Environment-based agent behaviour modelling

An understanding of how autonomous agents behave with respect to their environment is key for the deployment and management of agent systems. Relatively complex behaviour is seen to arise in mobile agents even when their behaviour is based on simple rules. The addition of more advanced forms of reasoning to mobile agents has been shown to benefit both the agent and the system that is hosting the agent. This paper discusses the work that has been done in studying how mobile software agents behave in economics based environments. The specific economic environment examined is a public goods threshold contribution game involving multiple players with multiple public goods to contribute towards. Models exist that predict or explain human behaviour in similar environments. We have developed a system so that we may compare these models with the trial results of mobile agent behaviour The system developed consists of multiple mobile agents that play the threshold contribution game. By varying the reasoning system in the mobile agents, different behaviour patterns emerged providing us with trial results for further analysis.