Sociological aspects of cooperation

The most prominent studies of cooperation in sociology proceed from some kind of rationality assumption. There are basically two groups of studies, one starting from the assumption of rational egoists and the other starting from the assumption of social rationality in which other regard has a prominent place. The studies based on the assumption of rational egoists is strongly inspired by (mostly game-theoretical) developments in economics and political science but add important sociological aspects to the analysis. Most important among these contributions are the dynamic analysis of cooperation, the link between formal and informal institutions, and the importance of social networks for embedding problematic transactions. Studies based on social rationality introduce assumptions on social preferences, learning, and framing. Important questions are the circumstances that strengthen or weaken social preferences; the evolution of signals and conventions that govern trust, and the effects of goals on the definition of the situation. It seems that studies using social rationality are on the rise. Sociological aspects of cooperation Andreas Diekmann and Siegwart Lindenberg 1. The problem of cooperation. Individuals cooperate if each willingly acts in a manner that contributes to the others‘ welfare. Cooperation is one of the oldest and most revered topics in sociology even though it appeared under a variety of headings, such as „social order“ and „solidarity“. The classical sociological answers to the question how cooperation among individuals comes about differ in detail but share a common core: Cooperation is the result of shared values and norms and normconforming behavior. A problem with this „obligation-centered“ solution is that there is no explanation of where norms come from and why the degree of norm conformity varies even for the same individual. over time. Theory that might solve these problems was not developed until the 1970s when a sea change in the basic assumptions on human action took place within sociology. At that time, analyses of cooperation among rational egoists (mostly rooted in game theory) began to shed new light on the way one might analyze cooperation, norms, and norm conforming behavior. For quite some time, game-theoretic analyses of cooperation had become the most serious attempt to explain cooperation. Big advances were made but many problems remained, especially the problem that individuals do not seem to be as self-centered Diekmann and Lindenberg Cooperation 2 2 as they are assumed to be in game theory. Still, the game-theoretic analyses were powerful enough to suggest that there was no way back to explanations of cooperation that ignored human rationality all together. More recently, analyses, based on some form of social rationality have come up in order to address some of the unsolved problems. In the following, we will first present what we consider to be the most important aspects of sociological research on cooperation among rational egoists. This will take up the larger part of the article. We will then discuss some contributions on cooperation with social rationality. 2. Cooperation among rational egoists. It may not be quite realistic to assume that individuals are always looking out for themselves and that they do so by weighing costs and benefits of their behavior. However, this assumption of „rational egoism“ seems to have four strong points in its favor. First, it allows one to pinpoint why cooperation is an interesting problem in the first place. Second, it allows rigorous theory formation with tractable structures of argumentation. Third, everyday experience and research give this assumption enough support to make it worth while to use it in the light of its analytic advantages, especially when analyzing interaction among interdependent individuals. Fourth, at later stages of theory formation, this assumption can be relaxed in favor of more realistic assumptions. Game theory has furnished the most important instruments for such analyses, and since the late 1970s – inspired by developments in political science and economics –sociologists have begun to use it to study questions concerning cooperation, bringing typically sociological aspects into the analyses, such as network embedding. The crucial first step in the „new“ analysis was to identify typical situations in which cooperation is problematic: social dilemmas. A social dilemma is – roughly speaking – a situation in which the rational decisions of individual actors are likely to lead to a collectively sub-optimal outcome. An outcome is suboptimal (not „Pareto optimal“) if one or more actors could improve their position without worsening the position of any other actor(s). In a social dilemma, individual and collective rationalities are in conflict. Cooperation can now be succinctly defined: an actor cooperates if and only if he or she chooses a course of action that will lead to a collectively rational outcome when other actors behave cooperatively as well. Social dilemmas occur when Pareto-superior behavior cannot be achieved by enforceable contracts. This may have many different reasons. For example, it may be impossible to communicate with others in order to come to an agreement; or, if communication is possible, transaction costs may be too high to actually reach an agreement. If agreements are within reach, it may be practically or epistemically impossible to specify all Diekmann and Lindenberg Cooperation 3 3 relevant future contingencies in the contract. Even if a contract can be drawn up, there may be no court with sanctioning power; or if there is such a court, it may not be possible to prove uncooperative behavior in court. . There are structurally different variants of social dilemmas, and game theory helps to distinguish them. For example, there is the common’s dilemma (Hardin 1968, Ostrom 1990) in which various actors have free access to a resource (such as fishers from different nations in international waters); there are step-level public goods (Taylor 1987) for which only a limited number of actors have to cooperate in order to produce the public good for a much larger group. Of this, a special case is the volunteer’s dilemma (Diekmann 1986) for which only one volunteer is necessary to produce the public good for all, such as rescuing a drowning child with a crowd of onlookers; there is the trust game (Dasgupta 1988) in which the trustee can honor or opportunistically exploit the cooperation of the other (as when you lend money to an acquaintance who promises to pay it back with interest). The Prisoner’s Dilemma (PD) is by far the best known variant of a social dilemma. It involves an arbitrarily large number of people. The special case of a two-person PD has been explored the best so far. In a two-person PD, both players have the option to cooperate or to defect (see Figure 1). If both cooperate, each gets a payoff of value R. If one defects while the other cooperates, the defector gets T and the cooperator gets S. If both defect, both will get P, with T>R>P>S. There is a dominant strategy for each player (i.e. an individually rational choice, no matter what the other does) leading to a so-called Nash equilibrium (in which no player can unilaterally improve his or her position by deviating from the equilibrium strategy) and to a Pareto-inferior result. The assumption of rational egoism leads to two motives in a PD: the defensive motive to shield against exploitation (if the other defects you better defect yourself), and the aggressive motive to exploit a cooperative player (if the other cooperates, you get even more by defecting). Together, these motive result in the conflict between individual and collective rationality. Mutual cooperation would be the Pareto-optimal result but the two motives lead to mutual defection. Diekmann and Lindenberg Cooperation 4 4 2.1 The Embedding of Cooperation Raub and Weesie (2000) have suggested that game-theoretic work on solutions to social dilemmas can be divided into three kinds of embeddings of potential cooperation: embedding in time, in institutions, and in social networks. We follow their suggestion for this overview. Time. When actors find themselves in a social dilemma that is iterated indefinitely over time, new sources for cooperation arise. First of all, defection can be sanctioned by the other player(s) in later moves, and the resumption of cooperation by the defector can be rewarded by a similar response of the other player(s). Friedman (1977) and Taylor (1987) have shown that if the value of future outcomes is high enough, this opportunity to make one’s cooperation conditional on that of the other leads to cooperative strategies with Nash equilibrium and Pareto optimality. Axelrod (1984) built on this result to ask the question which strategy would be most conducive to lead to cooperation in an iterated PD. In order to answer this question, he conducted computer simulations in which various strategies played against each other for about 200 rounds. Axelrod had invited suggestions for the strategies that would create the most cooperation. Winner was the conditional strategy „tit for tat“ (TFT) suggested by the mathematician and psychologist Anatol Rapoport. TFT is very simple: (a) always begin with cooperation; (b) for the present round, copy the move of your partner in the previous round. TFT has a number of distinct features. It is „nice“ because it starts with cooperation; it cannot be provoked because it sanctions exploitation; and it is forgiving because it returns to cooperation after the other has done so. Contrast this for example with „permanent retaliation“ in which defection is answered by unforgiving defection, a strategy C D