An Ontological Meta-Model for Game Research

Aarseth and Grabarczyk address the fragmentation of game ontology, a tradition of roughly sixty years running from Caillois through Avedon, Björk and Holopainen, and Zagal and colleagues. Disciplines bring incompatible methods and overlapping vocabularies such as "game," "mechanics," and "gameplay," and few models build on one another. Drawing on cognitive science's experience with interdisciplinary work, the authors propose a meta-ontology that treats games as mechanisms with multiple levels of description, so that differing ontological perspectives can be classified and related rather than homogenized.

The mechanistic view permits ontological reductionism without epistemological reductionism, much as software and hardware descriptions coexist. It also guards against descriptive isolationism, allows researchers to check the scope of broad claims, and exposes terms that function as tacit metaphors when carried across levels, as with "shooting" in game mechanics.

The resulting model is deliberately over-productive, with four layers of three sub-layers each: physical (platform, interface, behavioral), structural (computational, mechanical, economic), communicational (presentational, semantic, interface), and mental (phenomenal, conceptual, social). Applied to four existing ontologies, it shows where categories cut across layers, as with MDA's "dynamics" spanning structural and communicational, and confirms that Aarseth and Calleja's four categories align fully with the main layers.

Applied to nine game definitions from Caillois to Deterding, the model finds that none covers every layer, few address the physical layer, later definitions span more layers, and most pair mental with structural properties. Deterding's definition covers physical, communicational, and mental layers but omits the structural. The authors attribute such differences to ambiguities between game as process versus artifact and between play and game. A case study of Senet shows that only its physical and communicational layers survive, since rules and player experience are lost; here rules are treated as mental-level interpretations and mechanics as structural features. The model serves as both ontology and comparative tool, with potential for clarifying notions such as versions, ports, and clones.