Play is universal, and is generally understood to be educational.
Play is a universal phenomenon in the animal kingdom. From dolphins to dogs, countless species engage in playful activities, and these behaviors are generally understood to be a means of practicing essential life skills. As humans, we have developed an uniquely advanced form of play: video games. These interactive experiences are immensely popular, and offer a rich source of naturalistic, skillful, goal-oriented human behavior that is largely untapped by cognitive science. This workshop will explore the hypothesis that video games may serve as the ideal testbed for building and improving unified theories of cognition.
Chess and later go were held up as examples of complex tasks requiring intelligence, but aglorithms for playing both turn-based, perfect information board games have exceeded human levels -- without achieving human-level general intelligence. While special-purpose algorithms for playing modern video games (e.g., multiplayer online battle arenas (MOBAs) such as League of Legends) are capable of handling the real-time, partial information, and collaborative aspects of these games, it is an open question whether a general game-playing algorithm would approach human-level intelligence.
