The observation of the actual behavior by economic decision makers in the lab and in the field justifies that bounded rationality has been a generally accepted assumption in many socio-economic models. The goal of this paper is to illustrate the difficulties involved in providing a correct definition of what a rational (or irrational) agent is. In this paper we describe two frameworks that employ different approaches for analyzing bounded rationality. The first is a spatial segregation set-up that encompasses two optimization methodologies: backward induction and forward induction. The main result is that, even under the same state of knowledge, rational and non-rational agents may match their actions. The second framework elaborates on the relationship between irrationality and informational restrictions. We use the beauty contest (Nagel, 1995) as a device to explain this relationship.