This chapter adds learning to heterogeneity and interaction: agents are firms who do not only adaptively survive a system's changes, they also aim to improve their condition with anticipatory and reflexive capabilities. Agents shape the macroeconomic environment, adapt to its changes and also make decisions to fulfill this goal. From this perspective, the main aim of agents is to survive and to improve their capacity of withstanding changes in the environment.

This chapter provides another example of the usefulness of the ME approach as it demonstrates how the ME is able to mimic the behaviour of a dynamical system with a large number of equations, which cannot be analytically treated.

The structure of the chapter is as follows. Section 5.1 introduces the model and frames it within the literature, stressing the relevance of the reflexivity and anticipatory theory for a macro-model with learning agents. Section 5.2 presents the ABM model, detailing the behavioural rules for firms. Section 5.3 illustrates how the ME is built from the behavioural assumptions in order to make inferences based on the ABM. Section 5.4 provides the results of the simulations, contrasting the ABM and the ME solution. Finally, Section 5.5 offers some concluding remarks.

Introduction

This chapter considers agents that are not only adaptive but, to some extent, also able to think or, more generally, to learn, particularly with reference to the theory of George Soros (2013) on reflexive systems and of Robert Rosen (1985) on anticipatory systems.

The reflexivity principle applies only to systems of thinking agents and, briefly, it states that interpretations of reality have a deep influence on the behaviour of the agents participating in that reality. Such influence is so effective that changes in behaviours may modify the reality, creating uncertainty. As an example: if investors believe that markets are efficient, then that belief will change the way they invest, which in turn will change the nature of the markets in which they are participating (though not necessarily making them more efficient) (Soros, 2013, p.310).

The synthesis then comes with the human uncertainty principle: The uncertainty associated with fallibility and reflexivity is inherent in the human condition. To make this point, I lump together the two concepts as the human uncertainty principle (Soros, 2013, p.310).