Int. J. Human – Computer Studies (1997) 46, 17 – 30 Automation-induced monitoring inefficiency: role of display location* INDRAMANI L. SINGH, ROBERT MOLLOY AND RAJA PARASURAMAN† Cogniti  e Science Laboratory , The Catholic Uni  ersity of America , Washington , DC 20064 , USA (Recei ed 23 June 1995 and accepted in re ised form 25 July 1996) Operators can be poor monitors of automation if they are engaged concurrently in other tasks. However, in previous studies of this phenomenon the automated task was always presented in the periphery, away from the primary manual tasks that were centrally displayed. In this study we examined whether centrally locating an automated task would boost monitoring performance during a flight-simulation task consisting of system monitoring, tracking and fuel resource management sub-tasks. Twelve nonpilot subjects were required to perform the tracking and fuel manage- ment tasks manually while watching the automated system monitoring task for occasional failures. The automation reliability was constant at 87.5% for six subjects and variable (alternating between 87.5% and 56.25%) for the other six subjects. Each subject completed four 30 min sessions over a period of 2 days. In each automation reliability condition the automation routine was disabled for the last 20 min of the fourth session in order to simulate catastrophic automation failure (0% reliability). Monitoring for automation failure was inef ficient when automation reliability was constant but not when it varied over time, replicating previous results. Furthermore, there was no evidence of resource or speed accuracy trade-of f between tasks. Thus, automation-induced failures of monitoring cannot be prevented by centrally locating the automated task. ÷ 1997 Academic Press Limited 1. Introduction Automation has become more advanced in recent years. Automation has been implemented to reduce operators’ workload and fatigue, improve safety, and allow faster and more precise control of multiple simultaneous tasks. However, several potential human performance problems have also emerged relating to the user’s interaction with automation technology. These problems include (1) a reduction in the operator’s system awareness, (2) an increase in monitoring workload, and (3) a degradation in manual skills. A reduction in manual control skills appears particularly critical, as degradation in this skill limits the ability of the operator to quickly resume accurate manual control of a process following automation failure (James, McClumpha, Green, Wilson & Belyavin, 1991; Parasuraman, Molloy & Singh, 1993). Another potential cost of automation that has been noted is the problem of ‘‘complacency’’ (Wiener, 1981; Thackray & Touchstone, 1989; Singh, Molloy & Parasuraman, 1993). Wiener (1981) searched NASA’s aviation safety reporting * This paper was presented at the 33rd Annual Conference of The Psychonomic Society, held in St. Louis, MO, USA from November 13 to 15, 1992, by the first author. † Author for correspondence. 17 1071-5819 / 97 / 010017 + 14$25.00 / 0 / hc960081 ÷ 1997 Academic Press Limited