“As if it was moral”: The use of non-player characters (NPCs) to explore morality in video games

This work offers a set of design principles that might be useful to develop non-player characters (NPCs) to investigate moral decision-making in video games. Such principles might be ideally used in behavioural experiments to create a context where the consequences of a decision made by a player are limited to the in-game environments. To achieve this goal, we focus on how players interact and create in-game relationships with NPCs through video game mechanics. A key element of this analysis is the identification of particular human-like behaviours and human-like features, such as speech, facial expression, etc., which are central to putting the NPC and the player in a shared narrative context. We specifically look at several video games, such as “The Walking Dead” or “Spec Ops: The Line”, where NPCs behave as they were driven by agency and moral values. These NPCs face morally difficult decisions or engage in what in real life would be a crime, all while keeping the player in the center of the narrative of the game. The way NPCs are perceived in these video games, combined with the overall narrative of the game, allows players to make decisions that have a distinctly moral dimension. This suggests that there are indeed mechanisms that lead us to ascribe agency and moral values to NPCs and that we can study the mechanisms of moral decision-making with video games. However, this will be made possible only if NPCs are appropriately designed. The first design principle we focus on is the human ability to attribute agency. Human beings seem to have been provided by evolution with a cognitive architecture that makes them able to identify agency and intentionality in non-human entities. Arguably, this ability developed relatively early in human evolution and has helped our ancestors anticipate the behaviour of predators. Non-human agents, such as pets or robots, can be ascribed with specific beliefs and even decisional capacities, including the capacity for moral behaviour. Such interactions are always located in a determinate time and space and connected to the presence of a narrative, which places us in the centre. The second design principle we focus on is the co-location of the NPC within a narrative. There are some key insights about moral narratives, such as Jonathan Haidt’s moral foundation theory (and others), which can be used to identify morally laden content. In our work, we survey some of the existing theories and tools that can be particularly helpful in identifying the relevant principles and features to develop NPCs useful to investigate moral-decision making in behavioural experiments.