Two-person\u27s games with dynamic internal parameters

Recent video games have become so complex and highly qualified that non-player characters, called NPC, are required to behave more intelligently and naturally against their situations and environments. There are few theoretical ways to build NPSs as intelligent players, although most games apply ad-hoc rule-based methods to NPCs for their making decisions. Since such methods of NPCs play in a simple andtypical manner, human players tend to get bored as games go.For these problems, this paper proposes a decision-making model using game theory involving player\u27s internal parameters. In this model, the payoff functions and state transitions will be dynamically calculated with internal parameters of both players for maximizing internal parameters. Big difference from traditional game theory resides on such maximizetion method for dynamically changing. This paper demonstrates that two players\u27 best-response for a game with two internal parameters and two strategies is decided into a linear space problem. It also discusses the mathematical properties of a game, the long term gaming behavior (i.e. divergence or convergence of parameters) and the manipulation of theequilibrium coming from repetition of games on the border of strategic change. Experiments are carried out under a detailed condition of the relationship between player\u27s states and strategies using parameter matrices for cooperative or uncooperative features in a simple two-person repeated game