I have opined before about the imbalance between theoretical models and empirical data in the area of ethical gameplay. The past decade has produced a number of analyses of ethically significant commercial games, leading to several important collections of design principles, but there is little evidence of how these principles can be put into action and what effect they have on players ethical engagement.
This trend appears to be reversing, with a number of interesting papers coming out showing empirical data gathered from purpose-designed videogames, including this work by Sarah Hodge, Jacqui Taylor and John McAlaney at Bournemouth University. In this instance, the game is a 3D first-person role-playing game, depicting six moral choices corresponding to the six moral foundations Tamborini’s Model of Intuitive Morality and Examplars (MIME) based on Haidt’s Moral Foundations Theory (MFT). Each choice provides the player an opportunity to help an NPC with a problem, related to one of the six moral foundations. The player can choose either a pro-social or an anti-social action to take.
The research questions addressed are:
- How long do in-game moral decisions take?
- Which factors of real-life MFT domains, previous gameplay, in-game and post-game experiences will predict in-game moral alignment?
- How will real-life moral saliance relate to in-game moral choice and reaction time?
In game moral alignment was calculated as the number of pro-social choices minus the number of anti-social choices (although it is not clear why this is necessary, since it seems that these two options are the only ones available). As shown in previous research, players generally prefer to make pro-social choices (a mean of 5.19 out of 6). Reaction times were generally short (4-8 seconds). Interestingly, shorter reaction times correlated with higher moral alignment scores. The fastest choices were made on the Care/Harm and Sanctity/Degradation foundations (mean 4s) and the slowest choices were made on the Loyalty/Betrayal foundation (mean 6.6s).
Unfortunately, the other two research questions don’t really provide compelling results. My feeling as a designer is that this reflects on players’ relatively poor engagement with the game. The PANAS-X scale of positive and negative affect showed a strong negative result. The description of the game makes it clear why this might be the case: the game was deliberately designed without any overarching narrative “to reduce and remove biases”, and the moral choices are disconnected and lacking context. While this design decision makes sense as a psychological experiment, it removes a lot of the “gameness” of the game, and the choices seem very artificial and arbitrary.
This points to a general issue with empirical study of games: A game is a holistic experience. In-game decisions are highly contextualised and situational. In fact, this is one of the key design criteria in Sid Meier’s famous theory of “interesting choices“. Trying to pull a decision out of context in order to examine it in the laboratory destroys one of the important factors that makes it interesting, and therefore tells us less about what would happen in actual play. Likewise, ethics is highly situational. While we can study ethical choices in isolation to determine universal moral principles, real ethical choices are deeply embedded in a complex web of relationships, history and other situational factors that influence our choices.
I don’t mean this as a criticism of this paper, but as a general reflection on the problems we all going to face doing empirical research in this area. A good game is by its very nature a rich, complex, interconnected experience. This makes isolating cause and effect very difficult. The more we simplify in the name of good science, the less our experiments will be representative of real gameplay. I’m not sure what the answer to this paradox might be.