The more the better? The impact of number of stages of likelihood alarm systems on human performance

Responses to alarms involve decisions under uncertainty. Operators do not know if an alarm is more likely to be a hit or a false alarm. Likelihood alarm systems (LAS) help reduce this uncertainty by providing information about the certainty of their output. Unlike traditional binary alarm systems, they have three or more stages: each one represents a different degree of likelihood that a critical event is really present. Consequently, the more stages, the more specific is the information provided by the alarm system to reduce uncertainty. A laboratory experiment with 48 participants was conducted to investigate the effect of specificity of information of LAS on performances and responding behaviour. Specifically, a three-stage, four-stage, and five-stage LAS were compared using a multi-task environment. Results show higher percentages of correct decisions in the alarm task when participants used the four- and five-stage LAS than the three-stage LAS but no significant differences were found between the four-and five-stage LAS. Interesting differences in response patterns were also observed. This study suggests that four stages is the best degree of specificity for optimal performance