Reactions to Flight Deck Alerts in a Simulator

RESEARCH ARTICLE

Expertise and Responsibility Effects on Pilots’ Reactions to Flight Deck Alerts in a Simulator Yiyuan Zheng, Yanyu Lu, Zheng Yang, and Shan Fu ZHENG Y, LU Y, YANG Z, FU S. Expertise and responsibility effects on Similarly as in other fields, visual and auditory alerts pilots’ reactions to ﬂight deck alerts in a simulator. Aviat Space are most often used on the flight deck. Visual alerts Environ Med 2014; 85:1100–5. create a stimulus to capture a pilot’s attention through Introduction: Flight deck alerts provide system malfunction informadramatic color change, motion, or intense luminance tion designed to lead to corresponding pilot reactions aimed at guaranteeing flight safety. This study examined the roles of expertise and flight contrast. Color and motion are significantly salient beresponsibility and their relationship to pilots’ reactions to flight deck cause they provide strong cues for object segmentation alerts. Methods: There were 17 pilots composing 12 flight crews that in the natural visual environment (18). Flashing and were assigned into pairs according to flight hours and responsibilities. blinking signals are the most effective forms to grab the The experiment included 9 flight scenarios and was carried out in a CRJ200 flight simulator. Pilot performance was recorded by a wide angle operator ’s attention (26). However, the limitations of the video camera, and four kinds of reactions to alerts were defi for Delivered byned Ingenta to: UCLalert LIBRARY visual are obvious. Inattention blindness, which is analysis. Results: Pilots tended to have immediate reactions to uninterIP: 46.161.56.161 On: Mon,also 08 Aug 2016 known as 01:31:52 the looked-but-failed-to-see effect (8), rerupted cautions, with a turning off rate as high as 75%. However, this Copyright: Aerospace Medical Association sulting from the attention being focused only on central rate decreased sharply when pilots encountered interrupted cautions and warnings; they also exhibited many wrong reactions to warnings. vision or visual periphery, has been considered as a Pilots with more expertise had more reactions to uninterrupted caucommon cause of traffic accidents. Change blindness tions than those with less expertise, both as pilot flying and pilot moni(20) is another form of visual alert ignorance that the obtoring. Meanwhile, the pilot monitoring, regardless of level of expertise, server fails to detect. Nikolic and Sarter suggested that exhibited more reactions than the pilot flying. In addition, more experienced pilots were more likely to have wrong reactions to warnings while pilots are more likely to overlook the green box onset acting as the monitoring pilot. Conclusions: These results suggest that during changes in automation mode on the flight mode both expertise and flight responsibility influence pilots’ reactions to annunciator (14). alerts. Considering crew pairing strategy, when a pilot flying is a less Because of its omnidirectional nature, auditory stimexperienced pilot, a more experience pilot is suggested to be the monitoring pilot. The results of this study have implications for understanding uli are the most reliable attention grabbers for alarm syspilots’ behaviors to flight deck alerts, calling for specialized training and tems, without the high possibility of visual lapse (1). It design of approach alarms on the flight deck. would not be unheard because of the position or direcKeywords: expertise, flight responsibility, alerts.

T

HE FLIGHT DECK of a commercial aircraft is a complex human-in-the-loop system. Pilots should concentrate on aviating, navigating, communicating, and system management when they are flying. On advanced flight decks, the role of the pilot is not only defined as an individual operator, but also as a commander, a team member, and a flight deck occupant (13). Pilots are required to maintain full situational awareness during the entire flight. However, it is possible that they may fail to notice a system malfunction if the corresponding alert is not conspicuous enough, especially for an emerging situation. Therefore, the primary role of the alert is to direct a pilot’s attention from a current task to a particular system event. Normally, the alarm situation has a higher priority than the current task and requires the pilot’s immediate attention. The pilot must recognize the alert, analyze its meaning in relation to other information presented on the flight deck, and aim to resolve any abnormal or unsafe operating conditions (17). Moreover, the alert should express an explicit description or diagnosis of the malfunction by either a text instruction or voice prompt, or both, to assist the pilot in comprehending the situation (5). 1100

tion the pilot was facing. According to a Lufthansa survey (23), 98% of the pilots of Boeing 737-300 and Airbus 310-200 believed that auditory alerts were of vital importance to flight safety. Nevertheless, the distraction property sometimes annoys pilots, since spoken material has obligatory access to memory. The most common complaint about auditory alerts is excessive volume (10), and the sound might disrupt communication or even hinder normal operation. In order to make alerts more salient and comfortable, the Federal Aviation Administration (FAA) has published a related advisory: “Warning information must be provided to alert crew to unsafe operating conditions, and enable them to take appropriate corrective action. Systems, From the School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai, China. This manuscript was received for review in June 2014. It was accepted for publication in August 2014. Address correspondence and reprint requests to: Shan Fu, Ph.D., Professor, School of Aeronautics and Astronautics, Shanghai Jiao Tong University, 800# Dongchuan Road, Minhang District, Shanghai, China; [email protected]. Reprint & Copyright © by the Aerospace Medical Association, Alexandria, VA. DOI: 10.3357/ASEM.4078.2014

Aviation, Space, and Environmental Medicine x Vol. 85, No. 11 x November 2014

REACTIONS TO FLIGHT DECK ALERTS—ZHENG ET AL. as captains or co-captains for other types of aircrafts (8 controls, and associated monitoring and warning means must be designed to minimize crew errors which could for B737, 5 for A320, 3 for A330, 2 for A340, and 2 for create additional hazards.” In addition, some research B747). All subjects signed the consent form for the present has been performed on the influences of noise on perstudy before the experiment that was approved by the formance efficiency. Patterson suggested that alerts Institutional Review Board of the School of Aeronautics should be announced at 15 dB for detection and limited and Astronautics, Shanghai Jiao Tong University. to beneath 30 dB to prevent task disruption (15). FurtherEquipment more, after summarizing questionnaire feedback from 50 commercial pilots, Peryer proposed that the noise The experiment was carried out in a CRJ-200 full-flight level on the flight deck be reduced to approximately 10 simulator. It is a qualified flight simulator (level C) conto 15 dB (16). To overcome limitations, alerts are deforming to the guidance presented in Federal Aviation signed to be presented in both visual and auditory Administration Advisory Circular (AC 120-40B) - Airplane forms, and the flight manual recommends that the pilots Simulator Qualification (6). All of the alerts that were silence the alert and extinguish the corresponding light generated in the flight simulator were the same as those by pushing the switch when the cause is recognized, althat would be generated during actual flight. In addition though the alert would also disappear after the malto this experiment, the flight simulator has also been function was removed. used for pilot training for commercial airlines. However, although researchers and authorities have The main six alert lights on the CRJ-200 are located on proposed many alarm system design principles, the auboth sides of a glare shield. When a system malfunction thors were unable to find any published studies based occurs, the relevant alert light keeps flashing accompaon pilots’ practical reactions during flight. Since no relnied by an auditory alarm until the flight crew turns it off evant standard operating procedures are given, whether or UCL clearsLIBRARY the malfunction. In alert cases, MASTER CAUDelivered by Ingenta to: expertise and flight responsibility flying and On: pilotMon,TION and MASTER WARNING are the brightest. The IP:(pilot 46.161.56.161 08 Aug 2016 01:31:52 monitoring) affect pilots’ reactions on turning off alerts, Medical Copyright: Aerospace Association positions of alerts on the left glare shield and the interpreand whether the reactions have side effects on normal tations are represented in Fig. 1 and Table I, respectively. operations such as manipulating the aircraft or doing The performance of each pilot was recorded by a wide the checklist are inconclusive. angle video camera (pixel: 640 x 480, sampling frequency: The aim of the current investigation was to study the 10 Hz) which was installed behind the emergency exit behaviors and the potential influences of pilots’ reacdoor on the flight deck. The recording accurately detions to alerts, and the possibility of whether the results picted the reactions of pilots over time and space during could be used in pilot training and alert system design. the experiment. In this study, 17 pilots composed 12 flight crews based on their various flight hours. A total of 245 treatments of Procedure alerts occurring during 9 scenarios in a CRJ-200 flight Nine flight scenarios were designed for the experisimulator were recorded and analyzed. ment. Each scenario lasted approximately 12 min. These scenarios consisted of the emergencies and malfunctions METHODS that might be encountered and could lead to disasters Subjects during normal flight; for example, going around under a cross wind, one engine failure after V1, one engine on Seventeen Chinese male commercial airline pilots fire, fuel imbalance, etc. When the alert appeared, the ranging in age from 30 to 53 yr (Mean 5 39.1 6 7.75 yr) subject was required to determine what system was gowere invited to participate in this experiment. Their ing wrong, and then perform the relevant checklist to mean total flight hours were 7173.2 6 5270.9 h (range operate the aircraft into a safe configuration by either from 1000 to 18,000 h), and mean flight hours in the last removing the malfunction or landing successfully. The two weeks before the experiment were 10.82 6 11.66 h subjects were first distributed into two groups, less ex(range from 0 to 89 h). Each pilot has been either captain pert (LE) and more expert (ME), according to their exor co-captain of CRJ-200 for more than 1 yr (Mean 5 4.19 6 3.62 yr). Simultaneously, they had all been recruited pertise. The classification was based on flight hours

Fig. 1. The alerts positions on the left glare shield of the CRJ-200 flight deck. Aviation, Space, and Environmental Medicine x Vol. 85, No. 11 x November 2014

1101

REACTIONS TO FLIGHT DECK ALERTS—ZHENG ET AL. TABLE I. INTERPRETATION OF SIX ALERTS ON THE LEFT GLARE SHIELD OF THE CRJ-200.

analyzed separately. Then, the results of the remaining 50 alerts were considered. Among the 195 cautions, 149 occurred on normal flight Alert Switches / lights (uninterrupted cautions), and the other 46 emerged LH ENG FIRE PUSH Engine fire warning switch / light while flight crews were doing checklists (interrupted PULL UP/GND PROX Pull up / Ground Proximity switch / light cautions). In the first case, the subjects turned the alerts STALL Stall Warning switch / light off 111 times immediately (IR), delayed 22 times (DR), MASTER CAUTION Master Caution switch / light MASTER WARNING Master Warning switch / light did the checklists 16 times directly (NR), and did not ROLL SEL/CPLT ROLL Roll selection switch / light press the wrong switches (WR). The counterpart times in the interrupted cases were 16, 7, 23, and 0, where the percentage of IR was decreased dramatically, and rather than age, and 4000 flight hours was selected as NR was increased significantly as shown in Fig. 2. the threshold to ensure near equal cohorts. Therefore, Considering all immediate and delayed reactions of 9 pilots belonged to the more expert group (mean flight the uninterrupted cautions together, both expertise hours 5 11,222.2 6 3833.3 h), and the other 8 pilots were [F(1,20) 5 24.17, P , 0.001] and flight responsibility regarded as less expert (mean flight hours 5 2618.1 6 [F(1,20) 5 19.58, P , 0.001] had significant differences 1385.0 h). Further, these pilots were paired into 12 flight on the reactions and showed no interaction [F(1,20) 5 0, crews. The combinations of flight crews included four P 5 1]. The mean different reaction percentages and pairs because of different expertise distributions (Pilot standard deviations are shown in Fig. 3. The results of Flying / Pilot Monitoring: Less Expert / Less Expert; Less immediate reaction alone were similar to above. More Expert / More Expert; More Expert / Less Expert; and expert pilots responded more frequently than less exMore Expert / More Expert), and each pair contained pert pilots [F(1,20) 5 73.09, P , 0.001], and pilots monithree crews. Among the 17 subjects, 7Delivered pilots, including by Ingenta to: UCLwere LIBRARY toring more sensitive than pilots flying [F(1,20) 5 3 More Expert and 4 Less Expert, assigned with IP: were 46.161.56.161 On: Mon, 08 Aug 2016 01:31:52 38.03, P , 0.001]. The interaction was insignificant different flight responsibilities in different crews involved, Copyright: Aerospace Medical Association [F(1,20) 5 0.96, P 5 0.339]. Besides, more expert pilots e.g., as Pilot Flying in one crew and as Pilot Monitoring were more likely to help their partners to turn off the in the other. Before the experiment was carried out, cautions when they performed as the pilot flying. The each crew was given 6 h to become familiar with the difference of immediate reaction to cautions between scenarios and the operations in the flight simulator. For more expert as pilot flying and less expert as pilot monithe 7 pilots who performed the experiment twice, the toring was insignificant [t (11) 5 1.65, P 5 0.129]. This second training period was 2 h. In the experiment, one led to an overlapping phenomenon that both pilot flying flight instructor stayed with the flight crew in the simuand pilot monitoring responded to the cautions simultalator, was responsible for the configurations of each neously. The mean rate of overlapping operations when scenario, and acted in the role of ATC sometimes. The more expert performed as pilot flying was as high as order of scenarios was blind to the flight crew. The 0.19 6 0.07. experiment for one flight crew was finished in 2 days, Under delayed reaction conditions, the less expert pifour scenarios in the first morning and five in the next lot was more inclined to turn off the cautions over 2 s afternoon, or vice versa. Further, the pilot who was parthan more expert as the pilot monitoring [t (11) 5 5.11, ticipant in the first crew experiment did not take part in P , 0.001]. Nevertheless, the difference was insignificant another with a different partner until 14 d later. The exwhen they were pilot flying [t (11) 5 0.65, P 5 0.527]. periment was carried out over one month. Meanwhile, as long as pilot monitoring was performed All the operations of flight crews were recorded by by a more expert pilot, none of his partners had delayed the video camera, and four kinds of pilots’ reactions on reactions. alerts were defined as following: • Immediate Reaction (IR): turning off the alert in 2 s by pressing the switch; • Delayed Reaction (DR): turning off the alert over 2 s by pressing the switch; • No Reaction (NR): handling fault directly without turning off the alert; • Wrong Reaction (WR): turning off the wrong alert.

The threshold of IR and DR was defined as 2 s, because according to the previous studies, the mean reaction time of pilots to incentives was ;2 s (2,19). The results of the experiment were analyzed based on different flight responsibilities (pilot flying vs. pilot monitoring) and expertise (less expert vs. more expert). RESULTS Since cautions occupied the majority of the alerts (195/245), subjects’ reactions to the cautions were first 1102

Fig. 2. The percentage of three different reactions to the uninterrupted cautions and interrupted cautions. The three different reactions are Immediate Reaction (IR), Delay Reaction (DR), and No Reaction (NR).


REACTIONS TO FLIGHT DECK ALERTS—ZHENG ET AL. remaining 50 alerts of the experiment belonged to warnings. Therefore, it was reasonable that the rate of no reaction increased significantly. Pilots reacted more urgently to warnings than cautions, as the performance might be impaired with time pressure (22). Considering the two different roles of the flight crew, the pilot flying is concerned primarily with control of the aircraft, and the pilot monitoring should focus on whether the systems work properly. In consequence, determining the malfunctions, turning off the alerts, and solving the problem are all within the scope of pilot monitoring responsibilities. Thus, both more and less Fig. 3. The reaction percentages of Less Expert and More Expert pilots operating as Pilot Flying (PF) and Pilot Monitoring (PM) to the caution expert pilots when operating as the monitoring pilot in alerts. The error bars stand for the difference of reaction percentages of the experiment were more concerned with the alerts the pilots with the same expertise and in the same flight responsibility. than their partners. Furthermore, from the perspective of expertise, the results indicated that more expert pilots Among the remaining 50 alerts, the distribution of the were more sensitive than less expert pilots to cautions as 4 types of reactions was quite different compared with the monitoring pilot. This could be interpreted from two the cautions. The alerts were turned off immediately aspects. Firstly, the reactions of more expert pilots to only 8 times, and delayed twice. However, there was cautions were probably a kind of subconscious operano reaction 19 times, and the wrong reaction occurred tion after many flight hours. With practice and experi21 times. The rates of no reaction and wrong reaction ence, the resource demanded to accomplish a task could by Ingenta to: UCL LIBRARY were dramatically higher, especiallyDelivered wrong reaction. decrease until it became automatic (25). At this point, IP: 46.161.56.161 On: Mon, 08 Aug 2016 01:31:52 The detailed reaction results of the Copyright: different alerts are the reactions of more expert pilots were consistent, fast Aerospace Medical Association shown in Table II. Among all wrong reactions, only and error free, and automatically triggered. Secondly, 2 times were pilots pressing PULL UP when ENG FIRE more expert pilots were more self-confident than those occurred, and the rest were all pressing MASTER with less expertise. Having encountered the same situaCAUTION instead. tions many times, they believed they could handle the The difference of wrong reaction between more and emergencies with less or even without instructions. Simless expert pilots as monitors was insignificant to these ilarly, Kennedy and Taylor found that older pilots were alerts [t (11) 5 1.72, P 5 0.115], and the difference was more likely to land when visibility was inadequate (11). also insignificant when they were flying [t (11) 5 1.49, The conditioned responses and self-confidence of the P 5 0.166]. However, considering the wrong reaction to more expert pilots was even expressed when they operMASTER WARNING, the more expert pilots were more ated as the pilot flying and resulted in the overlapping likely to make mistakes when they were monitoring phenomenon. [t (11) 5 2.28, P 5 0.048]. This kind of inclination disapIn addition, different flight crew pairing strategies inpeared from the perspective of the pilot flying [t (11) 5 fluence the reactions to cautions. The least reactions 1.83, P 5 0.083]. In addition, the difference of no reaction happened when two less experienced pilots were was insignificant between less and more expert as monigrouped together (31/48), and the most reactions betor [t (11) 5 2.22, P 5 0.051]. longed to the two most experienced as a crew (41/47). Among all the reactions of the two most expert crews, a DISCUSSION delayed reaction only occurred one time. These results also demonstrated that more experienced pilots were Three categories of alerts (warning, caution, and advimore sensitive to cautions. What was surprising was sory) were normally applied to each operational situathat as long as the monitoring pilot was more experition or aircraft system malfunction (24). Among them, enced, there were no delayed reactions for his partner. warning alerts require immediate flight crew action, However, the reasons might not be the same. For the and caution alerts require immediate awareness (7). The subject with less expertise as the partner, the phenomenon might stem from the dependence and respect, and for the more expert partner. TABLE II. REACTION TIMES FOR THE 4 KINDS OF REACTIONS TO THE REMAINING 50 ALERTS. Some cautions occurred as interrupted alerts while the flight crews were doing checklists. Although more PULL UP MASTER WARNING ENG FIRE STALL expert pilots also had better performance than their less experienced counterparts, the reactions to the interIR 1 6 1 0 rupted cautions reduced sharply (20 times for more exDR 2 0 0 0 pert and 5 times for less expert of all 46 interrupted NR 4 9 5 1 WR 5 13 3 0 cautions) compared with the uninterrupted ones. Since Total 12 28 9 1 alerts tended to occur in phases of flight with a relatively high workload (27), the appearance of the interrupted IR5 Immediate Reaction; DR 5 Delayed Reaction; NR 5 No Reaction; caution could be considered as a secondary task, and the WR 5 Wrong Reaction. Aviation, Space, and Environmental Medicine x Vol. 85, No. 11 x November 2014

1103

REACTIONS TO FLIGHT DECK ALERTS—ZHENG ET AL. high workload of the primary task was detrimental to In future research, we will concentrate on developing the performance of the secondary task (12). The pilots different training strategies for less and more expert piwere so busy with troubleshooting that they could not lots, based on the results of different reactions to flight shift their attention to the new alert immediately. Condeck alerts. Meanwhile, better positioning and more temporarily, the influences between the primary and easily distinguishable sounds for warnings and cautions secondary tasks were mutual. It would result in the phethat may effectively reduce the wrong reactions of pilots nomenon of “preemption,” which was described as a will also be studied. discrete auditory message is more likely to attract attenACKNOWLEDGMENTS tion away from the ongoing visual tasks (4). In other This research work was supported by National Basic Research words, the interrupted alerts might compete with the Program of China (973 Program No. 2010CB734103) and National uninterrupted ones on occupying the resource channels Natural Science Foundation of China (61305141). Authors and afﬁliations: Yiyuan Zheng, Ph.D., Candidate, Yanyu (21). Although it seemed there were no fatal effects of Lu, Ph.D., Shan Fu, Professor, Ph.D., School of Aeronautics and the interrupted cautions in the experiment, some disasAstronautics, Shanghai Jiao Tong University, Shanghai, China. ters still happened from the suspension of doing checklists during the flight (3). More expert pilots showed REFERENCES more sensitivity to cautions. Meanwhile, they also made 1. Banbury SP, Macken WJ, Tremblay S, Jones DM. Auditory distraction and short-term memory: phenomena and practical more wrong reactions to the other alerts, especially to implications. Hum Factors 2001; 43:12–29. MASTER WARNING. These results could be attributed 2. Causse M, Baracat B, Pastor J, Dehais F. Reward and uncertainty to three different aspects. First, the switches of MASTER favor risky decision-making in pilots: evidence from cardioCAUTION and MASTER WARNING were positioned vascular and oculometric measurements. Appl Psychophysiol Biofeedback 2011; 36:231–42. so close to each other that it was easy to push the wrong 3. Degani A, Wiener EL. Cockpit checklists: concepts, design, and one by mistake. Second, with the emergence of the useLIBRARY . Hum Factors 1993; 35:345–59. Delivered Ingenta to: UCL warning, the increasing time and mental pressureby might K. The challenge of managing interruptions, distractions, IP: 46.161.56.161 On: Mon, 4.08Dismukes Aug 2016 01:31:52 have led to the human error (9). Third, the conditioned and deferred tasks. In: Jensen R, ed. Proceedings of the 11th Copyright: Aerospace Medical Association International Symposium on Aviation Psychology; March 5-8, reactions of more experienced pilots combined with 2001. Columbus, OH: Ohio State University; 2001. indistinguishable sounds between cautions and warn5. Endsley MR. Toward a theory of situation awareness in dynamic ings meant that they tended to press the MASTER systems. Hum Factors. 1995; 37:32–64. 6. Federal Aviation Administration. Airplane simulator qualification. CAUTION switch spontaneously when hearing whichFAA AC 120-40B. Washington, DC: FAA; 1991. ever alarm sound. Although in some other aircraft types, 7. Glover H. Terrain displays for enhanced ground proximity like the Boeing 777, warnings such as engine fire, stall, warning systems. Troy, MI: SAE international; 1997; SAE and ground proximity are all integrated into a warning Technical Paper 975557, doi:10.4271/975557. 8. Herslund M-B, Jørgensen NO. Looked-but-failed-to-see-errors in system, and only MASTER CAUTION and MASTER traffic. Accid Anal Prev 2003; 35:885–91. WARNING switches are left on the glare shield, it is still 9. Hollands JG, Wickens CD. Engineering psychology and human possible that pilots have the wrong reaction to the performance. Upper Saddle River, NJ: Prentice Hall; 1999. 10. James S. Audio warnings for military aircraft. Paper presented at warning, especially for more expert pilots. The effects the AGARD Aerospace Medical Panel Symposium on Audio of wrong reaction seem insignificant to flight. However, Effectiveness in Aviation. Copenhagen, Denmark. 1997. time and mental pressures are the critical factors for 11. Kennedy Q, Taylor JL, Reade G, Yesavage JA. Age and expertise warning conditions that require immediate and comeffects in aviation decision making and flight control in a flight simulator. Aviat Space Environ Med 2010; 81:489-97. plete attentions. Wrong reactions not only take crucial 12. Meshkati N, Hancock P. Human mental workload. Amsterdam: time but also divert pilots’ attentions. Therefore reducNorth-Holland/Elsevier; 1988. ing wasted time and providing a comfortable environ13. Palmer MT, Rogers WH, Press HP, Latorella KA, Abbott TS. A ment and better operating procedures for flight crew to crew-centered flight deck design philosophy for high-speed civil transport (HSCT) aircraft. Hampton, VA: NASA Langley cope with alerts are essential to flight safety. Research Center, 1995; NASA Technical Memorandum 109171. In this study, we investigated the effects of exper14. Nikolic MI, Sarter NB. Peripheral visual feedback: A powerful tise and flight responsibility on reactions of pilots to means of supporting effective attention allocation in eventdriven, data-rich environments.. Hum Factors. 2001; 43:30–8. alerts. More experienced pilots were more likely to re15. Patterson RD. Guidelines for auditory warning systems on civil move auditory disturbances of uninterrupted cauaircraft. London: Civil Aviation Authority; 1982. tions before they handled the malfunctions, even when 16. Peryer G, Noyes J, Pleydell-Pearce K, Lieven N. Auditory alert they were operating as the pilot flying. Although less characteristics: a survey of pilot views. Int J Aviat Psychol 2005; 15:233–50. expert pilots were also annoyed by the sounds, they 17. Pritchett AR. Reviewing the role of cockpit alerting systems. Hum showed better tolerance. Moreover, the reactions to Factors Aerosp Saf 2001; 1:5–38. interrupted cautions decreased sharply for both levels 18. Regan D, Gray R. Human perception of objects. Toronto: Sinauer Assoc.; 2000. of expertise. Expertise was not associated with better 19. Schriver AT, Morrow DG, Wickens CD, Talleur DA. Expertise performance of the reactions to warnings. More expert differences in attentional strategies related to pilot decision pilots made more mistakes by pushing the wrong switches making. Hum Factors. 2008; 50:864–78. when they tried to turn off the warnings. Besides, flight 20. Simons DJ, Levin DT. Change blindness. Trends Cogn Sci 1997; 1:261–7. crew pairing strategies also had influences on reac21. Stone G, Gulick R, Gabriel R. Use of task timeline analysis to assess tions to alerts. When the pilot flying is less experienced, crew workload. In: Roscoe AH, ed. The practical assessment of it is better to assign a more experienced pilot as the pilot workload. Amsterdam, The Netherlands: North Holland monitor. Press; 1987; NATO AGARD-AG-282. 1104


REACTIONS TO FLIGHT DECK ALERTS—ZHENG ET AL. 22. Svenson O, Maule AJ. Time pressure and stress in human judgment and decision making: New York: Springer; 1993. 23. Uchtdorf D, Heldt P. Survey on cockpit systems B737-300/A310200. Flight Crew Information. Frankfurt, Germany: Lufthansa, Flight Operations Division; 1989. 24. Veitengruber J. Design criteria for aircraft warning, caution, and advisory alerting systems. J Aircr 1978; 15:574–81. 25. Wickens CD, McCarley JS. Applied attention theory. London: Taylor & Francis; 2007.

26. Wickens CD, Rose PN. Human factors handbook for displays: Summary of findings from the Army Research Lab's Advanced Displays & Interactive Displays Federated Laboratory. Aberdeen Proving Ground, MD: Rockwell Scientific Company for the ARL Advanced Displays and Interactive Displays Federated Laboratory Consortium; 2001. 27. Woods DD. The alarm problem and directed attention in dynamic fault management. Ergonomics 1995; 38:2371–93.

Delivered by Ingenta to: UCL LIBRARY IP: 46.161.56.161 On: Mon, 08 Aug 2016 01:31:52 Copyright: Aerospace Medical Association


1105