A description of the area of Agent Based Social Simulation (ABSS) from a computer scientist's perspective is presented. We begin by defining ABSS by positioning it with respect to the three research areas that it is related to, i.e., agent-based computing, the social sciences, and computer simulation. We then discuss the role of ABSS and how it may aid cross-fertilisation between these areas.

Keywords:

Agent-based Computing; Agent-based Social Simulation; Computer Simulation; Social Science

What is Agent-Based Social Simulation?

1.1

One way of characterising the research area of Agent-Based Social Simulation (ABSS) is that it constitutes the intersection of three scientific fields, namely, agent-based computing, the social sciences, and computer simulation (see Figure 1).

Figure 1. The three areas constituting ABSS and their interrelationship

1.2

Agent-based computing is a research area mainly within computer science and includes, e.g., agent-based modelling, design, and programming. By the social sciences we here refer to a large set of different sciences that study the interaction among social entities, e.g., social psychology, management, policy, and some areas of biology. Finally, computer simulation concerns the study of different techniques for simulating phenomena on a computer, e.g.: discrete event, object-oriented, and equation-based simulation. The phenomenon simulated is an event or a sequence of events in a natural or an artificial system (or a combination of these), which could be either existing or non-existing at the time of the simulation. The reason for doing computer simulations is usually to gain a deeper understanding of the phenomenon, e.g., "debug" models of systems, predicting future behaviour, and performing experiments that cannot be carried out in reality for some reason or another. Although computer simulation can be seen as a computer science sub-area, much development is carried out within application areas, such as physics, mechanical engineering, biology, and to some extent even within the social sciences.

1.3

While the main focus of ABSS according to the characterisation outlined here is in the area where all the three fields intersect, much interesting work is carried out in the areas where just two of the fields intersect. For instance, the intersection between the social sciences and agent-based computing concerns Social Aspects of Agent Systems (SAAS) and includes the study of norms, institutions, organisations, co-operation, competition, etc. The activities belonging to the intersection between computer simulation and agent-based computing are often labelled Multi Agent Based Simulation (MABS) and study the use of agent technology for simulating any phenomena on a computer. Finally, the intersection between the social sciences and computer simulation is typically called Social Simulation (SocSim) and corresponds to the simulation of social phenomena on a computer using any simulation technique and is typically using simple models of the simulated social entities, e.g., cellular automata and objects, that are able to perform only very basic interaction. This type of models may be contrasted to the software agents used in MABS with possibly rich cognitive models and sophisticated communication languages and interaction mechanisms.

Figure 2. The intersections of the three areas defining ABSS

1.4

According to the view adopted here, ABSS can be said to investigate the use of agent technology for simulating social phenomena on a computer. However, this is a quite narrow definition and in some situations a wider definition may be useful. As most of the work in all the intersection areas (SAAS, MABS, and SocSim) is clearly relevant to ABSS, a natural extension would be to include them. However, it may be clarifying also to point out areas that clearly do not belong to ABSS, e.g.:

• social science that does not include an element of either agent technology or computer simulation,
• agent technology that does not include an element of either social science or computer simulation, and
• computer simulation that does not include an element of either agent technology or social science.

The Role of Agent-Based Social Simulation

2.1

Based on the characterisation above we may conclude that the main role of ABSS is to provide models and tools for agent-based simulation of social phenomena, and to apply these in different areas. However, through its inter-disciplinary flavour, ABSS has a unique potential for providing cross-fertilisation between the participating fields of research. (See Figure 3.) In this section we will study these possibilities.

Figure 3. Contributions mediated by ABSS from and to agent-based computing

Social Science and Agent-Based Computing

2.2

The subject of social science is a set of very large scale systems that are also very "messy" (Moss 2000). That is, they have ill-defined or unknown boundaries and individuals comprising the systems face constraints that are beyond their information processing capacities to define or to use in reaching decisions to act. In those areas where agent-based computing research is intended to provide the techniques and methodologies to build and support very large scale software systems, a natural source of analogy is the very large scale social systems. For instance, social systems are able to evolve and adapt to attain a certain robustness, which is a desirable feature of software systems.

2.3

The social sciences may benefit from agent based computing in several ways. For instance, when software engineers build multi-agent systems they implement agents that interact in ways that determine the properties of the whole software system. The ways in which macro-level system properties emerge from such interaction at the micro-level may be used to inform the development of models to describe actual social systems.

2.4

There is a second source of synergy between social science and agent based computing in the development and use of formalisms to specify software agents that are intended also to inform a new social theory (cf. Conte and Gilbert 1995). The formal structure of agent based computing clearly provides a supportive environment for the application of logical formalisms and the formalisms developed with a view to a new social theory are frequently found useful in specifications of agents for purposes of engineering multi-agent systems.

Social Science and Computer Simulation

2.5

The links between social science and computer simulation are largely methodological in character and the direct benefit seems in favour of social science. However, the concept of individual-based simulation has been partially developed within social science, e.g., dynamic micro simulation (Gilbert and Troitzsch 1999).

2.6

Social scientists have begun to convert social theories to computer programs. It is then possible to simulate social processes and carry out "experiments" that would otherwise be impossible. Simulation is useful when the phenomenon to be studied is not directly accessible or is difficult to observe directly. Also, computer simulation has been used as a method to clarify sociological theories. As Gilbert (1994) has pointed out, one advantage of simulation is that it is hard to do. To create a simulation model from a sociological theory stated in a natural language, all assumptions etc must be described explicitly and formally. Also, in order to run a simulation, every parameter in the simulation model must be given a value. You simply cannot be vague about what is being assumed.

2.7

However, simulation of complex social processes involves the estimation of many parameters, which can be difficult. Although using computer simulation for prediction of social phenomena, e.g. the consequences of a particular social policy, is usually difficult, it may be used for analysing and understanding the phenomena.

Agent-Based Computing and Computer Simulation

2.8

The contribution from agent based computing to the field of computer simulation mediated by ABSS is a new paradigm for the simulation of complex systems with much interaction between the entities of the system. As ABSS, and other micro simulation techniques, explicitly attempts to model specific behaviors of specific individuals, it may be contrasted to macro simulation techniques that are typically based on mathematical models where the characteristics of a population are averaged together and the model attempts to simulate changes in these averaged characteristics for the whole population. Thus, in macro simulations, the set of individuals is viewed as a structure that can be characterized by a number of variables, whereas in micro simulations the structure is viewed as emergent from the interactions between the individuals. Parunak et al. (1998) compared these approaches and pointed out their relative strengths and weaknesses. They concluded that "...agent-based modeling is most appropriate for domains characterized by a high degree of localization and distribution and dominated by discrete decision. Equation-based modeling is most naturally applied to systems that can be modeled centrally, and in which the dynamics are dominated by physical laws rather than information processing."

2.9

Possible benefits to agent based computing from computer simulation includes methods for evaluation of multi agent systems or for training future users of the system (cf. Davidsson 2000). Many new technical systems are distributed and involve complex interaction between humans and machines. The properties of ABSS makes it especially suitable for simulating this kind of systems. The idea is to model the behaviour of the human users in terms of software agents.

Conclusions

3.1

We have argued that ABSS lies in the intersection of three research areas, agent-based computing, the social sciences, and computer simulation, and that it provides a unique potential for cross-fertilisation between these areas. Although this potential has been partially explored, we believe that there are still many new and promising directions to explore.

References

CONTE, R. and Gilbert, N. (1995). Introduction: Computer Simulation for Social Theory. In Gilbert, N. and Conte, R. (Eds.) Artificial Societies: the Computer Simulation of Social Life, UCL Press.

DAVIDSSON , P. (2000). Multi Agent Based Simulation: Beyond social simulation, In Moss, S. and Davidsson, P. (Eds.) Multi Agent Based Simulation, Springer Verlag LNCS series, Vol. 1979.

GILBERT, N. (1994). Computer simulation of social processes, Social Research Update, Issue 6, Department of Sociology, University of Surrey, England.

GILBERT, N. and Troitzsch, K.G. (1999). Simulation for the Social Scientist, Open University Press.

MOSS, S. (2000). Messy Systems - The Target for Multi Agent Based Simulation, In Moss, S. and Davidsson, P. (Eds.) Multi Agent Based Simulation, Springer Verlag LNCS series, Vol. 1979.

PARUNAK, H.V.D., Savit, R. and Riolo, R.L. (1998). Agent-Based Modeling vs. Equation-Based Modeling: A Case Study and Users' Guide. In Sichman, J.S., Conte, R., and Gilbert, N. (Eds.), Multi-Agent Systems and Agent-Based Simulation, Springer Verlag.

Return to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, [2002]