Toward a cognitive experimental economics

For the last few years, economic literature has shown a new stream of analysis, at the crossroads of the heterodox tradition and cognitive sciences. A new approach is born, called Cognitive Economics, since attention is focused on the study of individual and organizational learning, seen as a key factor in shaping social phenomena (Rizzello, 2000). This new stream of analysis aims at refounding economic methodologies, starting from more realistic hypotheses than those at the basis of the mainstream. Therefore, empirical observation of real men and women's learning and decision making is necessary. Experimental economics, as the main way to gather data on these aspects, is essential for the development of such an approach. At the moment, though, there is still a lack of methodological discussion on how to perform experiments. This chapter aims at analyzing the relation between experimentalism and cognitive approach in economics. As it will be shown, cognitivism implies and requires methodological and procedural peculiarities as compared with the mainstream in the realization of experiments and in the analysis of the relative data. These differences depend, first of all, on the fact that, though it is deep-rooted in social sciences, this can be considered a new-born discipline with obvious different necessities from those of an older science. At the moment, in fact, empirical analysis should here be directed mainly towards the gathering of empirical evidence, instead of comparing alternative models to find the most fitted one. The importance of this descriptive interest is also related to the goal of understanding reality, and not just building models with predictive capacity. One of the main questions suggested by the cognitive literature, a question which raises many theoretical and empirical problems, is the heterogeneity of human agents. Ceteris paribus hypotheses cannot be simply posed and accepted, but should be tested and verified. Differences among subjects should, in fact, be studied and understood. But this goal requires new instruments and new kinds of analysis. For example, if individuals differ, a similar experimental treatment can bring to different behaviours. A macro analysis could then lead to conclude that a variable is not relevant, because positive and negative effects can compensate each other at a mean level. But the "mean agent" could make no sense. Some new devices and tools are available to carry on the study. New kinds of experiments, psychological tests and interviews have been recently developed and tested. Also the interaction between experimentalism and computational economics and its simulations can be very useful. Simulations with artificial agent is a growing and promising area of research; but the heterogeneity of human agents makes it difficult, at the moment, to perform such models, without looking at the real world. The risk would be, again, that of a science disconnected from reality.