This paper focuses on the link between the economic conceptions of rationality and learning. Traditionally, most economists believe that learning is just a way for agents to become fully rational. But being fully rational cannot describe a process, for there is only one way to be rational in the economic sense of the term. Therefore, what economists have in mind is not the process of learning, but the result of learning: ‘a fully rational agent'. Heterodox rationality conceptions such as the Simonian model of bounded rationality seem more compatible with the idea of learning. Bounded rationality implies that agents may act differently to the same stimulus; it is therefore compatible with the idea of diversity, one of the foundations of the evolutionary logic. But following Simon, learning should not be considered as a creative process that allows a lot of diverse answers. If diversity exists in the agents' behaviors, the way they learn appears to be unique. As a consequence, learning should decrease the strength of the selection forces, both processes being contradictory (Dosi et al. 2003). Our paper aims to overcome this contradiction by showing how intentionality and identity, and more broadly Fransisco Varela's ‘enaction' theory, can help to invent a concept of ‘rational learning' that is compatible with the evolutionary logic.
