Modelling Decentralised Decision Making

A set of agents t human or computer process} tire supposed lO dec~dc on an action given z, lmite set of possible straleglc., when the background inl’t~rnl:~titm is vague or numerically imprecise. The agents evaluate the stralegie.~ :rod communicate their opinions aboul their utdities, their opinion.’, about the credibilities of the reports of the other agents. :is well a.,, their opinions of the reliabilities o1" the other agents. A ..,igniticant property of the theory is that it admits the representation of imprecise inform.’~tion at all stages. The strategies are evaluated relative to generalisations of the p,’inciple of maxiraising the expected utility. It is also demonstrated how decision trees can he integrated into the framework :me we sugge,~t :m evaluation rule taking int¢~ act’~unt all strategies. credibilities, reliabilities, probabilities and utilitie.~ involved in the framework. This is done ~ itht~ut presuppo.~ing the exi.’,lencc o1 a central co-ordinating agent that e~.aluates the thfferent repurts of the agents. In :t distributed context there are no tl priori reasons Ibr such art agent and v..e shL~w how it f;.tn be eliminated in our framework.