Aesthetics for Birds

Aesthetics and Philosophy of Art for Everyone

JAAC X AFB Discussions #3: Tavinor on Video Games


Welcome to the third AFB x JAAC Discussion!

Today, we will be looking at “What’s My Motivation? Video Games and Interpretive Performance” by Grant Tavinor, available in JAAC’s Winter 2017 volume, 75 (1), online here. Grant is Senior Lecturer of Philosophy at Lincoln University, NZ and author of the book The Art of Videogames.

And big thanks to C. Thi Nguyen (Utah Valley University) for providing the critical précis. Grant’s response follows that, and they will both be available to discuss your questions and thoughts in the comments! Check it all out below the fold.


Video games are, in certain ways, rather odd creations: they can be both interactive and narrative. With a traditional non-interactive fiction, the author provides a narrative with a set of fixed events, and then reader interprets those events. But in video game, the player can be causally entangled with the narrative — their choices and actions partially determine the course of events. The player can interpret past events in order to make narratively informed choices, but this can shape the events to come. The result is a complex feedback system, where events drive player interpretation, and player interpretation drives events. Attending to this complex feedback system, says Grant Tavinor, will yield some rather exciting results — exciting, at least, for those concerned with the thrilling business of aesthetic ontology and the artistic nature of videogames.

But first: what, exactly, is interactivity? Tavinor offers us an account drawn from Dominic Lopes and Berys Gaut. Interactive works are those in which users are authorized, by the work, to take actions which determine its display features; and in which users then attend to the artistic features of that display. In less technical language: a player’s input helps determine the visuals and audio and the like, and then the player has to look at what the game spit out in response to their input, in order to have their tasty artistic experience. The notion of ‘authorization’ here is crucial. I could modify pretty much any art display; I could cut up the Mona Lisa and arrange the squares by color gradient to my heart’s content, but that is not the prescribed use of the object. In interactive works, the users are permitted, even required, to give input.


cut-scene from Final Fantasy X

Tavinor then singles out a special class of interactive games: interactive narratives, where the player’s actions determine features in the narrative. Not all interactive games are interactive narratives — Bioshock Infinite, for example has plenty of interaction for its action sequences, and player input can change the visual landscape, but the story elements are fixed and outside the player’s control. In fact, any video game where interactive sequences are interrupted with pre-scripted cut-scenes, and where all the narrative elements emerge in those cut-scene, is interactive, but not an interactive narrative.

On the other hand, says Tavinor, take a game like Until Dawn, where player choices deeply inform story events. Who lives and who dies, who succeeds and who fails, and how each character meets their probable untimely end, all emerge from player choice. That is an interactive narrative.


a plot decision in Until Dawn

Tavinor follows Gaut one step further: the player, says Tavinor, is doing something very much like a performer of, say, a musical work. With a classical musical work, there is a general form (“the composition”) and a specific instantiation (“a performance”). A performer must interpret the composition to create a specific instantiation, which is usually the only way the composition can be appreciated. Similarly, a game-player takes a general form (“the game”) and, through a performance, instantiates it as a particular playing. But there’s a crucial difference: in music, the performer and the audience are separate roles. In games, the game-player is both performer and audience. The game-player performs the work and then appreciates the results of their own performance.

Players of interactive narratives can thus be involved in what Tavinor calls an interpretive performance — where a player explores an interactive game as a way to get a grip on the game’s content and meaning. Interpretive performance isn’t the only way to play a game — if I’m playing just to win, and ignoring all the narrative elements, then I’m not involved in an interpretive performance. But if I’m playing for the sake of having an interpretive experience of the narrative elements and their meaning, then I’m involved in an interpretive performance.

This all becomes particularly interesting for Tavinor when we start to have interpretive relationships to the player characters under our control. We don’t have interpretive relationships like this in all games. In chess, for example, I do not try to imagine how my knight feels to be felled by a lowly pawn, nor to figure out the psychological states of the shadowy general-figure above it all. I simply pursue the goals of the game. Many video games suggest a similarly non-interpretive relationship — in Minecraft, for example, we largely occupy the player-character entirely first-personally and the game presents no cues about that character’s mental state. These sorts of player-character, says Tavinor, are merely epistemic and agential shells for the player. I don’t do any interpreting of my player-character at all; what actions I choose for my character are entirely a matter of my own motivations.


the first-personal view in Minecraft

But other games are quite different. Tavinor’s example here is, again, Until Dawn, in which various player-characters often speak and act independently of our control, are often viewed from a third-personal perspective with visible emotional reactions, and then are passed to us for occasional control.

In these cases, says Tavinor, my choice of action can be informed by my interpretation of that player-character. If the game has recently showed me how fearful a character looked, I might choose to have the character run away, not in order to win the game, but simply because it is in-character.


screenshot from Until Dawn

All of this hints at a rather fascinating feedback loop for players enmeshed in any sort of narrative relationship with the game. Players interpret the events of the game, including, perhaps, the player-characters’ motivations. But that interpretation can drive the choices a player makes, which may yield more story elements, which must be interpreted. Crucially, this is not a purely creative act, and is not to be assimilated to, say, telling a story. In most games, Tavinor points out, players do not always know what their choices will do, or what artistic effect it will have, and must learn from the game what story-events their actions have provoked. And past interpretations can drive future interpretations — the fact that I interpreted a character’s running away from a monster as, say, practical and sensible, rather than cowardly, will be added to the nest of cues the game gives me, and my later interpretations will be built on, among other things, my past interpretive contributions.

From this analysis, we get two rather dramatic outcomes, at least for those who care about aesthetic ontology. First, look to the debate about the multiplicity of interpretations. Do narrative works have a single correct interpretation, or multiple? For interactive narratives, the answer is clear, says Tavinor: since the interpretation depends on the narrative, and the narrative isn’t fixed, then the interpretation isn’t fixed. There must be multiple interpretations of a game, since the object of interpretation — the narrative — will vary from playing to playing.

Second: the narrative nature of some video games may supply norms for proper play. Let’s say that some games are meant to be played just for the sake of winning — abstract arcade games like Tetris are a good candidate. These games generate lusory warrant for playing — that is, the warranted actions are those that aim at winning. You’re supposed to try to win the game when you’re playing; otherwise, you haven’t really experienced the game. But, says Tavinor, lusory warrant is not the only kind of warrant. There is also narrative warrant for actions that fulfill some narrative function. Saving a character may help maximize points, but letting them die at a crucial moment may create a more satisfying and poetic narrative arc. Choosing to make my player-character attack, instead, may help me win the game, but may violate interpretation I have of her character. In fact, many games offer both lusory challenges and narrative cues and thus supply both lusory warrant and narrative warrant. Of course, some games may allow, or even force, conflict between these two kinds of warrant; this clash has been called ludonarrative dissonance.


Critical comments

First, a worry: Tavinor claims that there is no single correct interpretation of the work, which is the game. But I worry that this claim is driven by blurring together two distinct entities: the game, and a particular narrative generated by a particular playing of the game. Suppose one wanted to (for some reason) hang on to the view that a work had a single interpretation. I think there is a way to do this, against Tavinor’s analysis.

Tavinor’s argument that a game has multiple interpretations is based in the view that to interpret a game is to interpret a narrative, as it emerges in a playing. Call this the Game as Narrative view. Since the narrative changes from playing to playing, then the interpretation must change. But here is another, quite plausible, view: to interpret a game is to interpret the possibility space of narratives. Call it the Game as Possibility Space view. Tavinor himself flirts with the latter; in some games, he says, we need to play multiple times to interpret the game, to see how the causal networks operate. Lopes holds a similar view about interactive computer art. For Lopes, one can’t appreciate an interactive work by attending to a single set of displays generated by one interaction. Rather, one must interact with the work many times, and use the various different interactions and responses, in order to explore the possibility space, and thus to appreciate the work {Lopes 2010@53-61, 66}. For Lopes, the real object of appreciation isn’t the result of a lone interaction, but, rather, the stable algorithm underlying it all — the different interactions are just a way to come to grips with the algorithm.

Here’s a simplistic example: imagine I am playing a game called Lowlife. In my first play through, I choose to make my character drop out of school and trade drugs; one of my partners kills me, and my character ends up dead in a ditch. I interpret the games’ message as: be careful who your partners are. I then play the game a few more times, and discover that even if avoid taking on partners, once I enter the drug-trade, I end up dead in a ditch. I then interpret the game as saying: don’t deal drugs. But then I play the game more, and make my character stay in school and become a doctor, or a lawyer. But under all these choices, my character ends up dead in a ditch. Once the possibility space has been explored, the real interpretation seems to presents itself: no matter what you do, you’re going to end up abandoned and dead — life is a hopeless and terrible affair. If one held the Game as Possibility Space view, one might plausibly claim, contra Tavinor, that there was a single right interpretation forced by the causal structure of the game, which only came into view after a full exploration of the space.

What’s going on here, I think, is a hesitation between talking about interpreting a particular playing of a game and interpreting the game as a whole. So here’s a path that the defender of single interpretations could take, to resist Tavinor: they could say that, for a single playing, that there was only one narrative, and so only one interpretation. Of course, when we step back and look at the game as a whole, there are multiple narratives — but now the object of interpretation is no longer just one narrative, but the game – the algorithm, the possibility space as a whole. And that possibility space doesn’t change from one playing to another, nor does a player have any control over it. So that could also admit of a single interpretation. Or, to put it another way: the correct interpretation of the game depends on seeing the space of possible narratives, and that space is fixed. Of course, it may be impossible in a human life to see all the possible narratives of a game. Which doesn’t show us that the game has multiple interpretations; we could simply say that the game has one correct interpretation, which is beyond the ken of a single person. Which is also very interesting, but a very different place to end up.

Next, let’s consider ludonarrative dissonance. I find this topic painfully interesting, and here I simply have a number of questions. Tavinor’s point here is that games have different prescriptions for their proper experience, and these prescriptions can conflict. Is this a special property of games, or can other artistic object have such conflicting prescription? And does the conflict in prescriptions depend on the artist object being something of a hybrid of other forms? You might think this was a general feature of any case where two traditional art-forms, with coherent and typically internally consistent prescriptions, first get fused. Perhaps there are other such examples from the early days of, say, opera.

Or, alternately: is ludonarrative dissonance a particular outgrowth of interactivity and emergence? That is, in non-interactive cases, an audience member can’t get that far away from where the artist intended them to be. But because of the chaotic interaction of in-game causal mechanisms, game-players can end up in situations that the designer never imagined. Even if the game designer has tried to rid the game of ludonarrative dissonance, the unpredictability of an interactive narrative might create entirely new narrative possibilities that the designer had not foreseen. Accidental dissonance is more likely, because interactive narratives are harder to control.


Grant Tavinor’s Response

First, thanks to Thi for his comments and for finding these issue as interesting as I do!

On the first issue of the correctness of interpretations, one clearly has to be very careful in delineating playings and the games that are played (or works and displays of those works, to use a different terminology). I had hoped to make it clear that I view playings as displays of works, and as being one of the two objects of appreciation along with the games/works themselves (apologies if it wasn’t clear!). Hence I am generally sympathetic with the idea that the work in these cases is something like a “possibility space” explored through the various playings or displays of the game, as Thi suggests.

I find Thi’s comments about the mixed appreciative prescriptions of videogames also very interesting. The comparison to opera is apt; there, the initial combination of art forms with quite different modes of appreciation produced a very odd beast with its sometimes awkward mix of drama and music, particularly in the contrast of recitative and aria. Nevertheless, in Wagner’s conception of the Gesamtkunstwerk and especially in the Ring Cycle, this normative mix was embraced absolutely, and to extraordinary effect.


Niko Bellic, from Grand Theft Auto IV

The Grand Theft Auto series of games is similarly overstuffed work, in which different appreciative modes sit very closely together, and in which these mixed normative modes similarly seem in tension. In Grand Theft Auto 4, for example, the unrestrained violence of the player-controlled Niko, may clash with the reluctant criminal Niko of the narrative. It also leads to a very long and uneven narrative where Niko needs to play diverse narrative roles to fit with the diverse gameplay. This is perhaps a cost of combining ludic and narrative functions into a single player controlled character, and probably counts as another case of ludonarrative dissonance.

I’ve argued elsewhere that these contradictions often provide the substance for creative acts, constituting design problems that invite creative solutions within a tradition of creativity. The most recent Grand Theft Auto game provides three characters for the player to adopt during gameplay. This strategy allows the game to reconcile some of the normative tensions in the earlier game. It allows the game to partition some of the exaggerated violence typical of the franchise’s gameplay, by having it performed through the guise of the obviously psychotic Trevor. But is also enables the game to explore different kinds of stories through the other characters. Contrasting Trevor’s general insanity in the game, Michael’s story is that of a reformed mobster desperately trying to avoid relapse into his criminal ways. As such his typical early gameplay involves domestic activities such as yoga and managing his family conflicts. And of course, this multiple player-character strategy also opens up the interesting narrative and ludic dynamic of being able to switch between characters during a game and story.

Perhaps there is a deeper contradiction here, however. I know that Brock Rough has a very interesting paper where he contends that games, ontologically, cannot be artworks partly because of how the normative aspects of both would have to interact in such a case.¹ There is clearly much more to be said and argued about here.

¹[ed. note: Brock Rough’s paper is “Videogames as Neither Video nor Games: A Negative Ontology”, in The Aesthetics of Videogames, eds. Grant Tavinor and Jon Robson, Routledge, forthcoming; see a prior AFB post by Brock here]



  1. seems like the link to Tavinor’s JAAC article doesn’t work…

  2. Really cool paper, and neat discussion!

    I’ve got a general worry about the notion of interactivity employed, so maybe people here can help me figure out if I’m just barking up the wrong tree.

    Specifically, it looks to me like this notion of interactivity is too permissive, in that it entails that paradigmatic non-interactive media (e.g. novels) are in fact interactive. Proper engagement with a given novel requires that grasping what is true in the associated fiction. Obviously, this involves sussing out the explicit, primary fictional truth. But it also requires that we import certain things into the fiction, then, linking these up with the explicit fictional truths, our performing certain inferences in order to ‘generate’ implied (secondary) fictional truths. To me, this sounds an awful like a prescription to attend to the artistic features of the display (the explicit, primary fictional content), to provide input (the imported truths, the relevant bit of thinking), and, in so doing, change the display (the overall content of the fiction). But that would mean the relevant novel is interactive. Of course, novels are just an example – we could run the same point with e.g. films, or comics – anything that’s meant to be non-interactive. Basically, anytime proper appreciation of the work requires filling out the fictional content by import and inference, the work will turn out to be ‘interactive’.

    • Thanks for the comment. I guess that the defender of this view of interactivity would say that in your example the display is not really changed by the interpretative activities. What your example illustrates is that art displays/texts/vehicles typically under-determine interpretations, because the work is ambiguous in various respects. Interactivity in this view requires something more, i.e. that the display itself is undetermined before the appreciator gets to it. Interpreting a novel does not change the text itself, but playing a game generates one of a number of possible displays. And as you point out of art objects generally, this display itself needs elaboration before it is properly realized in an appreciation.

      • I should add that Arron Smuts has a 2009 paper in the Journal of Aesthetic Education where I recall him making a similar argument that this theory of interactivity is too permissive.

Leave a Reply

Required fields are marked *.