Cooperation and selfishness in strategies for resource management

The optimal management of resources is an important research issue in many disciplines, such as biology, economics, political sciences, and computing, In biology, the successful management of resources has a critical impact o n the fitness of individuals. In this study, an adaptive multiagent system was used to investigate the effect of environmental conditions on the succe ss of distinct strategies for optimal resource management. Individuals were represented as virtual robots that interacted with their environment by th eir sensors and motors, Their behaviour mas controlled by a rule-based syst em, which acted in respect to the motivational state and sensorial input. T he effect of each rule was further specified by a set of parameters which w ere encoded as artificial genetic code. Individuals transferred their genes to future generations according to their success in accomplishing resource s. Scenarios with varying parameters such as distribution of resources and spatial constraints for competition mere tested. Interestingly, agents lear ned to cooperate and to harvest their resources in a moderate may under cer tain environmental conditions, thus avoiding population crashes due to unco ntrolled exploitation. Moreover, the results indicate how multiagent system s could be used to develop and test policies for improved man-made resource management. (C) 2000 Elsevier Science Ltd, All rights reserved.