© Copyright JASSS

  JASSS logo ----

François Bousquet, Robert Lifran, Mabel Tidball, Sophie Thoyer, Martine Antona (2001)

Agent-based modelling, game theory and natural resource management issues

Journal of Artificial Societies and Social Simulation vol. 4, no. 2,
<http://jasss.soc.surrey.ac.uk/4/2/0.html>

To cite articles published in the Journal of Artificial Societies and Social Simulation, please reference the above information and include paragraph numbers if necessary

Received: 17-Mar-01      Published: 31-Mar-01


1.1
This special issue of JASSS presents a set of papers selected from a workshop held in Montpellier in March 2000[1]. The objective was to stimulate discussions between researchers who use and develop game theory (GT) or multi-agent systems (MAS) especially in the field of natural resource management and environment.

1.2
The research questions addressed in the field of the environment and resource management are very often questions of collective decision-making. Several actors have to coordinate the sharing of a common environment and manage the externalities generated by individual decisions. GT modelling has been developed since the end of 50's to understand strategic interactions between actors. Over the last few years, agent-based models have been proposed to simulate different kind of interactions between agents and some applications in the field of resource management and the environment have been published. For many researchers, especially resource and environmental social scientists involved in modelling of social and economic interactions, it is not clear whether they should use GT or multi-agent systems. Is MAS a new technical tool to simulate interactions between players? Are the underlying concepts and hypothesis different? What are the possible linkages between these methods?

1.3
On the one hand, GT provides a methodological framework to analyse strategic interactions between rational agents' behaviour. Each agent, aware of these interactions, must choose an action in a set of feasible actions and perform it. The strategic interactions correspond to choices that explicitly and directly take into account the behaviour of others agents. GT studies the solutions of a game i.e. the set of actions performed by the agents. A very well known example of GT modelling in environmental issues is the modelling of the Tragedy of the Commons by Garett Hardin (Hardin, 1968) using the prisoner's dilemma game, although other games may be relevant to study collective action (Heckathorn, 1996). This game describes how increasing the exploitation of shared resources can be an individual rational choice and a dominant strategy for all players that lead to disastrous collective outcome, the over-exploitation of the resources. In a broad scientific debate, many GT models reinforcing or criticizing this theory have been proposed. One important reference is the research of Elinor Ostrom (Ostrom, 1990,Ostrom, 1994) in the field of Common-Pool Resources (CPR) (Stevenson, 1991). GT theory is used to analyse actions situations applied to various types of CPR problems and to draw precise and logical conclusions. These conclusions are compared with experimental results. For the authors the structure of the games serves as a first step in organizing behavior data from a CPR and as a first approximation to the strategic and deontic considerations related to CPR such as forests, groundwater basins, etc. The methodological approach is to link game theory, institutional analysis and laboratory experiments.

1.4
On the other hand, many agent based models, have been developed over the last ten years, in two broad categories : models such as the SugarScape address the problem of resource management in a very theoretical way (Epstein, 1996). The objective is to emphasize the potentialities of MAS to study co-ordination between agents and the impact on the system dynamics. More applied models have also been developed in recent years (Bousquet, 1994;Lansing, 1994). These models are used for a better understanding of a present environmental problem or to explore past scenarios (Doran, 1993;Kohler, 2000). They focus on ecosystem management and land use change. These models often establish linkages between a dynamic resource and a society of interacting agents which have to coordinate to manage it. Models have also been used to analyse results in CPR experimental research and have already been published in JASSS (Deadman, 2000). The simulations explore the configuration of institution and individual behavioural characteristics so as to observe the group level outcomes.

1.5
Two communities of researchers have focused on developing the linkages between GT and MAS. Firstly Rosenchein and Zlotkin (Rosenchein, 1994) have conducted studies about agreements between computers (truly rational). The objective of the computer scientist is to design protocols for specific domains that will get the agents to interact in useful ways. Then the question is : for a given protocol what will the appropriate strategies be? In the same community Durfee has used GT to design individual strategies of agents taking into account the others players actions (Durfee, 1999). That is very close to the GT. The purpose is to use MAS as method to make the coordination practical, through communication for instance.

1.6
Secondly, there have been several publications on the use of multi-agent systems to simulate strategic dynamic interactions between agents. Hoffman made a review of part of the literature in JASSS (Hoffmann, 2000) with a special emphasis on the Axelrod's "evolution of cooperation"(Axelrod, 1984). This research has already been commented by Binmore in JASSS. The use of the evolutionary metaphor by researchers coming from the field of artificial life (Lindgren, 1997) strengthens the linkages with evolutionary game theory (Weibull, 1995).

1.7
The research context of the organizers and participants of the workshop, which led to this JASSS issue, is different. As modellers (using GT or MAS) our purpose is neither to build strategic and co-ordinated machines, nor to understand how evolution led to specific situations such as the co-adaptation of a set of strategies. Our objective is to observe and simulate societies where agents share a common resource, have representations, make individual and collective decisions, negotiate, exchange at given spatial and temporal scales. This leads to a better understanding of existing societies and we try to use our models for better co-ordination between actors.

1.8
Two kinds of papers were presented at the workshop. A first set of papers is composed of modelling applications on environmental issues. A second set of papers presents experiments and discussions about relevant issues for the research on collective management of the environment: some compare GT and MAS, some make proposals on the modelling of individual rationality and social control. A web site presents the objectives and gives the abstracts of most presentations:http://www.ensam.inra.fr/ESR/sma_tdj/. To facilitate the process each presented paper has been reviewed by a participant from the "challenging side" and a written comment has been presented together with the main paper.

1.9
Three categories of selected papers are presented in this special issue. Firstly three authors present papers comparing the two approaches. They analyse the arguments of each side and propose MAS as an alternative or as complement to GT. Secondly, three authors propose general ideas or experiences of intervention in the collective decision-making process. One paper proposes a classification of intervention strategies and two papers present experiences in the use of negotiation models or in the use of role games. Lastly, two papers address the question of MAS modelling for the dynamic linkages between individual rationality and social organization. The problems addressed by these two papers are of general importance in the domain of common pool resource management. Ostrom ( Ostrom, 1998) presents the questions treated by these papers (norms on the one hand and reciprocity, reputation and trust on the other hand) as the research issues for the second generation model of rationality.

1.10
Methodological comparison between MAS and GT :

1.11
The use of models to achieve co-operative environmental management.

1.12
Individual rationality and the modelling of social dynamics. These two papers can be viewed as two extremes on the society modelling debate: what is the relative weight of the individual decision-making process and the interaction system?

1.12
Rules of encounter (Rosenchein, 1994): the title of this pioneer book could be a relevant title for the attempt we made in favouring the interactions between GT and MAS users. By provoking this meeting, with a special emphasis on natural resource management, we tried to facilitate scientific discussions between these two communities. There are some strong opposition between researchers, some denying the scientific status of simulation, some denying the relevancy of strategic computation. However we made a step by putting more emphasis on questions addressed by both research groups such as group emergence and conditions for stability, models and experimentation in social sciences, individual rationality and organization of exchanges. By addressing collectively these questions, opposing or re-enforcing each other, one can expect significant advances on crucial issues in the field of natural resource and environmental management.


* Notes

1 The workshop was organised by INRA and CIRAD with support of the Region Languedoc-Roussillon.

* REFERENCES

AXELROD R. (1984). The evolution of cooperation. New York, Basic Books.

BOUSQUET F. (1994). "Distributed artificial intelligence and object-oriented modelling of a fishery." Mathematical Computer Modelling 2018: 97-107.

DEADMAN P. J., E. Schlager, et al. (2000). "Simulating Common Pool Resource Management Experiments with Adaptative Agents employing Alternate Communication Routines." Journal of Artificial Societies and Social Simulation 3(2) <http://www.soc.surrey.ac.uk/jasss/3/2/2.html>.

DORAN J. and M. Palmer (1993). The EOS Project : Integrating Two Models of Paleolithic Social Change. Artificial Societies. N. Gilbert and R. Conte, UCL Press.

DURFEE E. (1999). "Practically coordinating." AI Magazine Spring 1999: 99-116.

EPSTEIN J. and R. Axtell (1996). Growing Artificial Societies. Social Science from the Bottom Up, Brookins Institution Press/ The MIT Press.

HARDIN G. (1968). "The tragedy of the commons." Science 162: 1243-1248.

HECKATHORN D. D. (1996). "The dynamics and dilemmas of collective action." American Sociological Review 61: 250-277.

HOFFMANN R. (2000). "Twenty years on: The evolution of cooperation revisited." Journal of Artificial Societies and Social Simulation 3(2) <http://www.soc.surrey.ac.uk/jasss/3/2/forum/1.html>.

KOHLER T., J. Kresl, et al. (2000). Be there and then: a modeling approach to settlement determinants and spatial efficiency among late ancestral pueblo populations of the Mesa Verde region, US SouthWest. Dynamics in human and primate societies. T. Kohler and G. Gumerman, Oxford University Press: 145-178.

LANSING J. S., Kremer J.N. (1994). Emergent properties of Balinese water temple networks: coadaptation on a rugged fitness landscape. Artificial life III, Santa Fe, Addison-Wesley.

LINDGREN K. (1997). Evolutionary Dynamics in Game Theoretic Models. The economy as an evolving complex system II. W. A. a. S. D. a. D. Lane, Santa Fe Institute, Addison-Wesley.

OSTROM E. (1990). Governing the Commons, Cambridge University Press.

OSTROM E., R. Gardner, et al. (1994). Rules, games, and common-pool resources. Ann Harbor, University of Michigan Press.

OSTROM E. (1998). "A behavioral approach to the rational choice theory of collective action." American political Science Review 92(1): 1-22.

ROSENCHEIN J. and G. Zlotkin (1994). Rules of encounter. designing conventions for automates negotiation among computers, MIT Press.

STEVENSON G. G. (1991). Common Property Economics. A General Theory and Land Use Applications, Cambridge University Press.

WEIBULL J. (1995). Evolutionary Game theory, MIT Press.

----

ButtonReturn to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, 2001