Introducing the dilemma

The Gamer’s Dilemma (Luck, 2009a) is a paradox concerning the moral permissibility (henceforth permissibility) of two acts we might perform when playing computer games. The first is virtual murder; where a player intentionally causes their character to murder another in a computer game.Footnote 1 Virtual murder is commonplace,Footnote 2 but is it morally permissible? Perhaps it is, as no one is actually murdered,Footnote 3 it’s “just a game”.Footnote 4

However, the same can be said of a less commonplace act – virtual child molestation; where a player directs an adult character to molest a child character in a computer game. Again, no one is actually molested, it’s “just a game”.

Some may balk at the possibility that if virtual murder is permissible, so too is virtual child molestation. Yet it has proven difficult to find a relevant difference between these acts. This puzzle can be captured in the form of a paradox.

P1 - Virtual murder is permissible.

P2 - There is no relevant difference between virtual murder and virtual child molestation, in respect to being permissible.

P3 - Virtual child molestation is impermissible.

This is the Gamer’s dilemma in its general form; a set of jointly incompatible propositions that all seem true (or at least no single proposition seems obviously false).

However, in this form, the dilemma might apply too generally. For there may be fringe cases where some of the propositions do seem false (Bartel, 2012; Ali, 2015; Luck, 2022; Montefiore & Formosa, 2022). I’ve elsewhere (Luck, 2022, 1292) given the example of someone working for a government classification board. Such a person might permissibly engage in virtual child molestation, as it is their duty to play such games (to recommend an appropriate rating). In this case, P3 seems false. Given this, the dilemma may need to adopt a less general form.

P1 - X cases of virtual murder are permissible.

P - There is no relevant difference between X cases of virtual murder and Y cases of virtual child molestation, in respect to being permissible.

P3 - Y cases of virtual child molestation are impermissible.

Where X and Y refer to the types of cases the dilemma should be narrowed to. For example, they could be narrowed to cases where the player is treating the wrongdoing lightly (Luck, 2022), which presumably the ratings official isn’t. (These variables may also serve to help distinguish between different versions of the dilemma (Kjeldgaard-Christiansen, 2020; Ali, 2022) and different types of virtual acts (Öhman, 2020; Ramirez, 2020).)

In addition to narrowing the dilemma, it can also (in another respect) be widened; with the Gamer’s Dilemma possibly being an instance of a wider dilemma (Luck, 2022). To support this possibility, consider the following two examples.

Within some computer games, such as Skyrim, players can take very valuable items from innocent characters in circumstances such, were the game world actual, it would be stealing of a particularly egregious kind. Again, few eyes blink. However, if a game were introduced that enabled players to shout homophobic insults at gay characters, we would be less dismissive. (Luck, 2022, 1299)

This first example suggests the dilemma might still arise were we to replace virtual murder and child molestation with other virtual wrongdoings. And perhaps the dilemma might not even be limited to video games.

Why it is that we are happy to knowingly sing merrily along to The Beatles song “Maxwell’s Silver Hammer” (which is about a serial killer bludgeoning various people to death), but would be warier of singing merrily along to a song cheerfully describing the exploits of a child molester? (Luck, 2022, 1297)

So, a similar puzzle seems to arise if we replace playing a video game with different ways of engaging with fictional wrongdoings.

If these examples are all instances of the same puzzle, then the Gamer’s Dilemma may be an instance of a wider dilemma; a dilemma with the following schema.

P1 - X cases of engaging with fictional wrongdoing A are permissible.

P2 - There is no relevant difference between X cases of engaging with fictional wrongdoing A and Y cases of engaging with fictional wrongdoing B, in respect to being permissible.

P3 - Y cases of engaging with fictional wrongdoing B are impermissible.

This wider dilemma encompasses not only different types of wrongdoing (e.g. stealing and homophobic insults), but also different ways such fictionalized wrongdoings might be engaged with (e.g. singing songs about them, and virtually enacting them). In this paper we will focus on the Gamer’s Dilemma, rather than this wider dilemma. However, if the Gamer’s Dilemma is an instance of this wider dilemma, then the arguments offered here (suitably generalized) may apply more broadly.

Within the literature, an argument against P2 is referred to as a resolution to the dilemma. Resolutions are sought by those who hold that virtual murder may be permissible, but not virtual child molestation (Bartel, 2012; Luck & Ellerby, 2013; Patridge, 2013; Young, 2016; Nader, 2020; Kjeldgaard-Christiansen, 2020; and Milne & Ivankovic, 2021). Alternatively, some have offered dissolutions to the dilemma by arguing against P1 or against P3 (Ali, 2015; Luck, 2018; Ramirez, 2020). Dissolutions are sought by those who hold that either virtual murder is impermissible, or that virtual child molestation is permissible (but not both, otherwise this would create an equally puzzling inverse Gamer’s Dilemma).

Resolutions and dissolutions both constitute, perhaps confusingly, solutions to the dilemma; as they aim to tell us which of the paradox’s propositions are false. However, more recently, a third approach has emerged, which does not constitute a solution – resistance to the dilemma (Montefiore & Formosa, 2022; Formosa et al., 2023). This approach doesn’t strictly offer an argument against the truth of P1 or P3, but instead seeks to undermine the intuitions supporting P1 or P3.

It may not be immediately clear what the distinction is between dissolving and resisting the dilemma. For example, you might think if the intuitions supporting P1 are undermined this would constitute a dissolution to the dilemma, for we could then reject P1. But this move is too quick. For, although it is the case that, if P1 isn’t true then P1 it must be false, it is not the case that, if you don’t have the intuition that P1 is true, then you have the intuition that P1 is false (or that P1 is actually false). For example, I don’t have the intuition that you are hungry right now, but that doesn’t mean I have the intuition that you aren’t – my gut is telling me nothing about yours. So, undermining an intuition that something is true, doesn’t necessarily result in the intuition that it is false, let alone constitute a reason to think it is. However, as I will go on to argue, a successful resistance to the dilemma may (a) diminish the force of the dilemma, and (b) help lay the foundations for certain types of dissolutions.

Resisting the dilemma

Virtual murder is more commonplace than virtual child molestation. Why is this? Perhaps because we intuit P1 and P3 to be true.Footnote 5 In other words, the intuition that virtual murder is permissible, but not virtual child molestation, explains why we are more comfortable playing games where we can virtually murder, than where we can virtually molest.

But what if this is wrong, and we in fact don’t intuit P1 or P3 to be true at all. How would this affect the Gamer’s Dilemma? Some have argued that the dilemma “relies for its argumentative force on the claim that there is an intuitive moral difference between these acts” (Formosa et al., 2023, 1). If this is true then, were the intuitions supporting P1 or P3 undermined, then so too would the force of the dilemma. But what exactly does this mean?

To help us think more about the force of the dilemma, consider the following set of propositions.

  1. 1.

    Abortion is permissible.

  2. 2.

    Abortion is not permissible.

This is a set of inconsistent propositions, and I imagine you intuit one of these propositions to be true, but not the other. So, considering this set won’t reveal to you that your intuitions support inconsistent propositions (because they don’t).

Now consider a different set of inconsistent propositions,

  1. 1.

    Humanely killing cows for food is permissible.

  2. 2.

    There is no relevant difference between humanely killing cows for food and humanely killing dogs for food, in respect to being permissible.

  3. 3.

    Humanely killing dogs for food is impermissible.

Imagine that you currently intuit all three of these propositions to be true (perhaps you’ve just never put the pieces together before). Considering this set should help you see that your intuitions support inconsistent propositions; that at least one of your intuitions is unreliable (the one supporting the false proposition/s). In this way we can think of this latter set as having more revelatory force (for you) than the former set on abortion.

Now suppose that you discover that you actually don’t properly hold the intuition that killing dogs for food is impermissible. Perhaps you find out that what you mistook as your own intuition against dog killing was actually artificially induced by a microchip implanted in your brain.Footnote 6 Discovering this should reduce this set’s revelatory force; as it no longer reveals that at least one of your intuitions is unreliable. For there is no issue inuiting 1 and 2 (but not 3) to be true.

(I should quickly foreshadow that even if an inconsistent set doesn’t have any revelatory force – it may still be worth providing an argument to show which proposition is false. For example, I’m guessing the set on abortion has no revelatory force (for anyone). But it may still be important to provide an argument for, or against its permissibility.)

So, one motivation for resisting the Gamer’s Dilemma (by demonstrating we actually don’t intuit P1 or P3 to be true) is to reduce the dilemma’s revelatory force. To show that it doesn’t actually reveal that at least one of our intuitions is unreliable. With this in mind, let us examine two recent attempts to provide such resistance; each of which undermines the intuitive support for P1 in a different way.

We don’t intuit P1 to be true

Formosa et al. (2023) conducted an empirical study that could suggest most people don’t intuit P1 to be true. Put briefly, this study involved asking participants to consider various vignettes containing instances of virtual murder and virtual child molesation; after which their moral appraisal of these vignettes were recorded and analyzed. They found that,

…participants viewed virtual molestation as more morally unacceptable than virtual murder for all our conditions, however they also did not view virtual murder as morally unproblematic. (16)

This result seems compatible with (at least) the following two possibilities:

  • People intuit P3 to be true, but not P1 (as they inuit both actions to be impermissible, but virtual child molestation to be morally worse).

  • People intuit P1 and P3 to be true (as they intuit both actions to be morally wrong to some extent,Footnote 7 but only virtual child molestation to be sufficiently wrong to be impermissible).

The first possibility constitutes a resistance to the dilemma – for if most people don’t inuit P1 to be true, then the dilemma no longer reveals (for those people) that at least one of their intuitions is unreliable. In this manner the dilemma’s force is diminished.

Now consider a second way the dilemma might be resisted.

We do intuit P1 to be true, but the intuition is unreliable

Montefiore and Formosa (2022) argue that we do inuit P1 to be true, but these intuitions aren’t properly grounded. They point out that computer games often involve virtual justified killing – gamers playing “goodies” who are permissibly killing “baddies”. From such a norm a convention has arisen. And as a result,

…the conventional permissibility of justified killing in games may bleed into, and cause, gamers to construe many, if not most, instances of unjustified killing as being (at least potentially) conventionally permissible. (Montefiore & Formosa, 2022, 10)

The thought here is that the intuitions supporting P1 aren’t actually grounded by a sense of morality, but instead by a sense of convention. So, our intuitions are confused – they are latching on to the wrong thing.Footnote 8

If this is correct then it is no longer surprising that one of our intuitions is unreliable – we don’t need the dilemma to reveal this anymore. For the fact that P1 is grounded by social conventions, and social conventions don’t reliably track moral truths, already reveals this. In this manner, the force of the dilemma is again diminished.

These are two different ways that the intuitive weight of P1 might be undermined; either because we don’t actually hold this intuition in the first place, or we do, but it’s not a reliable indication of the truth of P1 (because it’s grounded by social conventions).Footnote 9 Let us now consider what follows from such resistance.

What follows from resisting the dilemma?

There are (at least) two different questions raised by the Gamer’s Dilemma. The first is,

Q1 - Why do we intuit virtual murder to be permissible, but not virtual child molestation?

What we are looking for here is a psychological explanation for seemingly conflicting intuitions. To answer this question, as Ekdahl (2023) states, “we need to turn an empirical eye towards understanding users’ video game intuitions” (61).

For those searching for such an explanation, resistance to the dilemma may prove very helpful. For one reply to Q1 is that we don’t in fact hold these intuitions (Formosa et al., 2023). In other words, the question itself is misguided. Another reply might be that, although we do hold conflicting intuitions, this is hardly surprising as they are grounded conventions, (Montefiore & Formosa, 2022), which can be inconsistent for a whole host of reasons.Footnote 10 Another reply might be that we hold conflicting intuitions because it is typically easier to think of occasions where killingFootnote 11 is permissible (e.g. in self-defense) than it is to think of cases where sexual-assault is permissible (Whitty, Young & Goodlings, 2011; Young & Whitty, 2011). Another reply might be that, typically “virtual pedophilia implies antisocial desire whereas virtual murder does not” (Kjeldgaard-Christiansen, 2020, 96). Another reply might be that we find virtual molestation to be disgusting, but not virtual murder (Coghlan & Cox, 2023). Another reply might be that, typically child molestation is an end unto itself, whereas murder is a means to an end (Ali, 2015). All such replies may suggest (at least on the face of it) that moral intuitions may not be playing as central a role in determining what we find permissible as we might otherwise presume.

However, the dilemma also raises a second question.

Q2 - Are these acts relevantly different in respect to their permissibility?

What Q2 demands is a solution (rather than an explanation). That is, we need an argument for why these acts are relevantly different (i.e. for why P2 is false – a resolution), or for why they aren’t (i.e. for why either P1 or P3 is false [but not both] – a dissolution). In other words, we are looking for an argument for the falsity of one of the dilemma’s propositions.

And, one might pursue such an argument even if the dilemma’s revelatory force has been eliminated. For instance, even if you don’t hold the intuition that virtual murder is permissible, it might still be worthwhile establishing an argument against P1. I alluded to this point earlier, when considering the following inconsistent set,

  1. 1.

    Abortion is permissible.

  2. 2.

    Abortion is not permissible.

Even though this set has no revelatory forceFootnote 12 it may still be worth providing an argument for, or against, the permissibility of abortion. Why? Perhaps, to engage in reflective equilibrium, to convince someone else of your position, or because a lot rides on which proposition is true, etc. The same may be true of providing an argument for, or against, P1, P2 or P3 of the Gamer’s Dilemma.

On first glance, addressing Q2 doesn’t require an empirical study of intuitions. As Formosa et al. (2023) themselves state,

…our study cannot solve or resolve the Gamer’s Dilemma, since while empirical claims about the intuitions of gamers regarding virtual actions are prominent in this literature, no survey can by itself conclusively settle the philosophical issues in dispute. (16)

But is this actually the case? I’m not so sure. I think there may well be possible routes from a resistance to the dilemma to a solution; although such routes might rely on some controversial moves. Let’s explore two such possibilities next.

From resistance to dissolution

Ethical intuitionists hold that moral intuitions can be a reliable source of moral wisdom. On this view, moral intuitions have independent credibility (Hooker, 2000, 12), such that a moral claim can be justifiably held to be true if it matches one’s intuitions. Yet intuitionists need not hold that such intuitions are indefeasible (Shafer-Landau 2003; Dancy, 2006; Audi, 2008). Rather moral intuitions can, and should, be undermined by relevant counter-considerations. But if moral intuitions (unless undermined) can reliably indicate the truth of moral claims, then it may be possible for resistance to the Gamer’s Dilemma to lead to its dissolution. This is because, by undermining the intuitive support of one of its propositions, we may be able to establish reason to think it is false.

Let us assume that we have successfully resisted the dilemma by undermining the intuitions supporting P1 (but we still intuit P2 and P3 to be true). Given the absence of additional reasons to think any of the propositions are true, or false, we may now have reason to think P1 is false. This is because we can mount the following type of argument,

  1. 1.

    We only intuit P2 and P3 to be true. [Given the resistance to P1.]

  2. 2.

    If we intuit Px to be true then we have reason to think Px is true.

  3. 3.

    If we only have reason to think P2 and P3 are true, then we have reason to think P1 is false.

So,

  1. 4.

    We have reason to think P1 is false.

3 holds because the Gamer’s Dilemma is an inconsistent set – in which case if we only have reason to think P2 and P3 are true, then by elimination we have reason to think P1 is false. So, in this manner we can move from undermining the intuitions supporting P1, to having reason to think P1 is false. That is, to move from resistance to dissolution.

However, we should note our “reason to think” in 4 (for the falsity of P1) is only as strong as our “reason to think” in 2 (for intuitions being a reliable indicator of truth). And much has been said regarding the reliability of such intuitions to at least be cautious of such strength (Folke, 2008; Sinnott-Armstrong, 2006, 2008; Stratton-Lake, 2020); with a number of empirical studies suggesting our intuitions may be affected by potentially arbitrary variables such as a person’s culture (Haidt et al., 1993; Nichols et al., 2003; Machery et al., 2004; Weinberg et al., 2001) or the order in which we think about things (Petrinovich & O’Neill, 1996; Swain et al., 2008). In defense of their reliability, some argue that such studies do not properly distinguish between the intuitions of the general population and expert groupsFootnote 13 (Hales, 2009). But it may be that even expert groups are similarly affected by such arbitrary variables (Weinberg et al. 2013; Vaesen et al., 2013; Horvath & Wiegmann, 2022).

A second note of caution should be raised regarding the type of resistance employed to undermine P1. In particular, it can’t be allowed to do too much work for this type of argument to work. For example, let us accept that Montefiore and Formosa (2022) have successfully undermined the intuitions supporting P1 (by arguing that the conventional permissibility of justified killings has bled into virtual murder). If we stop here, then the above argument could be used to suggest P1 is false. However, Montefiore & Formosa might be arguing that the intuitions supporting both P1 and P3 are conventionally grounded.

The intuitions that ground the narrow Gamer’s Dilemma must therefore be situated in the context of the development of conventional norms of appropriate videogame play that have developed over time around killing (or murdering) in games, and the very different history of conventional norms that have emerged around the enactment of sexual violence and child molestation in videogames. (31)

If the intuitions supporting P1 and P3 are both grounded in convention, then they would both be undermined. In which case we would no longer be able to use this argument to dissolve the dilemma, as we could no longer employ the process of elimination in premise 3Footnote 14 (at least insofar as a dissolution constitutes an argument against P1, or an argument against P3, rather than an argument against P1 or P3). However, with P2 left standing, we could at least be assured that the dilemma is to be solved by finding dissolution rather than a resolution.Footnote 15

From resistance to resolution

Young (2015, 2016, 2017, 2024) proposes that what makes an action morally wrong is a particular shared attitude towards it; a view he refers to as constructive ecumenical expressivism, or CEE. Under CEE the claim “virtual child molestion is impermissible” would be understood as something like “there is a shared negative moral attitude to virtual child molestion”.

CEE postulates the following: where a shared moral attitude occurs with regard to some object or event…even if different people have different reasons for this shared attitude, as a society we are able to create or construct a social norm that then acquires its own objectified moral standard. (2015, 215)

Put simply, if a sufficient number of people within a social group believe that virtual child molestation realizes a property (or properties) that they disapprove of (to a sufficient extent), then this act would be impermissible relative to that group. If this is correct, we may be able to move from a resistance to the dilemma to a resolution.

As discussed, Montefiore and Formosa (2022) suggest that most people (presumably a sufficient number of them) hold virtual murder to be permissible, but not virtual child molestation, as their intuitions are grounded by social conventions. This explains why we hold seemingly conflicting stances towards these acts – as conventions can conflict for a myriad of reasons (in this case because of the history of computer games).

According to CEE, provided a sufficient number of people within a group believe there is something about virtual molestation to disapprove of, then “within that socially constructed space…a social norm is established and objectified” (2017, 319). So, if Montefiore and Formosa (2022) are correct, and we find that the intuitions supporting these beliefs are grounded by widely held social conventions, then this may be sufficient to make virtual molestation actually morally impermissible. For morality, according to CEE, emerges from social norms and conventions. In other words, we may be able to mount the following argument.

  1. 1.

    There is a widely held social convention (within group X) that there is something about virtual child molestation to disapprove of.Footnote 16

  2. 2.

    If there is a widely held social convention (within group X) that there is something about an action to disapprove of, then this action is impermissible (relative to group X).

So,

  1. 3.

    Virtual child molestation is impermissible (relative to group X).

Where Montefiore & Formosa’s work supports 1, and Young’s work supports 2. And a similar argument could also be mounted for the permissibility of virtual murder.

From such arguments a resolution to the Gamer’s Dilemma arises. For, as Young (2016, ch.6) argues, there is now a relevant difference between these acts. Namely, that there are social conventions in place that make virtual child molestation impermissible, but no such conventions in place to make virtual murder impermissible.

However, employing such a resolution may require us to accept something akin to moral relativism; which will be unpalatable to some. This is because the moral permissibility of actions under CEE are ultimately determined by the norms of social groups. For example, if the permissibility of slavery were supported by widely held social conventions within a social group, then this practice would be permissible relative to that group. To be clear, it is not that slavery would merely seem permissible to the group, but it would be permissible relative to that group.Footnote 17

It should be noted that CEE is a robust form of moral relativism, insofar as it may be able to “withstand capricious changes to norms” (Young, 2024). This is because, according to Young, the shared attitude a group has regarding an action will typically be supported by a variety of reasons. In which case, the attitude is more likely to remain stable. For even if some of these reasons were undermined, providing a sufficient number of them remained, so too would the shared attitude. However, it is unlikely that those who hold that slavery is never permissible (at least regardless of any group’s attitude) will find much solace in the robustness of a possible moral norm to the contrary.

It should also be quickly noted that if this argument works, then Montefiore & Formosa’s work (2022) may not technically be resisting the dilemma. For if widely held social conventions “objectify” moral claims, then the discovery that our intuitions for a moral claim are grounded in such conventions no longer undermines the reliability of such intuitions (in fact it does the opposite).

Conclusion

The Gamer’s Dilemma raises at least two questions. The first question (Q1) is: why do we intuit virtual murder to be permissible, but not virtual child molestation? In other words, what explains these seemingly conflicting intuitions? Resistance to the dilemma may help us address this question. For example, perhaps social conventions explain our conflicting intuitions, or perhaps the question itself is misguided as we don’t in fact hold such intuitions.

However, the dilemma also raises a second question (Q2): are these acts relevantly different in respect to their permissibility? To answer this question we need a solution? Resistance to the dilemma seems less helpful here. However, it may be possible to move from a resistance to a solution – with two such possible routes outlined here. Yet these routes are controversial. For the first requires moral intuitions to be reliable and the second requires a form of moral relativism. Yet there may be less controversial routes from a resolution to an answer to Q2 (which I will leave to others to map out).

The focus here has been on determining how resisting the Gamer’s Dilemma might help us address two of the central questions it raises. However, these points may also generalize to other empirical approaches to the dilemma. That is, if we are seeking to answer Q1 then we should be “embracing empirically, subject-oriented informed approaches to the Gamer’s Dilemma” (Ekdah, 2023, 61). However, if we are seeking to answer Q2, then perhaps (without some controversial moves) such approaches cannot “conclusively settle the philosophical issues in dispute” (Formosa et al., 2023, 16).Footnote 18