In recent years, economists have shown an increasing interest in learning processes. It seems to be widely acknowledged that learning is a basic feature of human behavior and therefore must be important in all economic analysis that aims to advance our understanding of that behavior.
This is especially the case for a rational choice approach that tries to comprehend human behavior not only in a narrowly defined economic context, thereby contributing to social sciences more generally. Despite the basic and longstanding acceptance of its importance (see e.g., Schumpeter above), the phenomenon of human learning has mostly been neglected in economic theory. As will be outlined in Section 2, learning has traditionally not been recognized as a truly behavioral feature but was treated as a black box of automated adjustment, and it is only since the early eighties that learning has been analyzed more extensively by economists.
The shift from models that (implicitly) assume perfect adjustment of behavior to approaches that explicitly model adjustment processes originates on one hand in the rational expectations literature where theorists felt the need for a better understanding of how expectations are formed. The occurrence of multiple equilibria in many of these models led to theoretical efforts to model rational expectations equilibria as the result of some adjustment process. On the other hand, there was also an increasing demand from game theorists for a more comprehensive theory that allows to distinguish between multiple equilibria in games. After more than one decade of research, the question arises what can be learned from this learning literature.
It will be argued that most current approaches that stipulate learning mechanisms contribute to refinements of theoretical answers to theoretical questions in an often axiomatic or ad hoc way, but avoid to address truly behavioral questions. That is, they fail to enhance our understanding of learning as a basic process in economic behavior, a process that helps the individual to adjust to changing circumstances under a variety of conditions. Section 3 discusses the methodology involved with current approaches and presents the methodology of an alternative approach: contrary to standard theory that assumes learning to be perfect – and current approaches that study specific learning mechanisms either theoretically or experimentally – an alternative methodology is to identify the determinants for learning processes.[4]
Section 4 presents an approach of contingent learning (CLA) based on this idea and concludes with behavioral[5] hypotheses about the influence of situational factors on learning. The goal of this approach is on both levels of analysis. On the positive level it provides a framework that permits to account for the contingencies of a given situation with respect to learning in a way that is methodologically compatible with traditional economics. Thereby, the CLA aims to enlarge the economists’ analytical instrumentarium with theory elements that allow for deeper explanations and better predictions in cases where behavioral adjustments appear to be important, but were the conditions for these adjustments cannot be assumed to be as perfect as presumed in standard theory (see next section for a discussion of these assumptions).
Since these cases are likely to include processes of economic transition, privatization, the introduction (or design) of markets, as well as changes in economic policy, the CLA may also have implications on a normative level.
4 Note that the notion of “determinants” does not mean that learning is viewed as a deterministic process. Instead, the idea is that learning is influenced in a predictable way by situational factors, where the term “situational” refers to factors that are part of the situation rather then part of the psychological makeup of the individual.
5 Note that the term “behavioral” is used here to denote a theory (or a hypothesis) that is derived from evidence about actual human behavior. This evidence is based on both behavioristic and cognitive theories and empirics in psychology and economics. Hence, behavioral approaches in this sense may profoundly differ from behavioristic approaches in the tradition of Pavlov, Thorndike and Skinner.
Prof. Tilman Slembeck
Next: Learning in economics - a critical review
Summary: Index