2003/20 LEM Working Paper Series

Learning in Evolutionary Environments

Giovanni Dosi, Luigi Marengo and Giorgio Fagiolo
Learning, Evolutionary Environments, Economic Theory, Rationality

  JEL Classifications
D80, C70, A10

The purpose of this work is to present a sort of short selective guide to an enormous and diverse literature on learning processes in economics. We argue that learning is an ubiquitous characteristic of most economic and social systems but it acquires even greater importance in explicitly evolutionary environments where: a) heterogeneous agents systematically display various forms of “bounded rationality”; b) there is a persistent appearance of novelties, both as exogenous shocks and as the result of technological, behavioural and organisational innovations by the agents themselves; c) markets (and other interaction arrangements) perform as selection mechanisms; d) aggregate regularities are primarily emergent properties stemming from out-of-equilibrium interactions. We present, by means of examples, the most important classes of learning models, trying to show their links and differences, and setting them against a sort of ideal framework of “what one would like to understand about learning...”. We put a significant emphasis on learning models in their bare-bone formal structure, but we also refer to the (generally richer) non-formal theorising about the same objects. This allows us to provide an easier mapping of a wide and largely unexplored research agenda.

The paper is forthcoming in Dopfer, K. (Ed.), Evolutionary Principles of Economics, Cambridge, Cambridge University Press. It has been originally published as: IIASA Working Paper, WP-96-124, IIASA, Laxenburg (Austria). The present version has been updated in 2001.

download pdf