysis picks out one way in which possible-worlds semantics can be a powerful tool in thinking clearly about fictional worlds: namely, the relative closeness of worlds in which the fiction is true to our actual world gives us grounds for what sorts of inferences it is rational to make as readers when we introduce our background knowledge to the explicit propositions within the literary work. To borrow an example from Lewis, it is right to infer that Holmes had two nostrils and wrong to infer that he had three because worlds in which the stories of Sherlock Holmes are told as known fact and Holmes has two nostrils differ less from our actual world than do those worlds in which the stories of Sherlock Holmes are told as known fact and Holmes has three nostrils. This is a useful basis for determining the more clear-cut of implicit propositions in a work of fiction (Lewis’ analysis may have insight with regards to less clear-cut cases, such as the fact-of-the-matter about Cobb’s top, but this is not readily apparent; for the purposes of the current analysis, I will set this matter to the side.)
Yet although Lewis’ analysis offers us a robust way of making rational inferences about literary works, there are strong reasons to believe that Lewis fundamentally misconstrues what it means for something to be a ‘fictional world’. As Walton observes, “[fictional] worlds are sometimes impossible and usually incomplete, whereas possible worlds (as normally construed) are necessarily both possible and complete” (64). I am more concerned with the potential for impossible fictional worlds than I am with the incompleteness of fictional worlds. I mirror Lewis’ own assessment of his analysis in this regard: the analysis has no apparent problem with respect to completeness, because the world of a particular work of fiction can be described by a set of possible worlds, rather than a single possible world: the explicit propositional content of the work will hold fixed in all members of the set, and the remaining propositional content of the worlds will be filled by reasonable inferences, as described by the analysis itself.
Much more problematic, I think, is the demand for the possibility of possible worlds in the face of many fictional worlds that are clearly impossible. Lewis himself notes near the end of “Truth in Fiction,” “[according] to all three of [his] analyses, anything whatever is vacuously true in an impossible fiction. That seems entirely satisfactory. […] We should not expect to have a non-trivial concept of truth in blatantly impossible fiction, or perhaps we should expect to have one under the pretence—not to be taken too seriously—that there are impossible possible worlds as well as the possible possible worlds” (45-46). Yet it seems perfectly reasonable to expect that certain propositions in a fictional world are determinately true and that others are determinately false, even when the world is taken to be “impossible” in the relevant sense.
Take Escher’s Relativity (1953, pictured above) as an example. It is obvious that, according to our basic knowledge of the logic of spatial relations, the painting represents an impossible world. Yet it is equally obvious that certain propositions are false in this impossible world. For instance, it is not the case that walking in either direction on the stairway nearest the top of the canvas will lead one to arrive at the location of the person on the stairway nearest the bottom of the canvas. Insofar as we must take the painting to explicitly represent the content of its fictional world (a fairly standard assumption to make), we can tell this merely by looking at the ends of the top stairway in relation to the position of the man on the bottom staircase.
This is not an artifact of peculiarities of Escher of the visual medium of painting – in fact, the falsity of certain propositions within an impossible fictional world is even clearer when one turns to literature. For when a representational work of art includes a particular plot, one can merely say that it is not the case in the relevant fictional world that event phi does not at some point obtain, where phi is an event constitutive of the work’s plot. Authors like Borges and Calvino regularly crafted stories in worlds that were logically impossible, and yet it would be nearly impossible to discuss those stories at all on Lewis’ account of fictional truth (cf. The Garden of Forking Paths; Invisible Cities).
Lewis would probably object that his analysis is a useful basis because we make sense of fiction by indexing it relative to what we take as non-fiction – that is, relative to our actual world. That is why his counterfactual Analysis 1 is so useful, and is also why an analysis of impossible fictions cannot get off the ground in the first place. Taking this view of fiction, one might say that it is actually a feature, rather than a drawback, of Lewis’ analysis that it shows precisely why impossible fictions are ultimately incomprehensible to a reader.
This reply is unsatisfactory because it severely limits the scope of what we can take to be “meaningful fiction”: it may be the case that the Invisible Cities of Calvino are impossible, but that does not make the fiction incomprehensible to the reader. We can recognize that Escher’s Relativity in impossible in relation to our actual world, while also understanding how the world’s own “impossible” spatial relations work; this of course requires that the fictional world be internally consistent, but it demands no particular possibility relation between fictional and actual world. Examples like this lead me to conclude that an understanding of fiction which indexes fictional worlds relative to our actual world is misguided, because it artificially limits what we as appreciators of art take to be fictional worlds.[5]
Lewis’ possible-worlds semantics is therefore inadequate for fictional worlds as a general concept; however, it remains a powerful tool for analyzing the various possibilities for which a single work of fiction allow. As Lewis discusses, such possibilities come about in works of traditional works like Sherlock Holmes from different inferences a reader can rationally make, using their background knowledge, from the explicit propositions of the work (42-43). So, trivially, the Holmes stories describe possible worlds in which Holmes had an even number of hairs on his head, and possible worlds in which Holmes had an odd number of hairs on his head. Less trivially, this same semantics underpins how different types of counterfactual analysis function in thinking about a given work of fiction.
When an appreciator undertakes ‘predictive counterfactual reasoning’ while engaging a work of literature, she entertains different possible ways that a story could proceed, although she knows perfectly well (at least in traditional media) that there is only one explicit fact-of-the-matter about how it does proceed. In Lewisian terms, she is imagining distinct possible worlds in which the events of the work of literature, up to the point in the work at which she currently is, all obtain, but in which the chains of events evolve differently thereafter. The appreciator weighs different inferences about the work in her head, wondering which is most plausible, in order to include which of the contemplated possible worlds is most likely to obtain within the set of worlds described by the totality of the literary work (that is, in which of the imagined possible worlds the actual plot of the overall work obtains).
Lewis’ model also explains the potential for rational disagreement between two appreciators engaging in post hoc counterfactual analysis. Returning to our example of the two critics of Jack and the Bean Stalk, John and Susan, Lewis’ possible-worlds semantics would describe the disagreement as the two critics picking out distinct possible worlds that are counterfactually related to worlds in which the actual plot of the story obtains. Both are rational counterfactual analyses, but they pick out distinct possible worlds, probably as a result of each critic making inferences using different background knowledge from their own experiences, or else as a result of each critic weighting different aspects of the texts are more or less significant to resolving the counterfactual in question.
But where Lewis’ approach to fiction is most useful is in analyzing verifiable disjunctions. In the cases of predictive and post hoc counterfactual analysis, possible-worlds semantics are useful only in perspicuously explaining what is going on when the appreciator of a work of fiction makes inferences with respect to the explicit propositional content of the fiction; possible-worlds semantics are conceptually prior to such reasoning in the case of verifiable disjunctions, because the semantics here describe the dynamics of the work’s explicit propositional content itself: that is to say, Lewisian possible-worlds provide an analytic framework by which to make the real possibility of video games rigorous.
The utility of possible-worlds semantics in this regard is a result of real possibility making multiple possible worlds explicit within a work of literature. Recall that the inadequacy of Lewis’ counterfactual analysis of fictional truth is a result of indexing fictional worlds relative to the appreciator’s actual world; in the case of narratives containing verifiable disjunctions, possible-worlds semantics can be applied in such a way as to retain their utility without drawing implausible conclusions about the scope of fiction; this is because these semantics make no reference to the appreciator’s world as the actual world in the Lewisian sense – the ‘actual world’ is instead the current playthrough of the video game in question.
To see how such possible-worlds semantics might be leveraged in an analysis of literary works containing verifiable disjunctions, let us imagine some generic video game chi. It needn’t be so dynamic and choice-dependent a game as “Dishonored,” but it may just as well be. Call any distinct playthrough omega of chi some possible world with respect to the narrative of chi. That is to say, omega is a possible world within chi if and only if omega is some continuous sequence of events by which the player can proceed from the beginning of the game to its conclusion, or some portion thereof (in which case it is a ‘partial playthrough’). omega1 and omega2 are distinct possible worlds of chi just in case each ostensibly constitutes a complete fabula of chi, but the two differ by at least one event.[6] omega is the actual world of chi with respect to a given player alpha if and only if omega is a possible world of chi, the constitutive events of which describe the path by which alpha does in fact proceed from the beginning of the game to its conclusion. With these definitions in hand, we can offer a formal definition of verifiable disjunctions that puts it in terms of game narratives: a verifiable disjunction is any disjunction of two or more events within a narrative such that, given partial playthroughs of the game that are indistinguishable up to the point of narrative disjunction, a player knowledgeable about the game narrative can choose to advance the playthroughs via any constituent event of the disjunction, thereby differentiating the previously indistinguishable partial playthroughs as distinct possible worlds.
Lewis gives us a robust semantics by which to describe and analyze the verifiable disjunctions inherent to video games. Because the various possible worlds are indexed to a fictional world serving as the literary work’s ‘actual world’, the semantics does not run into the same problems that Lewis does by indexing fictions to the actual world of the appreciator. This allows us to coherently pose counterfactuals of impossible fictional worlds, such as: “Where would the man at the bottom of Escher’s Relativity end up, if he were to walk up the staircase on which he is standing?” And because possible worlds in video games are distinguished by verifiable disjunctions, we have the resources to determinately answer such questions as well.
Part III: Exploring the Aesthetic Implications of Possible Worlds in Video Games
The reader might agree with all that I have argued thus far, but still complain that the framework I have sketched of verifiable disjunctions and possible worlds in video games yield no interesting aesthetic possibilities that cannot be realized in other ways. I therefore wish to close by mentioning ways in which I believe these dynamics have the potential to enhance and alter the aesthetic experience in more general terms. I do not aim to pick out every way in which the preceding dynamics might influence aesthetics, nor do I aim to settle the matter of how the dynamics influence any one aesthetic phenomenon; rather, I point to the areas of aesthetics that I find most obviously susceptible to significant alteration by verifiable counterfactuals of games, and offer a first pass at the ways in which they might be so altered. In particular, I will focus on the following three areas of interest: resistance on the part of the appreciator; responsibility of the appreciator for the events of the narrative; and necessary ignorance of the appreciator during their first time engaging the work of literature.
Resistance to Restrictions on Possibility
In the last section of his “The Expression of Feeling in Imagination” (1994), Richard Moran addresses what he calls ‘imaginative resistance,’ which finds its origins in Hume’s idea that appreciators of aesthetics, while typically willing to accept the propositional content constituting a work of fiction, resist moral insinuations in the work that contradict one’s own moral sensibilities (cf. Hume, Of the Standard of Taste). Moran observes that such imaginative resistance is much more general a phenomenon than Hume takes it to be, that “it is part of any aesthetic response that claims a distinction in principle between the emotional solicitation of the work (whether as laughter, or pity, or horror) and the response one actually gives to it” (97). Regardless of how we ultimately want to explain such resistance, I take Moran to be correct about the scope of it: there exists at minimum a potential for friction between what a fictional work suggests is a “proper” response from the reader, be it emotional or moral, and how the reader feels in actuality about the work. So, for example, one aspect of an appreciator engaging Jack and the Bean Stalk will be whether the reader feels that Jack killing the giant and escaping with his riches is a proper happy ending to the narrative.
Verifiable disjunctions expand the domain of potential imaginative resistance beyond the realms of morality and feelings in Moran’s sense. In particular, they make possible resistance to what is possible and what is impossible within a given literary work. Consider by way of contrast the resistance that might come about through predictive counterfactual analysis: the appreciator might predict that a work will most likely evolve in a certain way, and therefore resist the textual matter-of-fact that it evolves in a different way (I take this to be a fairly common aesthetic phenomenon). Yet when an appreciator resists a work in this manner, it is not the case that she is resisting any sort of possibility within the work: what she is resisting is the actual world established by the explicit propositions of the work – or, if you prefer, the minimal set of possible worlds that are compatible with the total explicit content of the work. That this is distinct from questions of possibility is apparent from the fact that appreciators engage in post hoc counterfactual analysis: post hoc counterfactual analysis, as I discussed earlier, is fundamentally the act of entertaining different possible worlds that differ in one or more events from the explicit content of the narrative. The actual content of the narrative does not really restrict what counterfactuals the reader might entertain, except inasmuch as it makes the possible worlds they entertain closer or further away from the actual world(s) picked out by the explicit propositional content of the narrative.
But verifiable disjunctions do seem to set restrictions on what possible worlds the reader can rationally entertain within the context of a work of literature, for one phenomenological impact of distinct possible worlds represented explicitly within a literary work is that these possible worlds seem ontologically privileged over the sorts of post hoc “possible worlds” imagined by an appreciator outside the scope of the work’s explicit propositional content. Works that feature verifiable disjunctions tend to feel more modal realist in the Lewisian sense than our actual world does, precisely because they have the explicit resources, as defined in Part II, to present distinct possible worlds and describe their relations to one another, as well as to the relevant actual world (i.e., the current playthrough of the game in question, to which other possible playthroughs are indexed). In light of this, it seems that the appreciator is less justified in posing counterfactuals that are not represented by the literary work’s text. To see this, return to the example of killing the Lord Regent in “Dishonored”: the total possibilities of how the narrative proceeds from that juncture are as follows: ‘Corvo assassinates the Lord Regent’ v ‘Corvo publicly exposes the Lord Regent as a fraud’ v ‘Corvo dies and his quest is concluded in failure’.[7], [8] If an appreciator were to ask the question, ‘What would have happened at this point in the narrative, had Corvo not assassinated the Lord Regent?’, then someone with knowledge of the game would presumably answer, ‘Either Corvo would have publicly exposed the Lord Regent as a fraud, or Corvo would have died and his quest would have concluded in failure’. If the appreciator persisted by asking, ‘What would have happened in the narrative, had none of those events obtained?’, then the most plausible answer is that those are the only possibilities by which the narrative could proceed. Because what is possible is explicit within the literary work, verifiable disjunctions have the potential to restrict what the appreciator is justified in conceiving as possible within the narrative.
It is this restriction of possibility which grounds the potential for resistance on the part of the appreciator to what is possible and impossible within the scope of a narrative with verifiable counterfactuals. For example, in “BioShock Infinite” (Irrational Games, 2013), one of the early events in the narrative is a couple of characters prompting the player’s avatar to choose whether a coin flip will have the outcome of “heads” or “tails.” There are several such binary choices in the narrative, and the player typically is able to select one of the two outcomes; however, in this case, the avatar chooses the “heads” outcome without the player’s input. If the player expects to have agency over this choice, as she does elsewhere in the game, then the fact that there exists no such possibility here invites resistance in her response to the literary work. “BioShock” incorporates this sort of possibility constraint into its overall aesthetics: the game takes place within an explicit many-worlds quantum mechanical framework, and the narrative turns on the fact that there are some constants between discrete realities, as well as some variables. At the moment of the coin toss in the game, the player sees a tally that has been kept of previous coin tosses, which evidently were conducted in other realities; all previous outcomes were “heads,” implying that this moment is one of the constants in the universe. The ability of the game to literally restrict possibility through verifiable counterfactuals allows them to structure the set of possible worlds described by the game in such a way as to reflect that very notion of constancy in tension with variability. (Note, however, that, as “Dishonored” shows, verifiable disjunctions, rather than the particular subject of “BioShock” being many-worlds quantum mechanics, ground the added aesthetic dimension here.)
Responsibility of the Appreciator for the Events of the Actual World of the Narrative
Because video games bound possibility through verifiable disjunctions, and because the player’s actions determine which of the game’s possible worlds is made actual, the potential exists for the player of a video game to bear responsibility for the events that obtain in a playthrough of a video game. This is something that distinguishes video games from the traditional novel or film, in which, though the appreciator may certainly feel satisfaction or regret at the events that transpire, they would be confused to feel somehow responsible for those events (unless they happened to be the author of that work of literature).
Return to the example of Corvo killing the Lord Regent. There is an explicit possible world within the narrative of “Dishonored” such that Corvo advances through the narrative and the Lord Regent remains alive. Therefore, insofar as this is possible and the player knows that this is possible, the player, should she choose to kill the Lord Regent, is directly responsible for his death. She knows directly, by virtue of the game’s narrative, that it is not necessary for the Lord Regent die in order for the game’s narrative to progress; she is therefore culpable if Corvo assassinates him (as, contrariwise, she is culpable if he remains alive – for she also knows that it is possible for Corvo to kill him, and that he will survive otherwise).
Just as verifiable disjunctions make clear when the player is responsible for an event obtaining in the actual world of a narrative, so, too, they make clear when a player is not responsible for an event obtaining in that narrative. One test for this is simple: if a given event phi obtains in all possible worlds within a game, then the player is not responsible for phi. Beyond this test, the question of responsibility becomes increasingly complicated – for instance, a player may unknowingly bring about certain events in the actual world of a game by ignorance of the game’s overall causal structure – but it is intuitive that if some event occurs in all possible versions of the game’s narrative, then the player as an agent is not responsible for that event obtaining in the narrative. For example, in “Final Fantasy VII” (Square, 1997), the main antagonist kills one member of the player’s party of adventurers early on in the game, and this event is unavoidable. This contrasts starkly with the choice presented to the player of “Dishonored” as to whether or not Corvo kills the Lord Regent.
Because video games combine the agency of the player with a programmed text, there exists a large degree of flexibility with respect to just how much responsibility a player can have within the game. Insofar as games follow the possible-worlds semantics sketched above, they all espouse some type of compatibilism: the game’s programming determines the set of possible worlds and their explicit propositional content, while the player determines which of these possible worlds is made actual. How large the set of possible worlds is, how much they differ from one another, and the degree of transparency of the effects of a player’s choices are all examples of variables that a game can modify to alter the actuality and phenomenology of a player’s responsibility.
Ignorance of an Appreciator when First Engaging a Video Game
I mentioned above that player responsibility may be complicated by the fact that a player might unknowingly bring about certain events out of ignorance of the game’s overall causal structure; this points to a more general feature of games that is grounded in verifiable disjunctions: namely, insofar as more than one possible world exists within a game, a player cannot know all of the explicit content of the narrative, nor the total causal structure of the narrative, merely by playing the game once.
This contrasts starkly with traditional media: although one might reread a book, watch a movie more than once, or view a picture more than once, one would not expect the act of engaging such a work a second time to somehow instantiate new explicit representational content. But this is precisely what happens in the case of video games, assuming that the player makes different choices on a second playthrough and does not bring about the exact same actual world. Engaging the narrative once, in the case of narratives with verifiable disjunctions, is insufficient to glean the entirety of the narrative’s text.[9]
There is a variety of distinct ways in which this feature of game narrative can impact aesthetics and player phenomenology, and here I will focus only on what I take to be the two broadest such ways: transparent counterfactual relations, and opaque counterfactual relations. This picks out the difference between distinct possible worlds of the game, the existence of which is apparent to the player, and distinct possible worlds of the game, the existence of which is not apparent to the player.
Transparent counterfactual relations characterize games such as “Dishonored”: the game employs an obvious choice system, typically binary. In “Dishonored,” the player can either complete objectives via “high-chaos” choices (such as assassinating the Lord Regent) or “low-chaos” choices (such as exposing the Lord Regent as a fraud). The type of path that the player most often chooses influences the development of the world around Corvo, and ultimately leads to distinct outcomes to the narrative. The game makes this structure obvious to the player; thus, while a player proceeding through the game for the first time via a low-chaos path will not know what the game’s high-chaos worlds look like, he is fully aware that such worlds exist. He is consequently aware during the game that his choices influence the world with respect to the degree of overall chaos, and is motivated upon completing the game to explore the other path through the game with respect to chaos.
But verifiable disjunctions are under no requirement to be transparent to the player, and many games reflect this by presenting disjunctions that may be completely opaque to the player – what I call ‘opaque counterfactual relations’. For example, “Tales of Symphonia” (Namco Tales Studio, 2003) features a “relationship system,” whereby the player has the opportunity to associate the primary avatar of the game, Lloyd, with other members of the party more or less closely over the course of the game – for example, there are optional interactions Lloyd can have with certain characters, and at times the party must split into two groups, the members of these groups being determined by the player; these events influence the degree to which the secondary characters bond with Lloyd. No such relationship system is explicated within the game, so it is not surprising that its effects are also not obvious; however, the effects are quite significant with respect to the overall course of a playthrough. One of the secondary characters in Lloyd’s party is his father, Kratos, although Lloyd does not know that Kratos is his father until very late in the narrative. Eventually, Kratos parts ways with Lloyd’s party, and is subsequently replaced by a different character, Zelos, who fights in a role similar to that of Kratos. Zelos ultimately betrays the party, and the party must defeat him. Although it is not at all apparent to the player during her first time through the game, the defeat of Zelos can evolve in two distinct, mutually incompatible ways (bracketing the evolution of events in which Zelos defeats the party, and their journey ends in failure). In the more easily realizable outcome, the party defeats Zelos, but Zelos ultimately returns, revealing himself to truly be on the side of Lloyd and his friends, rejoining them once more on their quest. However, if the player has developed an especially close bond between Lloyd and Kratos over the course of the game via the relationship system, then the party kills Zelos when he betrays them, and, instead of Zelos joining them, Kratos ultimately returns to the party, joining his son &co on the final leg of their quest. In contrast to the obvious chaos dynamic of “Dishonored,” it is highly unlikely that the player would stumble upon the latter chain of events on her first time through the game; even on her second time through, she would likely only discover this path naturally if she decided to maximize the bond between father and son, once the familial relation between Lloyd and Kratos was revealed late in the player’s first playthrough of the game. The presence of such opaque counterfactual relations, though not as directly motivational for the player to replay the game as transparent counterfactual relations, are most likely more rewarding by virtue of requiring that the player discover their existence within the fabric of the game’s universe – as such, they make the overall universe of a game more of a mystery to the player, allowing the player to act more as a member of the universe than as an external agent with a god’s eye view of its causal structure.
The unifying thread of such verifiable disjunctive relations, which is the reason why I find them particularly promising for the development of future aesthetic theory more generally, is that these relations go above and beyond our actual experience of reality. If we take it to be the case that there is real possibility about different ways in which our world could proceed, then we must still resign ourselves to the fact that we only “play through” reality once, and see only one such set of possibilities made actual. Not so with video games: by playing through more than once, we can explore, discover, and respond to the causal relations of the total universe of the same; we can learn what is possible, what is not, and to what extent we are responsible for the actual.
[1] Nor should they be. By Lewis’ own lights, an account of fiction using his own possible-worlds semantics would render everything vacuously true in impossible fictions – a commitment which would render works like Escher’s paintings and Borges’ short stories unintelligible, which seems to me misguided (cf. “Truth in Fiction,” Lewis 1978).
[2] Indeed, if the work did provide an answer to the debate, then it is unlikely that it would be a debate in the first place. A novel might well contain an explicit counterfactual of the form ‘had it been the case that phi, then psi would have obtained’; but then, it stands to reason that there would be little motivation behind John and Susan debating what would have happened, had it been the case that phi: the work answers this for them.
[3] I do not claim that real possibility is unique to video games; it is obvious enough to me the choose-your-own adventure novels also contain real possibility. I focus on video games because real possibility seems to me a crucial concept by which to explain aesthetic features that are unique to video games.
[4] Of the other two analyses Lewis offers, Analysis 0 is a mere jumping-off point which Lewis himself deems unsatisfying, and Analysis 2 is a reformulation of Analysis 1 specifically designed to satisfy those who want to block the relevancy of little-known facts to the background against which readers critique fiction. I ignore the former since Lewis himself intends it not to be taken seriously, and I agree with him on that score; I ignore the latter because do not I find the degree to which a fact is well-known a compelling basis for excluding particular background knowledge of a particular reader with respect to relevancy within the fiction.
[5] In the specific case of Lewis, my guess is that his aversion to impossible fictional worlds is at least partly motivated by his commitment to modal realism. It is ontologically tidy to take a view of fiction as referring to other realizable versions of reality, but the price for this is rejecting impossible fictions on the pain of admitting impossible possible worlds into modal realist metaphysics.
[6] This condition is weak because there is a variety of ways in which distinct playthroughs of a game might be related to each other. Some playthroughs might begin at a single event and branch to different outcomes; some might begin at different events and converge to a single outcome. I prescind in this paper from any attempt to precisely catalogue all such modes of relation between playthroughs.
[7] Each proposition in this disjunction may in fact represent a set of possible worlds that are for-all-intents-and-purposes identical in this particular aspect. That is to say, there are different specific ways for Corvo to assassinate the Regent, as there are different ways for Corvo to be killed by guards or die accidentally during his mission; for the purposes of advancing the narrative with respect to a causal chain of events, however, such differences effectively wash out in analysis.
[8] The last proposition of this disjunction may be more controversial than the first two, because it tends to be understood as some type of ‘failure’ on the part of the player to advance the narrative, at which point he ought to start over at some earlier point in the narrative and attempt once again to advance the narrative without dying. Keeping strictly to our definitions of possible worlds, such “narratives of failure” still count as possible worlds, and I think this maps well onto intuitions about game narrative as a whole – if it were not strictly possible for the avatar within a video game to die, then the avatar’s being killed would be utterly contradictory. However, if the reader disagrees, then this disjunction may just as well be read without this last proposition; it makes no difference to the current argument.
[9] Here, I am ignoring such cases as players who read a guidebook for a video game, which describes all possible paths through the game, prior to playing the game. Such cases may complicate the dynamics discussed here, but my intuition is that the also violate prescriptions of “how to play the game by its rules,” to borrow Walton’s analysis of representational art as games of make-believe.
