Implications of the elaboration likelihood model for automation monitoring failure

As automation saturates the cockpit, pilot roles increasingly center on monitoring automated systems instead of controlling the aircraft directly (Edwards, 1976). Though cockpit automation Is highly reliable, nonvigilant monitoring may lead to several types of automation monitoring failure. The present study examines omission errors, in which pilots miss system failures undetected by monitoring automation, and commission errors, in which pilots inappropriately respond to false alarms generated by monitoring automation. Existing research has located few stimulus and organismic variables associated with successful, sustained automation monitoring. Further, little organized theory exists in the automation monitoring literature. The present study examines automation monitoring failures from the perspective of Petty and Cacioppo\u27s (1986) Elaboration Likelihood Model of persuasion (ELM). Three sets of hypotheses were proposed. First, trust in automation was hypothesized to correlate with omission and commission errors. Omission errors were hypothesized to be influenced by the amount of distraction present and the operator\u27s need for cognition (NFC), a variable associated with enjoyment of complex cognitive tasks. Commission errors were hypothesized to be jointly influenced by distraction, the operator\u27s NFC, and the quality of the arguments presented by the automated systems. The present study Included 120 participants who performed a low-fidelity flight simulation in which they had eight opportunities to commit omission errors and eight opportunities to perform commission errors. Results indicate that trust in automation was associated exclusively with commission error rates. Omission error rates rose with increased distraction or reduced NFC. Engine reset latency, the speed with which participants react to unannounced engine failures (omission error opportunities), increased with added distraction or decreased NFC-Commission error rates responded only to argument quality. Participants given high-quality arguments performed more commission errors than those given low-quality arguments. Two main conclusions were drawn. First, omission and commission errors may derive from different cognitive processes. Second, the present research did not contradict the ELM, which appears to offer a useful framework. More sophisticated tests of the ELM In the automation monitoring context, a better definition of attitudes toward automation, and a way to measure elaboration in the cockpit are required