Age-related differences in the low-prevalence effect of target and the contribution of interventions.

Abstract

Human-machine interaction (HMI) systems, such as automatic driving, often require people to detect errors in the machines. However, the better the system becomes, the less frequently system errors occur and the more easily they are missed. This is known as the low-prevalence effect (Wolfe et al., 2005). For social implementation of HMI, what kind of people easily miss and how much improvement can be expected through intervention should be clarified. This study examined the performance of 600 participants in 20s to 70s in an online visual search task with targets appearing rarely (2%) and the effect of a short-term intervention that increased target frequency (50%). The intervention temporarily reduced misses, but they still occurred more than 50% of the time. Moreover, younger ages had a faster increase in misses after the intervention. These characteristics should be considered in implementing a system with a lower frequency of events to be detected.