Recognition of coalition by computer in a three-person game

Abstract

An information processing model for recognition of coalitions in three-person games is developed by using a computer program, which simulates humans' move selection behavior, in Chinese checkers. The program plays as a participant against human opponents and decides if a coalition was formed after a game is over. It has two main parts: (a) a heuristic search algorithm for move selection and (b) a coalition recognition procedure which evaluates deviation patterns. The latter is defined as the difference between the move pattern of the selection algorithm and that of a human player. With this model, it is shown that humans' move selection behavior may be relatively simple even when the task environment involves human interaction and also that the moves of human players are influenced to some extent by the possible existence of coalitions. The system treated here is an information processing system, a computer program, which works as a member of a social environment and recognizes interactive behavior of other members. It models human behavior in such environments at the symbolic information processing level or the level of information processing psychology.