False Alarms and Human-Machine Warning Systems

This work illustrates that false alarms are likely to have significant frequencies as well as detrimental influence on the effectiveness of human-machine warning systems. Several factors are responsible for false alarm materialization, including the need to predict uncertain conditions in the future, variability of human perception, and low a priori probabilities of traffic collisions. The effect of false alarms on human trust in warning systems and on credibility of warnings could be considerable even for low false alarm rates. One way to decrease false alarm rates would be to focus on predicting possible conflicts, which are much more probable than actual collisions (thus increasing a priori probabilities). One of the suggested future research directions is to assess directly what different individuals will view as being a false alarm. In the majority of existing studies, the researchers defined this outcome. It is possible that people will change their behavior based on their perception of false alarms rather than on a researcher's definition of them.