A Feel for the Game: AI, Computer Games and Perceiving Perception

Ouellette and Conway's "A Feel for the Game: AI, Computer Games and Perceiving Perception" argues that current game artificial intelligence, built on symbolic manipulation and binary logic, cannot perceive or care about the significance of play, and that this shortcoming explains why human-level game AI remains largely unrealized despite early optimism among researchers such as van Lent and Laird. Drawing on phenomenology from Heidegger and Gadamer and on Bateson's account of play as meta-communication, the authors contend that affective intentionality, the "as-if" commitment that gives play its meaning, has been neglected in both AI models and game design, which have prioritized behavioral responses over the perception and emotional engagement players actually experience.

The analysis traces this neglect to the militaristic origins of game AI engines, which the authors argue embed a hegemonic masculinity oriented toward emotional reserve and instrumental rationality. Fun and realism have been treated as deterministic success factors, leaving the evaluative, bodily dimension of emotion largely unaddressed. The authors conclude that the mediated interface and the limits of current systems shape how players respond, so affective intentionality must be studied alongside cognition rather than subordinated to it.

The document's central proposal is that game AI should perceive players' affective responses, not merely their actions. Forza Horizon 4's adaptive drivetar difficulty is cited as an early practical instance of this approach. Support also comes from Kshirsagar et al. (2019), whose human-versus-robot competition experiment found that people's effort was driven by non-monetary attachments, a result the authors' own hypothesis had not anticipated. The authors maintain that computers, being "ontic" rather than "ontological," cannot attune to mood or play in the phenomenological sense, and therefore future AI and game design require a heuristic that accounts for affective intentionality.

The scope spans game studies, phenomenology, play theory, and game AI research, with the argument grounded in a VR virtual museum example in which a fire-tending interaction failed to elicit perceptually appropriate responses.