The Evolution of Coworker Networks

When selecting work team members several behavioral components concur. In this chapter we are interested in investigating the effects of these components in terms of team selection, agent aggregation and performance of groups. A computational model, together with a theoretical approach and the results of two human experiments where subjects interact in a similar game, allow us to identify some of the most important determinants. Our results suggest that the occurrence of two factors is crucial: the presence of leaders as aggregators of knowledge and agents being able to expand and improve their higher profit projects. It is particularly evident the threefold role the leaders have. First, they increase the social network of other agents making possible projects otherwise impossible. Second, they state the pace of a balanced growth in terms of social network, while taming the otherwise combinatorial explosion. Finally, they help selecting one of the theoretically possible equilibria.

Examining the Myth of Money with Agent-Based Modelling

This chapter presents an agent-based computational model of the emergence of money. It is based on classical economic theories of money, advocating that money is a symbol of credibility. The most interesting and mysterious feature of money is a departure of its face value from its intrinsic value. People accept and appreciate a piece of paper because it is believed as money. The model examines how such belief creates money in a society. Further more, by incorporating spatial activities of agents into the simulations, the model can examine various hypotheses which were difficult to be examined in previous approaches. The simulation results show that parameters such as credibility and communication between agents will affect the outcomes. The model not only provides the foundation for more generalized theory of money, but also demonstrates that agent-based modeling can be an effective tool to examine various hypotheses of social sciences.

Analyzing the Influences of Passive Investment Strategies on Financial Markets via Agent-Based Modeling

This chapter develops an agent-based model to analyze microscopic and macroscopic links between investor behaviors and price fluctuations in a financial market. This analysis focuses on the effects of Passive Investment Strategy in a financial market. From the extensive analyses, we have found that (1) Passive Investment Strategy is valid in a realistic efficient market, however, it could have bad influences such as instability of market and inadequate asset pricing deviations, and (2) under certain assumptions, Passive Investment Strategy and Active Investment Strategy could coexist in a Financial Market.

Fairness, Commitment, and Inequality

Conditional cooperation is a prominent explanation of reciprocal cooperation in repeated exchange. However, empirical evidence for commitment behavior indicates that people tend to build long-term cooperative relationships characterized by largely unconditional cooperation. Using an agent-based ecological model, earlier work showed that in competitive environments commitment can be a more successful strategy than fair reciprocity. We move further in two respects. First, we add the possibility of randomly mutating strategies under evolutionary pressures. Our results show the lack of evolutionary stable strategies but we also find that commitment strategies still outperform fairness strategies on average. Our second extension introduces inequality in individual capabilities. We find that inequality shifts the balance from commitment towards fairness strategies. Our explanation is that under inequality, strategies benefit from changing interaction partners from time to time because this gives more agents access to strong partners.

Probability Distribution Dynamics Explaining Agent Model Convergence to Extremism

This chapter studies continuous opinion models with extremists, and we use probability distribution models which approximate the behaviour of agents based models in order to explain their attractor patterns. The probability distribution is defined on a discrete grid in the opinion / uncertainty space. We compute the equations of probability flows between each couple of sites of the grid for different variants of the opinion influence model (bounded confidence, relative agreement and two others). The simulations show that the probability distribution models yield attractor patterns very similar to those obtained with the agent based models. Moreover, a study of the probability distribution evolution helps to better understand the process of convergence to single and double extreme attractors observed in agent based models.

Artificial Science

Science is important, not only in the knowledge it gives us, but also as an example of effective distributed problem solving that results in complex and compound solutions. This chapter presents a model of some of the social interactions in science, namely those between the body of published knowledge and the scientists’ individual knowledge. The structure of knowledge is modelled by a formal Hilbert system for a classical propositional logic. Individuals have limited selections of the total knowledge available which they use to derive new knowledge, which they may submit to the central journal to be published. This model shows how difficult it is to achieve the accumulation of knowledge as in science and also that publishing more does not necessarily lead to more important knowledge being discovered.

Cooperation as the Outcome of a Social Differentiation Process in Metamimetic Games

What are the principles underlying social differentiation ? Socio-economic models generally consider agents that pursue some particular ends given prior to their social activity. In this paper we propose an alternative in the framework of metamimetic games. We claim that the distribution of ends in a popultation is the outcome of social interactions and not only what drives them. We take the example of the prisoner’s dilemma in spatial games to illustrate how cultural co-evolution can lead to a spontaneous differentiation of ends in a population with a high level of cooperation. From this perspective, the question is not the traditional "How can altruists ’survive’ in a selfish worlds?" but rather to understand how heterogeneous ends can reinforce or limit each other to collectively entail the emergence of a social cooperative order.

Exploring the Effects of Campaigning Strategies for the Organisation of Collective Action Using Empirical Data

In a collective action, people act together with the intention of producing public goods. Public, or collective, goods are states or objects that benefit the many but only emerge if a sufficient number of persons make contributions. The present study explains de dynamics of participation in collective action campaigns by considering the interaction of different processes. With the resulting model it is possible to determine the optimal combination of diffusion measures for such a campaign. Before using the model for experimenting, we calibrate its parameters using data from a real world collective action. We find this to be a most important step in order to demonstrate that the model can be grounded empirically and to demonstrate the practical usefulness of simulation for consulting and design of real world processes. Finally, some “what if” scenarios reveal the model’s power of explanation and prediction.

Stress and Expectation in Double-Auction Market

This paper describes a multi-agent model of double-auction market in which simulations are led. In our study of market we focus on information processing and hence make assumptions about the cognitive use that agents do of these information. For some years now, experiments have been used to study auctions and now resulting data are used to make hy-pothesis about learning. We propose here simulations that are organised on the same model as experiments, as a succession of auctions session where each agent is either seller or buyer and has to exchange before the end. Communicaton is made of bids and asks that can be accepted by the others and lead to transactions. Our main result is the fact that we actu-ally obtain convergence although agents have no knowledge of others’ limit prices and only interact through a completely impersonal market. This corresponds to experimental data, which is a positive result in our search of the representation for economic rationality and is discussed methodologically in the paper.

Modelling Knowledge Production and Integration in Working Environments

This chapter introduces a formal model of a complex knowledge integration process named ‘thinking along’. Here the firm is modelled as a working environment consisting of agents arranged into work-practices, which provide the context for their interactions. The objective of the simulations reported here is to compare two different practice structures and test their effectiveness for solving problems by thinking along. To do so we will also introduce the notion of problem complexity as the basis for different experiments. From such comparison it emerged that complex problems are better tackled when practices group together agents with disparate skills (i.e. divisional practices) whereas simple problems can be more effectively addressed by organisational practices composed of agents with similar skills (i.e. functional practices). In either case, the simulated knowledge integration process played the dominant role.