Belief Updates in Multiple Agent Systems

We give a model for a multi-agent system which describes how the knowledge and beliefs of agents should change when they receive new information. The formal tool we use for this description is a combination of modal and dynamic logic. Two core notions in our model of belief update are the expansion of the knowledge and beliefs of an agent, and the processing of new information by an agent. An expansion has been deened as the change in the knowledge and beliefs of an agent when it decides to believe an incoming formula while holding on to its current propositional beliefs. To prevent our agents from forming inconsistent beliefs they do not expand with every piece of information they receive. Instead of that, our agents remember their original beliefs (the beliefs they had before receiving any information) and every piece of information they receive. After every receipt of information they decide which (consistent) subset of the received information should be incorporated into their original beliefs. The decision which subset should be used is based on the trust the agents have in the sources of the received information. This procedure is called the processing of new information. We show that our model of belief update behaves in an intuitive way and that it is not sensitive to the critique on comparable models.