Ergodicity economics in the Copenhagen experiments

Dr Ole Peters (London Mathematical Laboratory, Santa Fe Institute) explores how ergodicity economics rethinks decision-making and expected value, drawing on insights from Copenhagen experiments.

Abstract

Expected values play an important role in economic modelling. Many behavioural models propose that humans optimise expected wealth, income or utility. However, for non-ergodic observables, the expected value generally differs from the long-time average.

It is therefore often uninformative about what may happen in the future, and there is no immediate reason why decision-makers should optimise for it. This aligns with observations that the relevant models – expected utility theory and its counterparts – have shown limited predictive power beyond sample data.

Ergodicity economics (EE) is designed for situations where time averages differ from expected value. In recent years, EE models have been tested in a series of behavioural and neuroscientific experiments at the Danish Research Centre for Magnetic Resonance. Dr Peters introduces the general problem, explains the experimental setups, and reports recent findings and possible interpretations.

This event is part of the Financial Computing and Analytics Research Group seminar series at UCL Computer Science.