Social Dilemmas and Cooperation

Frontmatter

Social orientation analysis of the common and individual interest problems

Abstract

The purpose of this chapter is to show a new way to see social dilemmas and to explain people’s decision making in social dilemmas. The key concept in our analysis is social orientation. We will apply the social orientation analysis, as was done for 2×2 matrix games, to an analysis of decision making by people in social dilemmas. To do so, we will present new formulations for the two types of social dilemmas: “free rider” and “tragedy of the common.” Our analysis will show how our formulations are different from, interrelated with, and superior to the payoff matrix expression. Furthermore, we will also show how the two types of social dilemmas are different, how social dilemma situations are normalized, and how people’s decision making can be explained using a social orientation model.

Toshiaki Doi

Toward more locomotion in experimental games

Abstract

This chapter evaluates the traditional experimental gaming approach to the study of social interdependence. In addition to outlining several strengths, it is asserted that this approach is limited in two respects: (1) it does not enable a researcher to examine the ways in which individuals express their simple motivations and simple strategies when they are provided with more varied domain of options, and (2) it neglects an important domain of social interaction, namely those situations in which individuals are able to alter the underlying interdependence structure. The chapter reviews prior research that extends the traditional experimental gaming approach by offering subjects the possibility to alter the nature of interdependence. It is concluded that the ways in which individuals express their simple motivations and strategies may be importantly shaped by the availability of other options than a cooperative and noncooperative choice. To provide more insight into these processes, we should consider a greater locomotion in the way in which we use outcome matrices in our research on social interdependence.

Paul A. M. van Lange

Individual reasoning process in the participation game with period

Abstract

A game with period was introduced to examine the effect of revised feedback information on increasing participation, as it related to the individual reasoning processes. Five one-shot games and a game with period were played by the same subjects in the experiment 1. The result showed that the level of participation in the game with period was lower than the one-shot games, due to the change in players’ decision rule with respect to the information seeking. In the experiment 2, a persuasion procedure on reasoning was introduced to the same game with period. However, the persuasion effect on the individual reasoning process was not so evident.

Tetsuo Takigawa

The position effect: The role of a player’s serial position in a resource dilemma game

Abstract

We consider the following single-stage resource dilemma game with both strategic and environmental uncertainty: Members of group N are required to share a common resource whose size x is not known with precision. Rather, x is sampled randomly from a probability distribution which is common knowledge. Individual players make their requests privately; their requests are granted if and only if the total group request does not exceed x.

We examine a sequential protocol of play in which individual requests are made in a prespecified order. Players are not informed about previous requests in the sequence; they are only told their serial position in the sequence.

Data from 45 subjects, in correspondence with our previous studies, show significant effects due to resource uncertainty. In addition, they reveal a significant position effect with players appearing late in the sequence making smaller requests.

Ramzi Suleiman, David V. Budescu, Amnon Rapoport

Positive and negative mood effects on solving a resource dilemma

Abstract

A resource dilemma is a situation in which maximizing individual profits contradicts the conservation of a resource. How do people solve a resource dilemma if they are in positive or negative mood? There are some models which offer different explanations and predict different outcomes: (1) the cognitive capacity explanation expects that people in negative mood are hampered in performing a cognitive task because their cognitive capacity is reduced by thinking about their emotional state; (2) the mood repair explanation assumes that people in negative mood overexploit the resource to improve their mood; (3) the motivational explanation claims two different strategies depending on mood state. The hypotheses derived from the three models were tested in two experiments. The results lend some support to the motivational explanation.

Regina Vollmeyer

Fairness judgements in an asymmetric public goods dilemma

Abstract

Dawes (1980) and Marwell and Ames (1979) have suggested that decision making in social dilemmas is importantly influenced by considerations of fairness. The present study examined subjects’ fairness judgements in an asymmetric public goods dilemma. Subjects stated the fairest possible contribution each of six persons could make to provide a public good. These six persons differed in terms of individual wealth (resource asymmetry). The subjects’ attributions for resource asymmetries were manipulated so that for some, attributions were Internal (people are rich or poor due to their own efforts/abilities) and for others, External (people are rich or poor due to task difficulty/luck). This manipulation was the major independent variable. As expected, subjects making internal attributions stated that poor persons should contribute a larger proportion of their wealth than rich ones. The subjects making external attributions judged equal proportional contributions to be most fair. Fairness judgements appeared to influence subjects’ own contributions and their expectations of others’ contributions. The subject’s Social Orientation (Cooperative, Noncooperative) was also measured. Cooperators judged equal proportional contributions as fair, whereas Non-cooperators felt the poor should contribute a higher proportion of their wealth than the rich. Cooperators and Non-cooperators contributed equal proportions, in contrast to the hypothesis.

Jeffrey A. Joireman, D. Michael Kuhlman, Hidetaka Okuda

Group size effects in social dilemmas: A review of the experimental literature and some new results for one-shot N-PD games

Abstract

It is often argued that group size has negative effects on cooperative behavior in social dilemmas. However, this widely accepted view does not accurately reflect the findings of the experimental research. This paper presents an updated review of the experimental findings on group size effects in various social dilemmas. Also, new results concerning group size effects in the one-shot Prisoner’s Dilemma game are presented. The findings from the review and the experiment suggest that the effects of group size depend on the type of the game as well as on the experimental setting. While group size effects appear in infinitely iterated PD-type games, they do not affect the level of cooperation in the one-shot Prisoner’s Dilemma game. Reasons for the observed differences and the factors that drive the group size effect are discussed.

Axel Franzen

Provision of step-level public goods: Effects of different information structures

Abstract

We report the results of two experiments on social dilemmas in which each of n players receives an endowment and then decides privately whether to contribute it for the provision of a monetary public good. The good is provided if and only if at least m group members contribute. Decisions are made sequentially. We present and then test an equilibrium model under two different information conditions in which players are informed of 1) the previous decisions in the sequence; 2) either the number of previous contributions or the number of previous non-contributions. When the equilibrium solution yields unique predictions, most of the subjects behaved in accordance with it.

Amnon Rapoport, Ido Erev

Conditional contributions and public good provision

Abstract

The present article focuses on the effect that conditional contributions can have on the provision of public goods. First, a theoretical analysis is presented. It is argued that since promises will often be conditional, making promises can be regarded as a strategic move that alters the payoff structure of public good dilemmas. Second, an experiment is reported in which one member supposedly committed him/herself to contribute if at least one other member contributed too. The main focus in the experiment was on the reactions of the members confronted with this conditional contribution.

Eric van Dijk, Henk Wilke

Convergence in the orange grove: Learning processes in a social dilemma setting

Abstract

In a controlled social dilemma experiment conducted in an orange grove, Erev, Bomstein and Galili (in press) found that behavior converges to the game theoretical equilibria points. The present paper summarizes these results and explores learning models by which they can be described. Simulation results favor an imitation based learning rule over a fictitious play rule. In addition, it was found that the imitation heuristics can be learned by a simple reinforcement based learning rule.

Ido Erev

Leadership and group identity as determinants of resource consumption in a social dilemma

Abstract

Leaders who act on behalf of a group to access a common resource have been shown to “manage” the resource better, in that the average amount of the resource taken per persion is smaller, and distribution of resources in the group more equitable. However, research on leadership has usually involved only one group, thus eliminating any competition for the resource when the leader alone can access it. If several groups with leaders have access to a resource, will good management continue, or will leaders attempt to take more for their own group members? In this paper, we examine factors which might make the leader more or less resource conserving, with an emphasis on accountability to a subgroup, and on level of social identity. We predict that better management by leaders will be undermined if subgroup membership is made salient, but that conditions promoting dual accountability will also promote a more cooperative and resource conserving strategy.

An experiment was conducted to examine the relative impact of subgroup identification and leadership on resource use. Psychology undergraduates were led to believe they were connected by a computer network to a data-base in the Library, along with five other Psychology students (ingroup), and six Education students (outgroup); in a fourth, control condition, Psychology students believed they were in a network with 11 other unidentified students. Total time available on the data base was insufficient to meet the needs of all students, who required the data base to complete an assignment. In the first, baseline, phase, subjects made a booking for themselves individually; in the second phase, subjects were assigned to one of four conditions: ingroup leader, outgroup leader, individual ingroup member, or individual, non-ingroup member (n=20 subject per cell). Leaders made bookings on behalf of their group members in the second phase; individual booked only for themselves.

Results showed that both ingroup and outgroup leaders booked significantly fewer hours/person than did non-leaders; there was no difference between the two leader conditions. On the whole, leaders distributed hours equally, although need was also an important criterion. While our results suggest that the leadership effect overrides the effect of ingroup identification, our manipulation of group membership (common college major) may have made competition among individual ingroup members more salient than intergroup competition. Further research using a variety of means to create group identification is needed to establish the impact of leaders on resource consumption.

Margaret Foddy, Andrew Crettenden

Prisoner’s dilemma networks: Selection strategy versus action strategy

Abstract

To date, most research on Prisoner’s Dilemmas has dealt with isolated dyads. However, most PD-like relations in the real world take place in a network of relations where each player has a choice of partners. In this research project, we have created a situation where (1) every member of a group selects a partner (that is, two parties form a relationship by mutual choice, and each is free to leave the relationship), and (2) a PD game is played by members who have selected each other. We call this situation a prisoner’s dilemma network. We invited social dilemmas researchers to a computer contest of strategies. Nine strategies participated, and the winner of the contest was PURGE submitted by T. Kameda, which was a simple out-for-tat strategy. The major findings of this study are: (1) Action strategy (how to decide between C and D) is not as important as selection strategy (how to select a partner). (2) Out-for-tat seems to be the best strategy in PD networks. (3) In the situation where actions of the partner can be “mistaken,” forgivingness to defection in out-for-tat is important.

Toshio Yamagishi, Nahoko Hayashi, Nobuhito Jin

Choice of strategies in social dilemma supergames

Abstract

The purpose of the studies reported in this chapter is to examine if people prefer trigger-type payoff structure over NPD structure, and if trigger-type payoff structure is created at the supergame level as a result of strategy choices. Results of two experiments reported below are generally negative to these questions, and thus are inconsistent with the result of an earlier study (Watabe, 1992). Result of a computer simulation, however, suggests that trigger-like-payoff structures are more likely to result in larger groups than in smaller groups. This may explain the difference of the current result involving 4-person groups from the result of Watabe’s (1992) study involving a 10-person group.

Motoki Watabe, Toshio Yamagishi

Social dilemmas exist in space

Abstract

Proper representation of many kinds of social dilemmas requires the consideration of social space. Social interdependence is mainly limited to other people located in a local neighborhood. The concept of social space may be more fruitfull in some instances than the network description. Several geometries that may be usefull describing the social space are discussed: Euclidean, City Block, Fractal, One-dimensional, Railroad, Random, Probabilistic, and Multiple spaces metric. The outcomes of the conformity game for several types of the geometries of the social space are discussed.

Social dilemmas exist in social space, very much as do the societies that the dilemmas concern. The nature of this space should be taken into consideration when analyzing a social dilemma, since both the nature of the dilemma and social processes in the groups affected by the dilemma are dependent on the properties of the social space. In this chapter, we will use computer simulations to show some effects of space on social processes and propose some geometries relevant to representing social space. In the next chapter, the effects of the geometry of social space in an experimental setting with human subjects will be presented.

Andrzej Nowak, Bibb Latane, Maciej Lewenstein

Commuting by car or by public transportation? An interdependence theoretical approach

Abstract

Subjects from two different populations -- daily commuters and students -- were asked to make a single choice between going by car or by public transportation in a hypothetical commuting situation. On the basis of interdependence theory (Kelley & Thibaut, 1978), we assumed that the choice for car versus public transportation would be affected by the perceived structure of interdependence underlying the decision situation. Consistent with our predictions, it was found that commuters primarily perceived the decision situation as an accessibility problem -- with the basic structure of an N-person Chicken Dilemma -- whereas students primarily perceived the decision situation as an environmental problem -- with the basic structure of an N-person Prisoner’s Dilemma. Relative to commuters, students exhibited a greater preference for commuting by public transportation, and exhibited a weaker tendency to take expectations regarding others’ choices into consideration. These and other findings are discussed in terms of interdependence theory.

Mark van Vugt, Ree M. Meertens, Paul A. M. van Lange

Evolution of norms without metanorms

Abstract

A social dilemma can be resolved if group members sanction each other’s choices to defect. However, the provision of mutual sanctioning involves another social dilemma (i.e., a second-order dilemma), in which non-sanctioning is the dominant choice. Based on a series of computer simulations, Axelrod (1986) concluded that this second-order social dilemma can be resolved by metanorms (i.e., sanctioning the non-sanctioners). An alternative interpretation of Axelrod’s simulation results is proposed and tested through a series of computer simulations. Specifically, it is shown that the evolution of norms (which entail mutual sanctioning) does not require metanorms (i.e., sanctioning the non-sanctioners) insofar as the decisions to cooperate and the decisions to sanction are linked (i.e., cooperators punish defectors, and defectors do not punish other defectors). Furthermore, the linkage between these two types of decisions itself is shown to emerge through an “evolutionary” process.

Toshio Yamagishi, Nobuyuki Takahashi

Computer simulations of the relation between individual heuristics and global cooperation in prisoner’s dilemmas

Abstract

In this paper we explore the aggregate consequences of three simple individual choice rules in a simulated society in which the interaction between pairs of actors is constituted as a Prisoner’s Dilemma (PDG). After several simulations under different conditions, the most important conclusion is that in highly competitive PDG settings, using a competitive type of social comparison for performance evalution, cooperative behavior is maintained in large groups. There remain however several findings which need further study before we better understand the relation between individual heuristics and their global consequences.

David M. Messick, Wim B. G. Liebrand

What risk should a selfish partner take in order to save the life of a non-relative, selfish friend? A stochastic game approach to the prisoner’s dilemma

Abstract

A model of cooperation versus defection in a sequence of games is analyzed under the assumptions that the rules of the games are randomly changed from one encounter to another, that the decisions are to be made at each situation anew, according to the specific rules of the specific local game, and that the outcome of each such game affects the ability of a player to participate and, thus, cooperate (if in its own interest) in the next game. Players wish to maximise their total payment at the entire supergame. Under plausible assumptions it is shown that all Evolutionary Stable Strategy (ESS) of the supergame determine cooperation over a non-degenerate range of situations determining encounters of the P.D. type - A moderately altruistic cooperation is selected because it enables the survival of a potential partner for future cooperation. The model also explains the evolution of gratitude, rather than the assumption thereof, and predicts a qualitative difference between partnership altruism and kin altruism.

Ilan Eshel, Daphna Weinshall, Emilia Sansone

Learning models for the prisoner’s dilemma game: A review

Abstract

The paper provides a review of mathematical learning models for gaming behavior in sequences of prisoner’s dilemma games. The introductory part of the paper gives some basic information about the mathematics of traditional learning models. Three types of models are explained: Linear operator models, choice models and Markov-models. In the first part classical learning models for overt gaming behavior are presented. They were developed by Anatol Rapoport and his co-workers. Results for two Markov-models and a linear operator model are discussed. Van der Sanden developed a Markov-model with two latent value states and a transitional intermediate state. This model is based on earlier work of Meeker. Micko, Brückner und Ratzke developed a linear operator model with two latent dimensions called “cooperative inclination” and “trust”. The basic ideas of this model are very similar to Pruitt and Kimmel’s goal-expectation theory. Schulz and co-workers developed a series of learning models which account for players’ expectations and separate social value parameters and information processing parameters. In the most recent model of this type goals, expectations, and choices are overt variables. This model allows empirical tests of the goal expectation theory.

Hans Christoph Micko

Social capital and cooperation: Communication, bounded rationality, and behavioral heuristics

Abstract

Common-pool resources are natural or man made resources used in common by multiple users, where yield is subtractable (rival) and exclusion is nontrivial (but not necessarily impossible). The role of face-to-face communication in CPR situations, where individuals must repeatedly decide on the number of resource units to withdraw from a common-pool, is open to considerable theoretical and policy debate. In this paper, we summarize the findings from a series of experiments in which we operationalize face-to-face communication (without the presence of external enforcement). In an attempt to understand the high degree of cooperation observed in the laboratory, we turn to a bounded rationality explanation as a starting point for understanding how cooperative behavior can be supported in decision environments where game theory suggests it will not.

Roy Gardner, Elinor Ostrom, James Walker

Cooperation in an asymmetric volunteer’s dilemma game Theory and experimental evidence

Abstract

The symmetric Volunteer’s dilemma game (VOD) models a situation in which each of N actors faces the decision of either producing a step-level collective good (action “C”) or freeriding (“D”). One player’s cooperative action suffices for producing the collective good. Unilateral cooperation yields a payoff U for D-players and U - K for the cooperative player(s). However, if all actors decide for “freeriding”, each player’s payoff is zero (U > K > 0). In this article, an essential modification is discussed. In an asymmetric VOD, the interest in the collective good and/or the production costs (i.e. work) may vary between actors. The generalized asymmetric VOD is similar to market entry games. Alternative hypotheses abaout the behavior of subjects are derived from a game-theoretical analysis. They are investigated in an experimental setting. The application of the mixed Nash-equilibrium concept yields a rather counter-intuitive prediction which apparently contradicts the empirical data. The predictions of the Harsanyi-Selten-theory and Schelling’s “focal point theory” are in better accordance with the data. However, they do not account for the “diffusion-of-responsibility-effect” also observable in the context of an asymmetric VOD game.

Andreas Diekmann

Ten rules of bargaining sequences A boundedly rational model of coalition bargaining in characteristic function games

Abstract

The paper presents a first outline of a model of the bargaining process in characteristic function games. It is based on the conclusions from more than 1000 experimental games with free communication with 3, 4, 5 and sometimes more players. The essence of this approach is, that the bargaining process can be modeled as a sequence of states or proposals, each of which dominates the preceding one. And, that there are certain laws or rules of boundedly rational behavior, according to which players behave when they change coalitions. These rules are presented.

The concept is based on the following behavioral aspects:

• structure of sequences of proposals (Rule 1)
• phenomena of prominence (Rules 2, 3)
• the phenomenon of reciprocal loyalty (Rules 4, 5)
• conclusions from actions of others (attributed demands, Rule 6)

These phenomena restrict possible schemes of arguing to chains of proposals of which each dominates the preceding one. The space of all these chains is ordered by the natural inclusion of chains. It is a tree-shaped, finite set. On this space strategic considerations of the players are modeled by

• the recursive definition of stable states (Rule 7),
• the opportunity to break a coalition, and thereby force the game back to a former state of the process (Rule 8),
• principles of fairness, when forming the coalition of all players, and the stability of this coalition (Rules 9, 10).

Examples illustrate and motivate the conditions, and show phenomena that cannot be explained by traditional solution concepts.

Wulf Albers

Aspiration processing in multilateral bargaining: Experiment, theory and simulation

Abstract

A growing tradition in analysing bargaining behaviour uses an experimental setup to provide tools for controlling parameters of the underlying conflict. Game theory is used to provide a model of the conflict, to analyse the possible actions of the agents and the arguments that might be used within the bargaining process and to understand the results of the bargaining. In face-to-face bargaining, we usually find a great variety of actions that are not explainable and sometimes not even compatible with the usual rationality-assumptions on the agents. It is widely accepted that bounded rationality approaches are more adequate especially for the face-to-face settings. But it has not been really examined what concepts should be used in the corresponding models, and what tools are adequate for testing them. After a previous report on the use of the concept “social field” (Ostmann, 1992a), and another on a first attempt to explain bargaining results by aspirations and the social field (Ostmann, 1992b), this paper deals with aspiration processing in a more detailed way. We shall be exploring the limits of the concept “aspiration” with respect to explaining or predicting face-to-face bargaining processes.

Axel Ostmann

Resistance against mass immigration - An evolutionary explanation

Abstract

In Europe we face very different living standards in different countries. In former times such discrepancies did not induce mass migration since mobility was restricted, especially between Eastern and Western European countries. But now the relatively richer countries are confronted with mass immigration and also strong resistance against it. We will show that resistance against mass immigration can be explained as being genetically determined. Specifically, we will analyse a very simple game model of immigration with an undetermined preference parameter deciding whether an incumbent engages into opposition against mass immigration or not. It is shown that preference for fighting against mass immigration is the only evolutionarily stable strategy for all possible parameter constellations. In our view this has important political implications regardless whether one wants to argue for a more liberal immigration law or against it.

Werner Güth, Klaus Ritzberger

Backmatter