This paper aims to analyze and exemplify some methodological implications on the way to conduct experiments related to the adoption of a cognitive approach in Economics. Many differences arise in relation to a more traditional way. In fact cognitive economics has strong descriptive attention and aims at beeing closer to reality than the mainstream. Besides the idea of representative agents is questioned. Different kind of experiments, differents analysis and new tools are so required. The paper proposes also some notes on the relation between experimental economics and simulation with artificial agents.