Posted on March 25, 2020 at 5:00 AM
This essay was the winning entry in the undergraduate category of the 6th Annual Oxford Uehiro Prize in Practical Ethics.
Written by University of Oxford student, Eric Sheng.
In the computer game Red Dead Redemption 2 (henceforward, RDR2), players control a character in a virtual world. Among the characters represented by computer graphics but not controlled by a real-world player are suffragettes. Controversy arose when it became known that some players used their characters to torture or kill suffragettes. (One player’s character, for example, feeds a suffragette to an alligator.) In this essay, I seek to explain the moral disquiet – the intuition that things are awry from the moral perspective – that the players’ actions (call them, for short, ‘assaulting suffragettes’) provoke. The explanation will be an exercise in ‘moral psychology, philosophical not psychological’: I seek not to causally explain our disquiet through the science of human nature, but to explain why things are indeed awry, and thus justify our disquiet.
My intention in posing the question in this way is to leave open the possibilities that our disquiet is justified although the players’ actions are not wrong, or that it’s justified but not principally by the wrongness of the players’ actions. These possibilities are neglected by previous discussions of virtual wrongdoing that ask: is this or that kind of virtual wrongdoing wrong? Indeed, I argue that some common arguments for the wrongness of virtual wrongdoing do not succeed in explaining our disquiet, and sketch a more plausible account of why virtual wrongdoing is morally disquieting insofar as it is, which invokes not the wrongness of the players’ actions but what these actions reveal about the players. By ‘virtual wrongdoing’ I mean an action by a player in the real world that intentionally brings about an action φV by a character in a virtual world V such that φV is wrong-in-V; and the criteria for evaluating an action’s wrongness-in-V are the same as those for evaluating an action’s wrongness in the real world.
A common kind of argument that certain instances of virtual wrongdoing are wrong invokes the real-world consequences of those kinds of virtual wrongdoing. It might be argued, in the case under consideration, that assaulting suffragettes in RDR2 has such psychological effects on the player that he becomes more likely to harm women in the real world.
One difficulty that this kind of argument faces is empirical – there is little conclusive evidence that playing video games affects real-world behaviour in the way that it posits. Indeed, a recent study finds that those who play violent video games do not subsequently significantly differ from those who haven’t in a range of variables, including aggression and sexist attitudes.
A second difficulty is that raising the risk of harm (or wrong) isn’t sufficient for wrongness. Consider cases in which the risk-raising activity isn’t forbidden by any principle that no one can reasonably reject. Building a new railway line, for example, raises the risk of train accidents. But if, because overall it benefits everyone ex ante through facilitation trade, it isn’t forbidden by any principle that no one can reasonably reject, then it doesn’t seem to be wrong. Consider also cases where an action isn’t wrong even though it is forbidden by some principle that no one can reasonably reject. One such counterexample, perhaps, is McCormick’s example of someone who, by attending a football game and thus marginally increasing the audience’s size and thereby its level of arousal, raises the risk of vandalism of the property of neighbouring residents. If attendance at the game doesn’t, even ex ante, benefit some of these residents, perhaps it is forbidden by a principle no one can reasonably reject. Nonetheless, as McCormick claims, we don’t think attending the game is wrong, perhaps because the expected value of the harm to others is very small in proportion to that of the attendees’ pleasure. Those who argue that certain instances of virtual wrongdoing are wrong in virtue of raising the risk of harm must, therefore, show that those instances don’t fall in either category of counterexample – that certain instances of virtual wrongdoing don’t overall benefit others ex ante and that, even if they do raise the risk of harm to others, the risk to others isn’t small enough and the pleasure derived by the players great enough for the latter to morally outweigh the former. In the case of RDR2, these claims are far from obvious.
Even, however, if some consequential argument does show that certain kinds of virtual wrongdoing are wrong, it doesn’t succeed in justifying our disquiet about assaulting suffragettes. For one thing, our disquiet in this case is considerable, whereas the considerations adduced above suggest that the wrongness of the action is marginal. This is not decisive, although it does suggest that our intuitive reaction is at least miscalibrated. More to the point, our disquiet seems to remain even when consequences are bracketed. Suppose that some player undergoes a procedure after each time he plays RDR2 whereby his mind is restored to its antecedent state. Nonetheless, confronted with these players assaulting suffragettes in RDR2, many of us (I suspect) would still feel considerable disquiet.
Another prima facie attractive explanation invokes the notion of ‘dehumanization’ or ‘objectification’. At the time of the RDR2 controversy, for instance, a newspaper columnist wrote that RDR2 was failing ‘to enfranchise [male players] in notions and skills of talking to women as people, not objects’. It might be argued, then, that assaulting suffragettes in RDR2 is wrong because dehumanization is wrong and assaulting suffragettes in RDR2 is dehumanizing.
To dehumanize or objectify, however, is to treat a human as merely an object, but no human is thus treated when a player assaults a suffragette in RDR2, the suffragette being a fictional character, not a human. It might be objected that, while no real human is dehumanized, in the world of RDR2 a human being is objectified. But this is to rest the argument on an equivocation. Given that dehumanizing a human in the real(/fictional) world is wrong in the real(/fictional) world, whilst if it were true that assaulting suffragettes in RDR2 dehumanizes a human in the real world, doing so would be wrong in the real world, that isn’t true; and it being true that assaulting suffragettes in RDR2 dehumanizes a human in the world of RDR2 yields only that assaulting suffragettes in RDR2 is wrong in the world of RDR2.
Indeed, we can readily think of cases where a player’s character treats a suffragette, in the world of RDR2, in a way inconsistent with her being human, but the player is not guilty of morally objectionable dehumanization. Suppose a game developer, wishing to verify that the alligator in it gobbles things up as programmed, plays as a character who successively feeds every object and character (including the suffragette) in the game to the alligator. We don’t think she does anything wrong in the real world, even though she intentionally treats the suffragette as not a real human.
Suppose Sam shouts racist slogans when alone in her room without resulting in bad consequences. This action may be wrong on some accounts of right action, but its wrongness doesn’t seem to explain the disquiet we feel concerning it. Indeed, suppose Sam isn’t morally responsible for having developed racist attitudes; knows there’s a risk she’ll harm others because of having these attitudes; and shouts racist slogans in order, by venting these attitudes in private, to avoid harming others. Then, her action might not be wrong – but is nonetheless morally disquieting. One reason for this is that the action presumptively reveals that she has explicitly racist attitudes; and a state of affairs in which someone has explicitly racist attitudes is, pro tanto, morally bad. (By ‘X presumptively reveals that Y’, I mean: if one knows X, one can reasonably presume that it is likely that Y.) The presumption is defeasible. If one knows that Joe shouts racist slogans when he is alone in his room but also that he is an actor practising his lines for a film in which he plays a racist, it isn’t reasonable to presume that it is likely that he has explicitly racist attitudes. Indeed, in this case we would not, if we knew the relevant circumstances, feel disquiet.
The disquiet we feel concerning assaulting suffragettes RDR2 is, I hypothesise, explained by similar considerations as those that explain the disquiet we feel concerning privately shouting racist slogans – what explains our disquiet isn’t the wrongness of the action but what it presumptively reveals about the player. At least three considerations favour this hypothesis.
Firstly, whereas it’s implausible that the player dehumanizes anyone and at best unclear whether he impermissibly raises the risk of harm to others, it’s plausible that knowing that he assaults suffragettes in RDR2 permits one to reasonably presume that he has morally troubling desires concerning certain women. To motivate this claim, consider the question: why (i.e., for what [explanatory] reason) does the player assault suffragettes? Characters in fictional worlds might have reasons to do certain things: e.g., eat food to survive or to complete a mission set by the game. But a character in the world of RDR2 doesn’t have any such reason to assault suffragettes. So the desire of the desire-belief combination that explains the player’s action isn’t a desire that his character do what it has reason to do in the game-world. Rather, in virtual reality worlds like that of RDR2, there is characteristically an identification between the player and his character such that the player uses the character to do, in the virtual world, what he in some respect desires be done in the real world – many players, Marya Schechtman notes, ‘use a fictional context to give voice to a part of [their] personality, a trait, or set of desires’. I say ‘in some respect’, because the player may not, all things considered, be disposed to act to bring about something that he in some respect desires – as Schechtman adds, the trait or desire expressed virtually is often one that is ‘for one reason or another suppressed in real life’. What’s disquieting, then, about assaulting suffragettes in RDR2 is that the action presumptively reveals that the player in some respect desires, in the real world, that corresponding persons be similarly treated in the real world (a state of affairs where someone has this desire being, pro tanto, bad).
Secondly, this account allows us to explain why some kinds of virtual wrongdoing don’t cause disquiet. For instance, in strategy games players often use the countries they play as to illegally invade other countries. We are rarely disquieted by such actions, my account suggests, because usually, in such games, players merely act to bring about what their character has reason to do, given the goals and rules of the game, in the game-world. We no more think their action reveals a desire to invade other countries than we do that taking a knight in chess reveals a desire to kill horses.
Thirdly, my account makes sense of the intuition that in the case of virtual (but not of real) wrongdoing, numbers don’t count: whereas a state of affairs in which someone assaults twenty women in the real world is worse than one in which he assaults ten, a state of affairs in which someone assaults twenty suffragettes in RDR2, it seems, is not worse than one in which he assaults ten. This intuition would be difficult to explain if each action of assaulting a suffragette in RDR2 were wrong, even wrongness asymptotically decreases with the number of previous instances of the action-type committed; on my account, it is natural because (at least after enough instances of the action-type have occurred to make it unlikely that the assault of suffragettes is unintentional) more instances of the action-type don’t reveal something worse about the player than fewer.
An action being wrong isn’t necessary for our feeling moral disquiet about it, or for this disquiet to be justified. Assaulting suffragettes in RDR2 is morally disquieting, I suggest, because of what it reveals about the player who does so, not because it is wrong.
 John Rawls, Political Liberalism (New York, 1993), p. 86.
 I set aside virtual worlds shared by characters controlled by more than one player and the wrong that one player might do another through bringing about virtual actions in such worlds.
 Simone Kühn, Dimitrij Tycho Kugler, Katharina Schmalen, Markus Weichenberger, Charlotte Witt and Jürgen Gallinat, ‘Does playing violent video games cause aggression? A longitudinal intervention study’, Molecular Psychiatry, 24 (2019), pp. 1220–1234
 Such counterexamples exist only if contractualism is incorrect.
 Matt McCormick, ‘Is it wrong to play violent video games?’, Ethics and Information Technology 3 (2011), pp. 277–287: 280.
 Perhaps such instances are a safety valve for emotions that might otherwise be dangerously expressed.
 Van Badham, ‘Red Dead Redemption 2: calls to ban violence against women in games are too simplistic’, The Guardian, 8 November 2018, theguardian.com/games/2018/nov/08/red-dead-redemption-2-calls-to-ban-violence-against-women-in-games-are-too-simplistic.
 Marya Schechtman, ‘The story of my (second) life: virtual worlds and narrative identity’, Philosophy & Technology 25 (2012), pp. 329–343: 332.
 I leave it open that on some accounts of wrong action, perhaps virtue-theoretic ones, revealing a bad desire in this way also makes an action wrong.