Too little or too much: The effect of feedback on risk assessment in a missile defense task

Abstract

Operators must often assess risk and take actions in proportion to that risk. This is not an inherently easy task and based on previous experiments, visualization aids may not be the entire solution to aiding operators. This experiment explored the impact of feedback paradigms on performance in a national missile defense (NMD) task. Participants had extensive practice with one of three types of feedback styles: averaged results after a set of scenarios, constant feedback after every scenario, or constant feedback with forced reflection. This study has three main contributions to decision-making research. 1) Although the trust in automation literature shows that averaged feedback increases understanding of the relative frequency of errors, this NMD task showed that constant feedback is more effective than averaged feedback. 2) Participant performance was still improving after three sessions, suggesting the need for more intense training. 3) NMD provides a concrete domain with viable metrics for investigating the perception of risk and uncertainty.