Approximation complexity and human decision-making
How does approximation complexity affect the quality of decision-making?
It is often suggested that decision-makers use heuristics to overcome complexity, and that the resulting behaviour closely approximates optimal behaviour. However, little is known about the efficacy of heuristics. We study decision-making in a setting in which participants are faced with decisions that differ in their degree of approximation complexity. We find that decision quality decreases as approximation complexity increases, suggesting that decision quality is a consequence of the inherent 'hardness' of the decision task. Claims about decision-makers' ability to approximate optimal solutions of normative models such as expected utility maximisation or Bayesian inference therefore need to be evaluated carefully. Our findings are can also be used for building refined models of bounded rationality.