‘Homo economicus’ as an intuitive statistician (3): Data-driven causal inference

Abstract

The bounded rationality programme, as understood in new classical economics, views the economy as a society of intuitive statisticians – the intuitive statistician hypothesis. This hypothesis raises the question of whether there is a ‘tight enough’ theory of statistical inference.Without a tight enough theory of statistical inference we will not learn much about the economy by studying the dynamics of an economy of intuitive statisticians. As a general framework for studying this question, we conjectured that the agent (statistician) first seeks to learn the probability distribution of the variables representing his or her choice situation and next uses the probabilistic information to learn about the causal structure of the situation. The last two chapters studied some of the issues relating to learning the probability distribution of a set of variables. This chapter studies in detail the second general stage of inference that is concerned with inferring the causal structure of a set of variables from their joint distribution.