A review of game theoretic and social psychological models of coalition formation / BEBR No. 372
Year: 1977
Type: article
Abstract: This paper reviews three classes of game theoretic solution concepts (solutions, subsolutions, and the core, bargaining set models, and the Shapley value) and four social psychological models (Minimum Resource theory.Minimum Power theory.Bargaining theory, and the Weighted Probability model) of coalition formation.The research that has been conducted on characteristic function games and on coalition situations involving more than three players has been summarizedand the models have been evaluated.The advantages of collaboration between the two disparate but similar areas are discussed.Formal theory concerning coalition behavior has been studied by game theorists since 1944 (von Neumani\ and Morgenstern) and by social psychologists since 1956 (Caplow) .The two areas have adopted different philosophical approaches and have made limited progress in their relatively independent' pursuits of knowledge.Game theorists have developed elaborate and elegant mathematics (i.e., normative theory) but have paid little attention to the applicability of their results to human behavior.Social psychologists, on the other hand, have collected a large amount of data on three-person groups, but have only recently expanded their descriptive models in attempts to pre- dict bargaining behavior in larger, more complicated conflict situations.This paper sumnarizes selected portions of the research and theory from the two areas and attempts to locate the common group between them.A concerted effort bridging the two areas holds increased promise for breakthroughs which each field alone might findunattainable.This paper limits its review to aspects of the two areas that are most similar and that, therefore, have the greatest collaborative potential.Extensive reviews of the game theory literature can be found in Luce and Raiffa (1957) or Rapoport (1970).Reviews of earlier social psychological research can be found in Chertkoff (1970) or Stryker (1972) .Present coverage of game theoretic models is restricted to characteristic function models of cooperative games; coverage of social psychological models is restricted to those v^hich make a_ priori predictions for either the formation of or the payoffs to certain coalitions.These restrictions preclude, for instance, discussion of the Nash equilibrium (1951) and Laing and Morrison's (1973) model of sequential games.Models o.Coalition F. 2 Game Theoretic Models A major distinction in the study of n-person games is made between cooperative and non-cooperative games.Cooperative games are those in which the players have the opportunity to communicate with one another and to form binding agreements.Non-cooperative games do no permit binding agreements, and may even forbid communications.Thus, for example, the usual form of the prisoner's dilemma game is non-cooperative.(Cooperative games, on the other hand, are those where the players bargain to determine their pay offs.)In describing cooperative games, most theoretic models utilize the characteristic function-The characteristic function of a game specifies a value or payoff to each possible coalition, including one-person coalitions.For instance, a three-person game with outcomes of 1 for each two-and three-person coalition and for all one-person coalitions would have the following characteristic function:where v( ) indicates the value or payoff to a coalition, and A, B, and C are actors or parties in the game.The characteristic function, then, emphasizes the payoffs that coalitions receive: coalitions are differentiated on the basis of the payoffs they can obtain.Solutions and Subsolutlons von Neumann, and Morgenstern's (1944) presentation of n-person conflict situations led them to the question of x^Jhat outcomes might be considered stable.They introduced the notions of imputations Models of Coalition Fo 3 and domination.An imputation is a payoff configuration that satisfies conditions of individual rationality (i.e., a party will not accept a payoff from a coalition that is less than the payoff it can receive playing alone) and Pareto optimality (i.e., a payoff will not be considered if another payoff increases the outcomes of at least one of the included parties while not reducing the payoffs of any of the other included parties) .
Author: J. Keith Murnighan
Language: English
Subfield: Economics and Econometrics
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W2592881070
