The Representational Challenges of Fictional Worlds

Modern computer games often have a strong narrative structure to the progress of their gameplay. In some genres, story is at the forefront of the interaction. As game makers seek to add new capabilities for procedurally generated content to their new releases, work on interactive narrative is growing in significance both as a basic research topic as well as a source for insight into the creation of content for games. As a result, there are now many compelling challenges for researchers working in AI that involve interactive computer games as their area of application. A significant different exists between application contexts familiar to most AI researchers and those found in games, however. Games are fictional worlds. Of course, we are all aware that games are fictional worlds. What is not so obvious is what this means to the relationship between a player and his or her game system and what consequences the relationship holds for the use of representations like logic and other formal computational approaches. Games are expected to be fictional in much the same way that other narrative media are. Their worlds share many properties in common with the real world, but their experience is specifically not realistic. Games are not simulations, where system behaviors are required to be identical to real world correlates. When a game player begins to play a game, he or she steps inside a magic circle (Huizinga, 1955), effectively entering into a contract with the game’s designers where players agree to suspend disbelief and play according to the rules and designers agree to manipulate the game world in ways that facilitate that effective play rather than recreate realistic interaction. Understanding narrative in these game worlds, as in other media, involves the active construction and experience of an illusion (Gerrig, 1993; Johnston and Thomas, 1995). This construction process is particularly participatory in computer games, where players not only