Automation and Situation Awareness - Endsley
Automation and Situation Awareness - Endsley
Automation and Situation Awareness - Endsley
In each of these cases, the human operators overseeing the automated systems were unaware of critical features
of the systems they were operating. They were unaware of the state of the automated system and were unaware of the
aircraft parameters the automation was responsible for.
In addition to difficulties in detecting these types of automation errors, it is also frequently difficult for
human operators to correctly understand what the problem is, once they have detected that something is amiss. In
the US Air accident, for instance, both crew members tried to gain control of the aircraft, but were unable to due to
the mis-trimmed rudder. The lost time associated with trying to overcome the aircraft control problem without
understanding what was causing it fatally delayed aborting the take-off until it was too late.
With many automated systems, partially due to their complexity, understanding the meaning of displayed
information, once attended to, can represent a significant difficulty. For instance, in aircraft systems, pilots have
reported significant difficulties in understanding what their automated flight management systems are doing and why
(Sarter & Woods, 1992; Wiener, 1989). Similarly, the accident at the Three Mile Island nuclear power plant was
attributed to an erroneous over-ride of the automated emergency handling system by the human operators. They had
misdiagnosed the situation based on displayed information and believed an excessive coolant level was causing the
problem rather than too little coolant (Wickens, 1992). They did not correctly understand the meaning of the
information that was displayed to them.
Each of these problems can be directly linked to a lower level of situation awareness that exists when
people operate as monitors of automated systems. Situation awareness (SA), a person's mental model of the world
around them, is central to effective decision making and control in dynamic systems. This construct can be severely
impacted by the implementation of automation.
SITUATION AWARENESS
Originally a term used in the aircraft pilot community, situation awareness has developed as a major
concern in many other domains where people operate complex, dynamic systems, including the nuclear power
industry, automobiles, air traffic control, medical systems, teleoperations, maintenance, and advanced manufacturing
systems. Achieving situation awareness is one of the most challenging aspects of these operators jobs and is
central to good decision making and performance. Hartel, Smith and Prince (1991) found poor situation awareness to
be the leading causal factor in military aviation mishaps. In a recent review of commercial aviation accidents, 88
percent of those with human error involved a problem with situation awareness (Endsley, 1994a). Situation
awareness clearly is critical to performance in these environments.
Situation awareness is formally defined as the perception of the elements in the environment within a
volume of time and space, the comprehension of their meaning and the projection of their status in the near future
(Endsley, 1988a). Situation awareness involves perceiving critical factors in the environment (Level 1 SA),
understanding what those factors mean, particularly when integrated together in relation to the person's goals (Level
2 SA), and at the highest level, an understanding of what will happen with the system in the near future (Level 3
SA). These higher levels of situation awareness are critical for allowing decision makers to function in a timely and
effective manner.
For instance, in an aircraft environment, operators must be aware of critical flight parameters, the state of
their on-board systems, their own location and the location of important reference points and terrain, and the location
of other aircraft along with relevant flight parameters and characteristics. This information forms the elements
they need to perceive to have good Level 1 SA. But a great deal has to do with how the operators interpret the
information they take in. They need to comprehend that a certain pattern of flight parameters indicates that they are
near stall point, or that the displayed altitude is below their assigned altitude. This understanding forms their Level 2
SA. At the highest level, Level 3 SA, their understanding of the state of the system and its dynamics can allow
them to be able to predict its state in the near future. A group of enemy aircraft flying in a particular formation will
thus be projected to attack in a given manner. With accurate and complete situation awareness, operators can act to
bring their systems into conformance with their goals.
may not know if their requests have been received, if the actions are being performed properly, or if problems are
occurring (Norman, 1989, p. 6). He attributes this problem largely to an erroneous belief by system designers that
information on certain parameters is no longer needed by system operators once relevant functions are assumed by
automation.
In some cases, critical cues may be eliminated with automation and replaced by other cues that do not
provide for the same level of performance. In many systems, important cues may be received through auditory,
tactile or the olfactory senses. When processes are automated, new forms of feedback are created, frequently
incorporating more accurate visual displays; yet the fact that information is in a different format may make it harder
to assimilate with other information or less salient in a complex environment. Young (1969) and Kessel and
Wickens (1982) found that proprioceptive feedback received during manual control was important to performance and
denied in automated tracking tasks in which information was only presented visually. The development of electronic
fly-by-wire flight controls in the F-16 led to problems in determining airspeed and maintaining proper flight control,
as the vibration information that usually came through the flight stick was suddenly missing (even though the
needed information was clearly indicated on traditional visual displays) (Kuipers, Kappers, van Holten, van Bergen,
& Oosterveld, 1989). Artificial stick-shakers are now routinely added to fly-by-wire systems to put back in the
feedback that operators are accustomed to (Kantowitz & Sorkin, 1983).
In some cases, the design of the automated system intentionally conceals information from the operator.
Some autofeathering systems, for instance, have neglected to notify pilots of their actions in shutting down engines,
leading to accidents (Billings, 1991). In some notable accidents, the fact that the automated system had failed was
not clearly indicated to the operator, as in the Northwest Airlines accident in Detroit (National Transportation Safety
Board, 1988). In addition, there is a tendency for some displays to eliminate raw system data, in favor of processed,
integrated information. Important information as to the source of information, its reliability, or the value of
constituent data underlying the integrated information may be unavailable. Billings (1991) notes that the clarity of
the integrated displays may be seductive, yet highly misleading if such underlying information is not known.
A noted problem in many systems is the lack of information salience that may accompany automation.
Frequently displays associated with complex automated systems involve computerized CRT screens with information
imbedded in hierarchical displays that may be associated with various system modes. Problems with getting lost in
menus, finding the desired display screen, and interpreting cluttered displays have been noted. The increased display
complexity and computerized display format reduces the perceptual salience of information, even if it is available. In
a complex environment with many activities going on, it is easy for operators to loose track of such information.
Either intentionally or inadvertently, the design of many systems posses a considerable challenge to
situation awareness through the elimination of or change in the type of feedback provided to operators regarding the
systems status. Unless very careful attention is paid to the format and content of information displays, these issues
can easily sabotage situation awareness when operators are working with automated systems.
Lack of Understanding of Automation
One of the major impediments to the successful implementation of automation is the difficulty many operators
have in understanding automated systems, even when they are attending to them and the automation is working as
designed. This may be partially attributed to the inherent complexity associated with many of these systems, to poor
interface design and to inadequate training.
The development and maintenance of situation awareness involves keeping up with a large quantity of
rapidly changing system parameters, and then integrating them with others parameters, active goals and ones mental
model of the system to understand what is happening and project what the system is going to do. This allows
operators to behave proactively to optimize system performance and take actions to forestall possible future
problems. As complexity increases (as it is apt to do with automation), this task becomes even more challenging.
The number of parameters increases, and they change and interact according to complex underlying functions.
Achieving an accurate mental model of the system can be very difficult, and this taxes the ability of the operator to
attain the higher levels of situation awareness (comprehension and projection) from information that is perceived.
By adding to system complexity, therefore, automated systems may make achieving good situation awareness more
difficult.
Wiener (1989) has documented many problems with a lack of understanding of automated systems in aircraft
by the pilots who fly with them. McClumpha and James (1994) conducted an extensive study of nearly 1000 pilots
from across varying nationalities and aircraft types. They found that the primary factor explaining variance in pilots
attitudes towards advanced technology aircraft was their self-reported understanding of the system. While
understanding tended to increase with number of hours in the aircraft, this also was related to a tendency to report that
the quality and quantity of information provided was less appropriate and more excessive. Although pilots are
eventually developing a better understanding of automated aircraft with experience, the systems do not appear to be
well designed to meet their information needs. Rudisill (1994) reported from the same study that whats it doing
now, I wonder why its doing that, and well Ive never seen that before are widely heard comments in advanced
cockpits, echoing similar concerns by Wiener (1989).
Many of these problems clearly can be attributed to standard human factors short-comings in interface
design. For instance, transitions from one system mode to another may not be salient, designated by small changes
in the displayed interface, yet creating very different system behavior. Although the systems are operating properly,
operators may be confused as they misinterpret observed system behavior in light of their mental model of a different
system mode.
With increased complexity, proving information clearly to operators so that they understand the system
state and state transitions becomes much more challenging. Operators may rarely see certain modes or combinations
of circumstances that lead to certain kinds of system behavior. Thus, their mental models of the systems may be
incomplete. This leaves operators unable to properly interpret observed system actions and predict future system
behavior, constituting a significant situation awareness problem.
Although problems with complexity and interface design are somewhat peripheral to automation per se (i.e.
these problems also exist in many systems quite apart from any automation considerations), these issues often
plague automated systems, and can significantly undermine operator situation awareness in working with automated
systems.
Benefits to Situation Awareness
It should be noted that automation does not always result in these types of problems. Wiener (1985)
points out that automation has, for the most part, worked quite well, and has accompanied a dramatic reduction in
many types of human errors. Furthermore, he believes that it may improve situation awareness by reducing the
display clutter and complexity associated with manual task performance, and through improved integrated displays
(Wiener, 1992; 1993).
It has been suggested that automation may also improve situation awareness by reducing excessive
workload (Billings, 1991). Curry and Ephrath (1977) found that monitors of an automatic system actually performed
better than manual controllers in a flight task. As monitors, the authors argued, subjects may have been able to
distribute their excess attention to other displays and tasks.
Recent research, however, demonstrates a certain degree of independence between situation awareness and
workload (Endsley, 1993a). Workload may only negatively impact situation awareness at very high levels of
workload. Low situation awareness can also accompany low levels of workload. If workload is reduced through
automation, therefore, this may not translate into higher situation awareness.
Furthermore, whether automation actually results in lower workload remains questionable. Wieners (1985)
studies showed that pilots report automation does not reduce their workload, but actually may increase it during critical
portions of the flight. Many recent studies are beginning to confirm this. Harris, Goernert, Hancock, and Arthur
(1994) and Parasuraman, Mouloua, and Molloy (1994) showed that operator initiation of automation under high
workload may increase workload even more. Riley (1994) augments this with his finding that a subjects choice to use
automation for a task is not related to the workload level of the task. Grubb et. al. (1994) showed that workload
actually was fairly high in tasks where humans must act as monitors over a period of time, as they do with automated
systems. Automation in many ways may serve to increase workload, particularly when workload is already high.
Bainbridge (1983) called it the irony of automation that when workload is highest automation is of the least assistance.
Despite this, however, workload remains the fundamental human factors consideration in many automation decisions.
format. As subtle changes in the form of information can impact situation awareness, it is critical that proposed
designs be tested thoroughly (within the context of the operators problem domain and in conjunction with multiple
task demands) for possible insidious effects on situation awareness. Careful consideration needs to be given to
providing interpretable and comprehensible information (that maps to the operators goals), as opposed to volumes
of low level data, in order to better meet operator needs.
New Approaches To Automation
Many of the issues surrounding the negative impact of automation on situation awareness and human
performance may be attributable not to automation itself, but the way that automation has traditionally been
implemented in many systems. In most cases, the performance of some task has been given over to the automation
and the humans job has become that of monitor. New approaches are currently being explored that challenge this
division. These approaches redefine the assignment of functions to people and automation in terms of a more
integrated team approach. Two orthogonal and possibly complementary approaches can be defined along the axes of
Figure 1. One approach seeks to optimize the assignment of control between the human and automated system by
keeping both involved in system operation. The other recognizes that control must pass back and forth between the
human and the automation over time, and seeks to find ways of using this to increase human performance.
Level of
Automation
(Control)
Adaptive
Automation
When?
How
Much?
What?
Tasks
Figure 1
Level of control. One way to minimize the negative effects of automation is to devise implementation
schemes that keep the human actively involved in the decision making loop while simultaneously reducing the load
associated with doing everything manually. This can be accomplished by determining a level of automation that
minimizes negative impacts on operator situation awareness (Endsley, 1987b; 1993b).
Wiener and Curry (1980) and Billings (1991) have both discussed the fact that automation does not exist in an
all or none fashion, but can be implemented at various levels. With regard to the automation of cognitive tasks
(through artificial intelligence or expert systems), a task may be accomplished (a) manually with no assistance from
the system, (b) by the operator with input in the form of recommendations provided by the system, (c) by the system,
with the consent of the operator required to carry out the action, (d) by the system, to be automatically implemented
unless vetoed by the operator, or (e) fully automatically, with no operator interaction (Endsley, 1987b; 1993b). This
can be viewed as five possible levels of automation from none to full, as depicted in Figure 2.
Endsley and Kiris (in press) implemented automation of an automobile navigation task at each of these five
levels. They found that the out-of-the-loop performance problem was significantly greater under full automation
than under intermediate levels of automation. This corresponded with a greater decrement in situation awareness
under full automation than under intermediate levels, as compared to manual control. By implementing functions at
a lower level of automation, leaving the operator involved in the active decision making loop, situation awareness
remained at a higher level and subjects were more able to assume manual control when needed.
Thus, even though full automation of a task may be technically possible, it may not be desirable if the
performance of the joint human-machine system is to be optimized. Intermediate levels of automation may be
preferable for certain tasks, in order to keep human operators situation awareness at a higher level and allow them to
perform critical functions.
Roles
Level of Automation
Figure 2
Human
System
None
Decide, Act
Decison Support
Decide, Act
Suggest
Consentual AI
Concur
Decide, Act
Monitored AI
Veto
Decide, Act
Full Automation
Decide, Act
Levels of Control and Automation (Adapted from Endsley and Kiris, 1995)
Adaptive automation. Recent work on adaptive automation has also been found to aid in overcoming the
out-of-the-loop performance problem (Parasuraman, 1993). Adaptive automation recognizes that over the course of
time, control of tasks may need to pass back and forth between an operator and an automated system in response to
changing demands. Adaptive automation attempts to optimize this dynamic allocation of tasks by creating a
mechanism for determining in real-time when tasks need to become automated (or manually controlled) (Morrison,
Cohen, & Gluckman, 1993). In direct contrast to historical efforts which have featured fixed task allocation
assignments, adaptive automation provides the potential for improving operator performance with automated systems
by continually adjusting to operator needs and keeping operators in the loop. In recent research, Carmody and
Gluckman (1993) found Level 2 SA to be impacted by adaptive automation of certain tasks.
Issues. Many questions remain in exploring the problem space set forth by the approach in Figure 1.
Most notably, the characteristics of tasks that determine their optimal level of control and suitability for adaptive
automation need to be investigated. Gluckman et. al. (1993), for instance, found different effects on workload and
performance for adaptive automation involving a static (system monitoring) versus a dynamic (resource management)
task. Carmody and Gluckman (1993) also found situation awareness to be more effected by adaptive automation of a
dynamic task than a static task. Lewandowski, Durso and Grounlund (1994) have proposed that if parts of an
integrated task are automated, more performance decrements will occur than if the whole task is automated. Endsley
and Kiris (in press) propose that psychomotor, perceptual and cognitive tasks may be differentially affected by
automation level. Considerable work is needed to determine the critical dimensions of tasks for successful
implementation of adaptive automation and specifying an optimal level of control for a given task.
Secondly, when adaptive automation should be invoked needs to be determined. In a first look at this issue,
Parasuraman (1993) examined whether manual control implemented at a pre-set periodic interval differed in effect
from manual control implemented on the basis of poor monitoring performance (indicating a loss of situation
awareness). He found no differences between the two conditions in terms of their effect on subsequent human
monitoring performance under automation. The insertion of a period of manual control was equally beneficial in
both cases. This type of work needs to be extended to address questions of periodic insertion of automation into
manual tasks. Research is also needed to explore the interaction between adaptive automation and level of control
how much automation needs to be employed may be a function of when it is employed.
How adaptive automation should be implemented is also a question. Various schemes have been proposed
from purely manual control for turning the automation on and off, to system control for invoking automation based
on real-time monitoring of human performance, physiology, or flight conditions. This is not a simple matter,
however. Many systems have left it up to operators to invoke automation at their discretion. In critical situations
however, the operator may be (1) so overloaded as to make this an extra burden, (2) incapacitated or otherwise unable
to do so, (3) unaware that the situation calls for automated assistance, or (4) a poor decision maker. Leaving the
system with the ability to turn itself on and off may be even more problematic, as this taxes the operator with the
task of keeping up with what the system is doing. A major question lies in determining how adaptive automation
should be implemented so as to provide the potential benefits without leading to new problems of loss of system
awareness.
Thus, an increased level of human control and adaptive automation may provide means for keeping
operators sufficiently in-the-loop to avoid the situation awareness decrements that can occur with traditional all-ornothing function allocations. Significant research issues still need to be resolved to determine how to best
implement these concepts within operational settings.
Evaluation of Automated Systems
A direct consideration of the operator interface, the proper level of control for a task, and means of
transitioning between automated and non-automated states needs to made during the design process in order to address
fundamental problems associated with automation. Although these concepts have shown merit for improving
situation awareness, research to establish precise guidelines for such prescriptions is, to date, just beginning.
Lacking such guidance, and because the effects of automation can be quite insidious, careful testing during the
development of automated systems is imperative.
Several methods have been established for the measurement of situation awareness. (For a complete review see
Endsley (1994b).) Most involve the creation of a simulation of the system under consideration. Impact of a particular design
concept on situation awareness can be measured directly through either objective or subjective means, or it can be inferred
through less direct performance measures.
Objective measurement. The most commonly used means of objectively evaluating a design concepts
impact on situation awareness involves directly questioning operators as to their perceptions of critical aspects of the
system they are operating. The Situation Awareness Global Assessment Technique (SAGAT) (Endsley, 1987a;
1988b) is a technique wherein the simulation is frozen at randomly selected times, the system displays blanked and
the simulation suspended while subjects quickly answer questions about their current perceptions of the situation.
Subject perceptions are then compared to the real situation based on simulation computer databases to provide an
objective measure of situation awareness.
SAGAT includes queries about all operator situation awareness requirements, including Level 1 (perception
of data), Level 2 (comprehension of meaning) and Level 3 (projection of the near future) components. This includes
a consideration of system functioning and status as well as relevant features of the external environment. This
approach minimizes possible biasing of attention, as subjects cannot prepare for the queries in advance since they
could be queried over almost every aspect of the situation to which they would normally attend.
SAGAT provides an objective, unbiased assessment of operator situation awareness that overcomes memory
problems incurred when collecting data after the fact, yet minimizes biasing of subject situation awareness due to
secondary task loading or artificially cueing the subjects attention. Empirical, predictive and content validity has
been demonstrated for this technique (Endsley, 1989; 1990a; 1990b).
Subjective measurement. Subjective measures of situation awareness are easier and less expensive to
administer than objective measures, but may lack the same degree of accuracy and diagnosticity. The most commonly
used method is to have operators provide ratings of their situation awareness with system concepts along a designated
scale.
Taylor (1989) developed the Situational Awareness Rating Technique (SART) which has operators rate
system designs on the amount of demand on attentional resources, supply of attentional resources and understanding
of the situation provided. As such, it considers operators perceived workload (supply and demand on attentional
resources) in addition to their perceived understanding of the situation. While SART has been shown to be correlated
with performance measures (Selcon & Taylor, 1989), it is unclear whether this is due to the workload or the
understanding components.
Performance measurement. In general, performance measures provide the advantage of being objective and
are usually non-intrusive. Simulation computers can be programmed to record specified performance data
automatically, making the required data relatively easy to collect. Hansman, et al. (1992), for example, used detection
of clearance amendment errors as a measure of aircrew situation awareness in evaluating the use of an automated
datalink system for updating the onboard flight management computer. As many other factors can act to influence
subject performance measures (such as decision making, workload impacts, actions or individual strategy differences),
these are also limited for inferring subject situation awareness by themselves.
General measurement considerations. One of the biggest difficulties associated with automation is its
insidious effect on situation awareness and performance. An increased tendency for out-of-the-loop performance
problems may be difficult to detect during testing. For this reason, it is preferable to measure situation awareness
directly, in addition to evaluating operator performance with an automation concept.
It also needs to be recognized that automation can effect situation awareness in a global, not readily
predicted manner. A new system may provide more situation awareness on one factor, but simultaneously reduce
situation awareness on others. Assessment of the impact of automation on situation awareness needs to take into
account operator situation awareness accross the range of requirements.
Finally, it can take quite a bit of training for operators to feel comfortable with an automation concept and
be able to adequately perform with it. Assessments of any automation concept need to be made after operators have
become proficient with the system, if a fair evaluation is to be made. Of course, if that proficiency takes an
inordinate amount of time (years in the case of previously mentioned aircraft systems), system deficiencies are
certainly indicated.
CONCLUSION
In conclusion, the successful implementation of automation is a complex issue. The traditional form of
automation which places humans in the role of monitor has been shown to negatively impact situation awareness
and thus their ability to effectively perform that function. Losses in situation awareness can be attributed to the
unsuitability of humans to perform a monitoring role, assumption of a passive role in decision making, and
inadequate feedback associated with automation. As a result many automated systems have been suboptimized, with
infrequent, but major errors attributed to a failure of the human component. This unsatisfactory state of affairs
currently plagues many automation efforts. New approaches to automation design that seek to fundamentally alter
the role of the human operator in interacting with the automated system provide a great deal of promise for
surmounting this problem. Careful test and evaluation of proposed automation concepts is imperative for
establishing that adequate situation awareness is maintained to keep human operators in the loop and able to fulfill
their functions.
REFERENCES
Bainbridge, L. (1983). Ironies of automation. Automatica, 19, 775-779.
Billings, C. E. (1991). Human-centered aircraft automation: A concept and guidelines (NASA Technical
Memorandum 103885). Moffet Field, CA: NASA Ames Research Center.
Carmody, M. A., & Gluckman, J. P. (1993). Task specific effects of automation and automation failure on
performance, workload and situational awareness. In R. S. Jensen & D. Neumeister (Eds.), Proceedings of the
Seventh International Symposium on Aviation Psychology (pp. 167-171). Columbus, OH: Department of
Aviation, The Ohio State University.
Cowan, N. (1988). Evolving conceptions of memory storage, selective attention, and their mutual constraints within
the human information processing system. Psychological Bulletin, 104(2), 163-191.
Curry, R. E., & Ephrath, A. R. (1977). Monitoring and control of unreliable systems. In T. B. Sheridan & G.
Johannsen (Eds.), Monitoring behavior and supervisory control (pp. 193-203). New York: Plenum Press.
Danaher, J. W. (1980). Human error in ATC system operations. Human Factors, 22(5), 535-545.
Endsley, M. R. (1987a). SAGAT: A methodology for the measurement of situation awareness (NOR DOC 87-83).
Hawthorne, CA: Northrop Corporation.
Endsley, M. (1987b). The application of human factors to the development of expert systems for advanced cockpits.
In Proceedings of the Human Factors Society 31st Annual Meeting (pp. 1388-1392). Santa Monica, CA:
Human Factors Society.
Endsley, M. R. (1988a). Design and evaluation for situation awareness enhancement. In Proceedings of the Human
Factors Society 32nd Annual Meeting (pp. 97-101). Santa Monica, CA: Human Factors Society.
Endsley, M. R. (1988b). Situation Awareness Global Assessment Technique (SAGAT). In Proceedings of the
National Aerospace and Electronics Conference (NAECON) (pp. 789-795). New York: IEEE.
Endsley, M. R. (1989). A methodology for the objective measurement of situation awareness. In Situational
Awareness in Aerospace Operations (AGARD-CP-478) (pp. 1/1 - 1/9). Neuilly Sur Seine, France: NATO AGARD.
Endsley, M. R. (1990a). Predictive utility of an objective measure of situation awareness. In Proceedings of the
Human Factors Society 34th Annual Meeting (pp. 41-45). Santa Monica, CA: Human Factors Society.
10
Endsley, M. R. (1990b). Situation awareness in dynamic human decision making: Theory and measurement.
Unpublished doctoral dissertation, University of Southern California, Los Angeles, CA.
Endsley, M. R. (1993a). Situation awareness and workload: Flip sides of the same coin. In R. S. Jensen & D.
Neumeister (Eds.), Proceedings of the Seventh International Symposium on Aviation Psychology (pp. 906911). Columbus, OH: Department of Aviation, The Ohio State University.
Endsley, M. R. (1993b). Situation awareness: A fundamental factor underlying the successful implementation of AI
in the air traffic control system. In D. J. Garland & J. A. Wise (Eds.), Human Factors and Advanced
Automation Technologies (pp. 117-122). Daytona Beach, FL: Embry-Riddle Aeronautical University Press.
Endsley, M. R. (1994a, March). A taxonomy of situation awareness errors. Paper presented at the Western European
Association of Aviation Psychology 21st Conference, Dublin, Ireland.
Endsley, M. R. (1994b). Situation awareness in dynamic human decision making: Measurement. In R. D. Gilson,
D. J. Garland, & J. M. Koonce (Eds.), Situational awareness in complex systems (pp. 79-97). Daytona Beach,
FL: Embry-Riddle Aeronautical University Press.
Endsley, M. R., & Kiris, E. O. (1995). The Out-of-the-Loop Performance Problem and Level of Control in
Automation. Human Factors, 37(2), 381-194.
Ephrath, A. R., & Young, L. R. (1981). Monitoring vs. man-in-the-loop detection of aircraft control failures. In J.
Rasmussen & W. B. Rouse (Eds.), Human detection and diagnosis of system failures. New York: Plenum
Press.
Gluckman, J. P., Carmody, M. A., Morrison, J. G., Hitchcock, E. M., & Warm, J. S. (1993). Effects of allocation
and partitioning strategies of adaptive automation on task performance and perceived workload in aviation
relevant tasks. In R. S. Jensen & D. Neumeister (Eds.), Proceedings of the Seventh International Symposium
on Aviation Psychology (pp. 150-155). Columbus, OH: Department of Aviation, The Ohio State University.
Grubb, P. L., Miller, L. C., Nelson, W. T., Warm, J. S., & Dember, W. N. (1994). Cognitive failure and perceived
workload in vigilance performance. In M. Mouloua & R. Parasuraman (Eds.), Human performance in automated
systems: Current research and trends (pp. 115-121). Hillsdale, NJ: LEA.
Hansman, R. J., Wanke, C., Kuchar, J., Mykityshyn, M., Hahn, E., & Midkiff, A. (1992, September). Hazard
alerting and situational awareness in advanced air transport cockpits. Paper presented at the 18th ICAS Congress,
Beijing, China.
Harris, W. C., Goernert, P. N., Hancock, P. A., & Arthur, E. (1994). The comparative effectiveness of adaptive
automation and operator initiated automation during anticipated and unanticipated taskload increases. In M.
Mouloua & R. Parasuraman (Eds.), Human performance in automated systems: Current research and trends (pp.
40-44). Hillsdale, NJ: LEA.
Hartel, C. E., Smith, K., & Prince, C. (1991, April). Defining aircrew coordination: Searching mishaps for
meaning. Paper presented at the Sixth International Symposium on Aviation Psychology, Columbus, OH.
Kantowitz, B. H., & Sorkin, R. D. (1983). Human factors: Understanding people-system relationships. New York:
John Wiley & Sons.
Kessel, C. J., & Wickens, C. D. (1982). The transfer of failure-detection skills between monitoring and controlling
dynamic systems. Human Factors, 24(1), 49-60.
Kuipers, A., Kappers, A., van Holten, C. R., van Bergen, J. H. W., & Oosterveld, W. J. (1989). Spatial
disorientation incidents in the R.N.L.A.F. F16 and F5 aircraft and suggestions for prevention. In Situational
Awareness in Aerospace Operations (AGARD-CP-478) (pp. OV/E/1 - OV/E/16). Neuilly Sur Seine, France:
NATO - AGARD.
Lee, J., & Moray, N. (1992). Trust, control strategies and allocation of function in human-machine systems.
Ergonomics, 35(10), 1243-1270.
Lewandowski, S., Durso, F. T., & Grounlund, S. D. (1994). Modular automation: Automating sub-tasks without
disrupting task flow. In M. Mouloua & R. Parasuraman (Eds.), Human performance in automated systems:
Current research and trends (pp. 326-331). Hillsdale, NJ: LEA.
McClumpha, A., & James, M. (1994). Understanding automated aircraft. In M. Mouloua & R. Parasuraman (Eds.),
Human performance in automated systems: Current research and trends (pp. 183-190). Hillsdale, NJ: LEA.
Morrison, J., Cohen, D., & Gluckman, J. P. (1993). Prospective principles and guidelines for the design of
adaptively automated crewstations. In R. S. Jensen & D. Neumeister (Eds.), Proceedings of the Seventh
International Symposium on Aviation Psychology (pp. 172-177). Columbus, OH: Department of Aviation,
The Ohio State University.
National Transportation Safety Board (1988). Aircraft accident report: Northwest Airlines, Inc., McDonnell-Douglas
DC-9-82, N312RC, Detroit Metropolitan Wayne County Airport, August, 16, 1987 (NTSB/AAR-99-05).
Washington, D. C.: Author.
11
National Transportation Safety Board (1990). Aircraft accident report: US Air, Inc., Boeing 737-400, LaGuardia
Airport, Flushing New York, September 20, 1989 (NTSB/AAR-90-03). Washington, D.C.: Author.
Norman, D. A. (1989). The problem of automation: Inappropriate feedback and interaction not overautomation (ICS
Report 8904). La Jolla, CA: Institute for Cognitive Science, U. C. San Diego.
Parasuraman, R. (1987). Human-computer monitoring. Human Factors, 29(6), 695-706.
Parasuraman, R. (1993). Effects of Adaptive Function Allocation on Human Performance. In D. J. Garland & J. A.
Wise (Eds.), Human factors and advanced aviation technologies (pp. 147-158). Daytona Beach, FL: EmbryRiddle Aeronautical University Press.
Parasuraman, R., Molloy, R., & Singh, I. L. (1993). Performance consequences of automation-induced
complacency. International Journal of Aviation Psychology, 3(1), 1-23.
Parasuraman, R., Mouloua, M., & Molloy, R. (1994). Monitoring automation failures in human-machine systems.
In M. Mouloua & R. Parasuraman (Eds.), Human performance in automated systems: Current research and
trends (pp. 45-49). Hillsdale, NJ: LEA.
Riley, V. (1994). A theory of operator reliance on automation. In M. Mouloua & R. Parasuraman (Eds.), Human
performance in automated systems: Current research and trends (pp. 8-14). Hillsdale, NJ: LEA.
Rudisill, M. (1994). Flight crew experience with automation technologies on commercial transport flight decks. In
M. Mouloua & R. Parasuraman (Eds.), Human performance in automated systems: Current research and trends
(pp. 203-211). Hillsdale, NJ: LEA.
Sarter, N. B., & Woods, D. D. (1992). Pilot interaction with cockpit automation: Operational experiences with the
flight management system. The International Journal of Aviation Psychology, 2(4), 303-321.
Selcon, S. J., & Taylor, R. M. (1989). Evaluation of the situational awareness rating technique (SART) as a tool
for aircrew systems design. In Situational Awareness in Aerospace Operations (AGARD-CP-478) (pp. 5/1 -5/8).
Neuilly Sur Seine, France: NATO - AGARD.
Singh, I. L., Molloy, R., & Parasuraman, R. (1993). Automation-induced complacency: Development of the
complacency-potential rating scale. International Journal of Aviation Psychology, 3(2), 111-122.
Slamecka, N. J., & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental
Psychology: Human Learning and Memory, 4(6), 592-604.
Stein, K. J. (1983). Human factors analyzed in 007 navigation error. Aviation Week & Space Technology,
119(October 3), 165-167.
Taylor, R. M. (1989). Situational awareness rating technique (SART): The development of a tool for aircrew
systems design. In Situational Awareness in Aerospace Operations (AGARD-CP-478) (pp. 3/1 - 3/17). Neuilly
Sur Seine, France: NATO - AGARD.
Wickens, C. D. (1992). Engineering Psychology and Human Performance (2nd ed.). New York: Harper Collins.
Wickens, C. D., & Kessel, C. (1979). The effect of participatory mode and task workload on the detection of
dynamic system failures. IEEE Transactions on Systems, Man and Cybernetics, SMC- 9(1), 24-34.
Wiener, E. L. (1985). Cockpit automation: In need of a philosophy. In Proceedings of the 1985 Behavioral
Engineering Conference (pp. 369-375). Warrendale, PA: Society of Automotive Engineers.
Wiener, E. L. (1989). Human factors of advanced technology (glass cockpit) transport aircraft (NASA Contractor
Report No. 177528). Moffett Field, CA: NASA-Ames Research Center.
Wiener, E. L. (1992, June). The impact of automation on aviation human factors . Paper presented at the
NASA/FAA Workshop on Artificial Intelligence and Human Factors in Air Traffic Control and Aviation
Maintenance, Daytona Beach, FL.
Wiener, E. L. (1993). Life in the second decade of the glass cockpit. In R. S. Jensen & D. Neumeister (Eds.),
Proceedings of the Seventh International Symposium on Aviation Psychology (pp. 1-11). Columbus, OH:
Department of Aviation, The Ohio State University.
Wiener, E. L., & Curry, R. E. (1980). Flight deck automation: Promises and problems. Ergonomics, 23(10), 9951011.
Young, L. R. A. (1969). On adaptive manual control. Ergonomics, 12(4), 635-657.