The effects of heterogeneous interaction and risk attitude adaptation on the evolution of cooperation

This paper addresses the evolution of cooperation in a multi-agent system with agents interacting heterogeneously with each other based on the iterated prisoner’s dilemma (IPD) game. The heterogeneity of interaction is defined in two models. First, agents in a network are restricted to interacting with only their neighbors (local interaction). Second, agents are allowed to adopt different IPD strategies against different opponents (discriminative interaction). These two heterogeneous interaction scenarios are different to the classical evolutionary game, in which each agent interacts with every other agent in the population by adopting the same strategy against all opponents. Moreover, agents adapt their risk attitudes while engaging in interactions. Agents with payoffs above (or below) their aspirations will become more risk averse (or risk seeking) in subsequent interactions, wherein risk is defined as the standard deviation of one-move payoffs in the IPD game. In simulation experiments with agents using only own historical payoffs as aspirations (historical comparison), we find that the whole population can achieve a high level of cooperation via the risk attitude adaptation mechanism, in the cases of either local or discriminative interaction models. Meanwhile, when agents use the population’s average payoff as aspirations (social comparison) for adapting risk attitudes, the high level of cooperation can only be sustained in a portion of the population (i.e., partial cooperation). This finding also holds true in both of the heterogeneous scenarios. Considering that payoffs cannot be precisely estimated in a realistic IPD game, simulation experiments are also conducted with a Gaussian disturbance added to the game payoffs. The results reveal that partial cooperation in the population under social comparison is more robust to the variation in payoffs than the global cooperation under historical comparison.     

Efficiency of continuous double auctions under individual evolutionary learning with full or limited information

In this paper we explore how specific aspects of market transparency and agents’ behavior affect the efficiency of the market outcome. In particular, we are interested whether learning behavior with and without information about actions of other participants improves market efficiency. We consider a simple market for a homogeneous good populated by buyers and sellers. The valuations of the buyers and the costs of the sellers are given exogenously. Agents are involved in consecutive trading sessions, which are organized as a continuous double auction with order book. Using Individual Evolutionary Learning agents submit price bids and offers, trying to learn the most profitable strategy by looking at their realized and counterfactual or “foregone” payoffs. We find that learning outcomes heavily depend on information treatments. Under full information about actions of others, agents’ orders tend to be similar, while under limited information agents tend to submit their valuations/costs. This behavioral outcome results in higher price volatility for the latter treatment. We also find that learning improves allocative efficiency when compared to outcomes with Zero-Intelligent traders.     

Spatial Evolution of Social Norms in a Common-Pool Resource Game

We study the conditions for the emergence of cooperation in a spatial common-pool resource (CPR) game. We consider three types of agents: cooperators, defectors and enforcers. The role of enforcers is to punish defectors for overharvesting the resource. Agents are located on a circle and they only observe the actions of their two nearest neighbors. Their payoffs are determined by both local and global interactions and they modify their actions by imitating the strategy in their neighborhood with the highest average payoffs on average. Using theoretical and numerical analysis, we find a large diversity of equilibria to be the outcome of the game. In particular, we find conditions for the occurrence of equilibria in which the three strategies coexist. We also derive the stability of these equilibria. Finally, we show that introducing resource dynamics in the system favors the occurrence of cooperative equilibria.      

Endogenous games and equilibrium adoption of social norms and ethical constraints

We consider a situation in which games are formed endogenously in two senses: (1) there is a pregame in which agents choose to learn a subset of all feasible strategies and can then employ only these strategies in subsequent play, and (2) agents choose their game partners through a costly search process. We show that at any subgame perfect equilibrium, agents will constrain their action sets in the pregame in such a way that a single social norm prevails. Thus, all agents in a society will abide by the same ethical standard, although what standard this will be cannot be predicted. We also show that these are essentially the only SPE outcomes. We suggest that this provides at least a partial explanation for experimental observations that agents apparently choose strategies that do not maximize their payoffs.     

Coordination and delay in global games

What is the effect of offering agents an option to delay their choices in a global coordination game? We address this question by considering a canonical binary action global game, and allowing players to delay their irreversible decisions. Those that delay have access to accurate private information at the second stage, but receive lower payoffs. We show that, as noise vanishes, as long as the benefit to taking the risky action early is greater than the benefit of taking the risky action late, the introduction of the option to delay reduces the incidence of coordination failure in equilibrium relative to the standard case where all agents must choose their actions at the same time. We outline the welfare implications of this finding, and probe the robustness of our results from a variety of angles.     

Naive learning and cooperation in network experiments

In this paper we study learning and cooperation in repeated prisoners' dilemmas experiments. We compare interaction neighbourhoods of different size and structure, we observe choices under different information conditions, and we estimate parameters of a learning model.We find that naive imitation, although a driving force in many models of spatial evolution, may be negligible in the experiment. Naive imitation predicts more cooperation in spatial structures than in spaceless ones—regardless whether interaction neighbourhoods have the same or different sizes in both structures. We find that with some interaction neighbourhoods even the opposite may hold.     

Conformism and diversity under social learning

Summary. When there are competing technologies or products with unknown payoffs an important question is which technology will prevail and whether technologies with different payoffs can coexist in the long run. In this paper, we use a social learning model with local interactions to study this question. We show that the adoption of technologies as well as the prospects of conformism/diversity depend crucially on the nature of interaction between individuals and the heterogeneity of preferences in a society. Received: May 10, 1999; revised version: February 4, 2000     

Negative Externalities and Evolutionary Implementation

We model externality abatement as an implementation problem. A social planner would like to ensure efficient behaviour among a group of agents whose actions are sources of externalities. However, the planner has limited information about the agents' preferences, and is unable to distinguish individual agents except through their action choices. We prove that if a concavity condition on aggregate payoffs is satisfied, the planner can guarantee that efficient behaviour is globally stable under a wide range of behaviour adjustment processes by administering a variable pricing scheme. Through a series of applications, we show that the concavity condition is naturally satisfied in settings involving negative externalities. We conclude by contrasting the performance of the pricing mechanism with that of a mechanism based on direct revelation and announcement dependent forcing contracts.     

Learning with noiseless information and payoff-relevant signals

Summary. In models of active learning or experimentation, agents modify their actions to affect the distribution of a signal that provides information about future payoffs. A standard result in the experimentation literature is that agents experiment, if at all, to increase information. This finding is a direct consequence of Blackwell's theorem: one experiment is more informative than another if and only if all expected utility maximizers prefer to observe the first. Blackwell's theorem presupposes, however, that the observed signal only conveys information and does not directly affect future payoffs. Often, however, signals are directly payoff relevant, a phenomenon that we call signal dependence. For example, if a firm is uncertain about its demand and uses today's sales as a signal of tomorrow's demand, then that signal may also directly affect tomorrow's profit if the good is durable or if consumers form consumption habits. Datta, Mirman and Schlee [9] and Bertocchi and Spagat [4] show that, if the signal is payoff relevant, experimentation may indeed reduce information. Here we show that, despite the inapplicability of Blackwell's Theorem, agents always experiment to increase information if the information structure is noiseless: given the true value of the unknown parameter, the signal realization is deterministic. We then apply our framework to analyze Lazear's [16] model of retail clearance sales, a model with both signal dependence and noiseless information. Received: February 19, 1999; revised version: August 11, 1999     

Bayesian learning in social networks

We extend the standard model of social learning in two ways. First, we introduce a social network and assume that agents can only observe the actions of agents to whom they are connected by this network. Secondly, we allow agents to choose a different action at each date. If the network satisfies a connectedness assumption, the initial diversity resulting from diverse private information is eventually replaced by uniformity of actions, though not necessarily of beliefs, in finite time with probability one. We look at particular networks to illustrate the impact of network architecture on speed of convergence and the optimality of absorbing states. Convergence is remarkably rapid, so that asymptotic results are a good approximation even in the medium run.     

Network Games

In contexts ranging from public goods provision to information collection, a player's well-being depends on his or her own action as well as on the actions taken by his or her neighbours. We provide a framework to analyse such strategic interactions when neighbourhood structure, modelled in terms of an underlying network of connections, affects payoffs. In our framework, individuals are partially informed about the structure of the social network. The introduction of incomplete information allows us to provide general results characterizing how the network structure, an individual's position within the network, the nature of games (strategic substitutes vs. complements and positive vs. negative externalities) and the level of information shape individual behaviour and payoffs.     

Learning from private information in noisy repeated games

We study the perfect type-contingently public ex-post equilibrium (PTXE) of repeated games where players observe imperfect public signals of the actions played, and both the payoff functions and the map from actions to signal distributions depend on an unknown state. The PTXE payoffs when players are patient are determined by the solutions to a family of linear programming problems. Using this characterization, we develop conditions under which play can be as if the players have learned the state. We provide a sufficient condition for the folk theorem, and a characterization of the PTXE payoffs in games with a known monitoring structure.     

A spectrum of equilibration processes in game theory

In game theory, four dynamic processes converging towards an equilibrium are distinguished and ordered by way of agents' decreasing cognitive capacities. In the eductive process, each player has enough information to simulate perfectly the others' behavior and gets immediately to the equilibrium. In epistemic learning, each player updates his beliefs about others' future strategies, with regard to their sequentially observed actions. In behavioral learning, each player modifies his own strategies according to the observed payoffs obtained from his past actions. In the evolutionary process, each agent has a fixed strategy and reproduces in proportion to the utilities obtained through stochastic interactions. All along the spectrum, longer term dynamics makes up for weaker rationality, and physical relations substitute for mental interactions. Convergence, if any, is towards an always stronger equilibrium notion and selection of an equilibrium state becomes more sensitive to context and history. The processes can be mixed if associated to different periods, agents or mechanisms and deepened if obtained by formal reasoning principles.     

Renegotiation-proof relational contracts

We study infinitely repeated two-player games with perfect monitoring and assume that each period consists of two stages: one in which the players simultaneously choose an action and one in which they can transfer money to each other. In the first part of the paper, we derive simple conditions that allow a constructive characterization of all Pareto-optimal subgame perfect payoffs for all discount factors. In the second part, we examine different concepts of renegotiation-proofness and extend the characterization to renegotiation-proof payoffs.     

Social welfare for an economy of angelic agents

This paper analyzes an economy in which all agents are pursuing the common good (or social welfare) but choices are decentralized, i.e., each agent can choose his/her action in the set of the actions that he/she can perform. One wonders if it is enough the common goal of maximizing social welfare to their will be achieved. The paper examines both the cases in which the choice made by each agent does not directly influence those of other agents, as in the competitive equilibrium analysis, and the case in which there is a direct influence, as in the game theory analyses. In the first case, we get that the common goal of maximizing social welfare is not enough to reach it, but it is necessary to coordinate the actions of individual agents by extending information to redistribute initial endowments and by introducing an appropriate social organization. We get the maximum social welfare without further intervention for the cases describable with the theory of games, but only for games of complete information. If the information is incomplete, some further coordination is generally required.     

Between-game rule learning in dissimilar symmetric normal-form games

Rule learning posits that decision makers, rather than choosing over actions, choose over behavioral rules with different levels of sophistication. Rules are reinforced over time based on their historically observed payoffs in a given game. Past works on rule learning have shown that when playing a single game over a number of rounds, players can learn to form sophisticated beliefs about others. Here we are interested in learning that occurs between games where the set of actions is not directly comparable from one game to the next. We study a sequence of ten thrice-played dissimilar games. Using experimental data, we find that our rule learning model captures the ability of players to learn to reason across games. However, this learning appears different from within-game rule learning as previously documented. The main adjustment in sophistication occurs by switching from non-belief-based strategies to belief-based strategies. The sophistication of the beliefs themselves increases only slightly over time.     

Information structures and decentralizability of equilibria

Summary Decentralizability with respect to an equilibrium concept means that those equilibria for an extensive game and its agent normal form game coincide for any given payoffs. We consider decentralizability of Nash equilibrium, subgame perfect equilibrium, and perfect equilibrium. For each equilibrium concept we give a necessary and sufficient condition on the information structure of an extensive game for decentralizability to hold. When it holds it does not matter if agents with the same objectives decide independently or have someone coordinate their actions.The author thanks Satish Chand, Mamoru Kaneko, Akira Okada and participants at seminars at the Australian National University, Kyoto University, University of Tsukuba, and The First Decentralization Conference in Japan held at Keio University for valuable suggestions and comments.     

The cost of useful knowledge and collective action in three fisheries

In a complex environment knowledge is valuable and its acquisition is costly; as a result people are careful about what to learn and how to learn it. We suggest that the dynamics of the “local” environment strongly influences the method that individuals choose to acquire useful knowledge and is one of the principal determinants of the way they compete and cooperate. We focus on the way different environments lead to different costs, especially the relative opportunity costs, of search and communication and, consequently, to the emergence of different patterns of persistent cooperation and competition. In predictably regular and in predictably random environments, the cost of autonomous search is low and little social structure emerges. In complex environments, the relative costs of communication are high, leading to persistent social structure. Our presumption is that the characteristics of the emergent, or informal, social structure are a major determinant of successful collective action. We investigate the hypothesis through a comparison of three fisheries in which the costs of acquiring useful knowledge are different. Because of these differences, fishers' acquisition of useful knowledge leads to different social structure and different preconditions for successful collective action in each fishery. The lobster fishery is characterized by strong collective action and appears sustainable; the urchin and groundfisheries, worked by the same communities, are not even though almost all their participants are familiar with and often participate in the lobster fishery.     

Dynamic coordination with timing frictions: Theory and applications

We present a general framework of dynamic coordination with timing frictions. A continuum of agents receive random chances to choose between two actions and remain locked in the selected action until their next opportunity to reoptimize. The instantaneous utility from each action depends on an exogenous fundamental that moves stochastically and on the mass of agents currently playing each action. Agents' decisions are strategic complements and history matters. We review some key theoretical results and show a general method to solve the social planner's problem. We then review applications of this framework to different economic problems: network externalities, statistical discrimination, and business cycles. The positive implications of these models are very similar, but the social planner's solution points to very different results for efficiency in each case. Last, we review extensions of the framework that allow for endogenous hazard rates and ex ante heterogeneous agents.     

Think Globally,Teach Locally: Experiencing the Foreign Aid Debate Through Service Learning

Abstract

Novel curricular strategies are required if institutions want all students to actively experience the benefits of global knowledge and civic engagement, as financial and practical commitments frequently make study abroad inaccessible to many students. In this paper, we outline an innovative service-learning course, where local action coupled with an international target, offered a parallel and novel learning strategy that capitalized on the strengths of experiential education, while providing a practical and more inclusive student engagement opportunity available to a larger subset of students. We also describe our teaching strategy, which emphasizes the social context of the classroom: discovery, self-exploration, and shared learning. Together, service learning and a critical pedagogy can better help students relate to the otherwise abstract processes of foreign aid. In 2013 and 2014, approximately 30 undergraduate students participated in a student-led outreach project soliciting bicycle donations to support human development efforts in Uganda and Ghana. In addition to making reasonable progress toward learning outcomes during the two-year pilot, we found that the everyday challenges our students encountered in their service-learning project were microcosms for some of the large-scale, global challenges that foreign aid delivery faces.