An Investigation of Extreme Alarm Response Patterns in Laboratory Experiments

Abstract

Response patterns to unreliable alarm systems often conform to probability matching theory. However, some participants have been observed to respond to all or no alarms in a given set. In this paper, we analyze extreme response patterns within seven alarm mistrust experiments. In half of those experiments, participants had no redundant cues regarding individual alarm validity, and relied only on an estimate of overall alarm system reliability. In the other half, participants knew overall system reliability and could check individual alarm validity. Results show that participants without alarm validity information were more likely to respond to all alarms. Those who could check individual alarm validity matched the true alarm rate with their responses. Our results confirm the importance of designing information redundancy within alarm systems. Practical implications include providing underlying system information to operators to ensure appropriate response patterns. Further research should be conducted to investigate lack of responding for low alarm system reliability rates.