In the real world, usually, peoples are coming together for sharing their knowledge and talking from their good and bad experiences and more or less everybody has something to say. Although we cannot ignore anybody's knowledge but it's common sense to assign more weight on the most experienced person's knowledge when we are going to decide what we need to do based on consultation from people. The achievements of this research have the same philosophy, that everybody needs to be heard. Fuzzy integrals are one of the most powerful and flexible methods for hearing everybody's knowledge and extract knowledge which is useful for everybody. One of the challenges is that how to fairly answer the "what is the agents' expertise and how to determine the most and least expert agent?" question. To answer this question, in this thesis, we have proposed «the hypothesis of expertness» which defines a framework for "expertness criteria" definitions, and based on this framework we have introduced a new expertness criteria and showed that the defined framework and criteria are much more efficient than the state of the art criteria "Shortest Experienced Path" criteria. Also, the power of using fuzzy integrals for intelligence aggregation and non-additive measuring/knowledge is demonstrated Key words: Multi-agent Systems, Cooperative Learning, Reinforcement Learning, Fuzzy Integral, Non-additive Knowledge
