Behavior Research Methods
2009, 41 (1), 118-127
doi:10.3758/BRM.41.1.118

ATC-labAdvanced: An air traffic control
simulator with realism and control
SELINA FOTHERGILL
University of Queensland, Brisbane, Queensland, Australia

SHAYNE LOFT
University of Western Australia, Perth, Western Australia, Australia
AND

ANDREW NEAL
University of Queensland, Brisbane, Queensland, Australia
ATC-labAdvanced is a new, publicly available air traffic control (ATC) simulation package that provides both
realism and experimental control. ATC-labAdvanced simulations are realistic to the extent that the display features
(including aircraft performance) and the manner in which participants interact with the system are similar to those
used in an operational environment. Experimental control allows researchers to standardize air traffic scenarios,
control levels of realism, and isolate specific ATC tasks. Importantly, ATC-labAdvanced also provides the programming control required to cost effectively adapt simulations to serve different research purposes without the need
for technical support. In addition, ATC-labAdvanced includes a package for training participants and mathematical
spreadsheets for designing air traffic events. Preliminary studies have demonstrated that ATC-labAdvanced is a flexible tool for applied and basic research.

Air traffic control (ATC) simulations are frequently used
for both applied and basic research. There is a growing
need for ATC simulations, to identify factors that influence
the workload and performance of air traffic controllers
(Athenes, Averty, Puechmorel, Delahaye, & Collet, 2002;
Lamoureux, 1999) and to build theories of the representations and processes that underlie performance on specific
control tasks (Gronlund, Ohrt, Dougherty, Perry, & Manning, 1998; Rantanen & Nunes, 2005). In addition, ATC
simulations are frequently used to address more basic issues of human cognition, such as the associative learning

mechanisms involved in relative judgment (Loft, Neal, &
Humphreys, 2007), the processes that underlie memory in
the performance of intended actions (Stone, Dismukes, &
Remington, 2001), the effects of time pressure on processing load (Hendy, Liao, & Milgram, 1997), and individual
differences in complex skill acquisition (Ackerman, 1992).
Consequently, ATC simulations are effective tools for evaluating the generalizability of broader theories about basic
cognitive processes and capacities, thus explaining human
performance more generally. In this article, we describe a
new ATC simulation package called ATC-labAdvanced that
can be used for both applied and basic research. In doing
so, we highlight the improvements it offers over currently
available ATC simulators.
Existing ATC simulators have typically been developed
so as to have the level of realism and experimental control

required to investigate specific research questions. Realism refers to the extent to which experiences encountered
in the simulation occur in the field of interest (DiFonzo,
Hantula, & Bordia, 1998; Ehret, Gray, & Kirschenbaum,
2000). Experimental control refers to the degree to which
a simulation can provide control over variables and thus
support the conclusion that the effects obtained are due to
experimental manipulations (Boring, 1954; Brehmer &
Dorner, 1993). To maximize efficiency, existing simulators have typically been designed to compromise between
the extent to which they can mimic field experience (realism) and the experimental control that they can provide.
High-fidelity ATC simulators typically have high realism
but lack experimental control. Medium-/low-fidelity simulators can provide this control but often lack realism.
This trade-off between realism and experimental control presents a problem when both are required. For example, many research groups are developing theories and
models designed to predict controller performance in field
settings (for a review, see Loft, Sanderson, Neal, & Mooij,
2007). For this type of research, it is crucial to use simulations that are representative of the environmental context in which experts make decisions (Brunswick, 1956;

Simon, 1956). At the same time, experimental control
is required in order to isolate the effects of independent
variables on specific ATC control tasks. In contrast, the
purpose of more basic research may be to test a specific

S. Fothergill,

© 2009 The Psychonomic Society, Inc.

118


ATC-LABADVANCED
theoretical issue that is prevalent in a range of applied settings in which individuals monitor dynamic multi-item
displays (e.g., military command, radar system operators).
In these circumstances, it may be desirable to have low
correspondence (cf. Gray, 2002) between the simulation
and the operational environment, so that the research can
be generalized to other systems (Berkowitz & Donnerstein, 1982; Mook, 1983). In other circumstances, ATC
simulations may be conducted to assess the effectiveness
of controller team performance or training programs, and
increased experimental control would add little to improving the outcomes of the research.
This highlights a need for an ATC simulation package
in which realism and control can be systematically varied
according to the research question(s) under investigation.
In the present article, we present a new ATC simulator
called ATC-labAdvanced that provides this. Importantly,
ATC-labAdvanced also provides the programming control
required for researchers to customize the exact levels of
realism and control they require in their simulations. The

aim of the present article is to introduce ATC-labAdvanced
and indicate how it can be used for research. First, we will
detail the features of ATC-labAdvanced that provide realism,
experimental control, and programming control. These
features will then be compared with those of existing
simulators. We will then provide examples of applied and
basic research programs that have used ATC-labAdvanced.
Next, we will outline the training package available to familiarize participants with ATC-labAdvanced simulations.
Finally, data logging features and system requirements
will be provided.
Realism in ATC-labAdvanced
The primary duties for air traffic controllers are to enforce separation standards between aircraft and ensure that
aircraft reach their destinations in an orderly and expeditious manner. One of the more common separation standards set by the International Civil Aviation Organization
(ICAO) is that aircraft are required to maintain either a
1,000-ft vertical separation or 5 nautical miles horizontal
separation from all other aircraft. Consequently, a pair of
aircraft is considered to be in conflict if they will, given
their current speeds, flight levels (altitudes), and bearings,
simultaneously violate vertical and horizontal separation
standards in the future. Controllers are required to perform
a range of control activities to ensure the safe and efficient
flow of aircraft. When logical, practical, or logistical considerations constrain field experimentation or observation
in an applied work context (DiFonzo et al., 1998; Gray,
2002), high-fidelity simulators can be used to simulate
these tasks. Examples include the FAA Academy Training
Simulator (Jones & Endsley, 2000), TRACON (Ackerman, 1992), the EUROCONTROL Simulation Capability
and Platform for Experimentation, ATCoach (UFA Inc.,
n.d.), and FIRSTplus (Raytheon, 2005). ATC-labAdvanced
simulations can also be designed so that participants perform tasks in a manner similar to field controllers.
Display realism. The first requirement for achieving

realism was to ensure that the components of the ATC-

119

labAdvanced display resembled the ATC operational environment in as many ways as possible (Schiff, Arnone, &
Cross, 1994). To achieve this, the ATC-labAdvanced display
was based on the Australian Air Traffic Management System and was developed in close collaboration with subject
matter experts.
Figure 1 illustrates a generic example of the display
used in a high-fidelity ATC-labAdvanced simulation. The
sector that the participant controls (the active sector) is
made up of a series of flight paths, waypoints, and airports
presented against a light gray background. The surrounding darker gray background represents adjacent and approach sectors (sectors that contain airports). Small green
circles symbolize aircraft track symbols, and each aircraft
has a data block label that displays the call sign, aircraft
type, ground speed, current flight level, and cleared flight
level. These aircraft track symbols and data blocks can be
fully customized. ATC-labAdvanced uses nautical miles for
distance, knots for ground speed, and feet for altitude.
Every 5 sec, each aircraft’s position and data block label
information is updated. Aircraft enter the active sector on
inbound flight paths from adjacent sectors or take off from
airports in approach sectors. They then proceed as denoted
in their flight plan through the series of waypoints and either
land at an airport or exit to adjacent sectors on outbound
flight paths. Aircraft that cruise at flight levels below or
above the sector flight level boundary of the active sector (over flights) can also be simulated. Importantly, ATClabAdvanced simulates aircraft performance data (e.g., climb
and descent rate, speed rate) accurately for commercial
jets, turbo propeller aircraft, and military aircraft. As a result, aircraft can transit through sectors in a manner similar
to that for an ATC operational environment.

The notification system used to denote transitions in
aircraft states can be closely based on ATC operational
environments. That is, the attributes (e.g., colors, flashing) of aircraft track symbols and data block labels can be
set to represent different phases of flight, which change
dynamically as aircraft move though sectors. For example, an aircraft approaching an active sector from an adjacent sector may be set to turn from black to blue when
it reaches a certain distance (e.g., 10 nm) from the active
sector. As the aircraft travels closer to the active sector, it
may be set to flash orange until the controller officially
“accepts” the aircraft, using a specific sequence of actions, at which point it would turn green to denote that it
is under the jurisdiction of that controller. When the aircraft is handed off to the adjacent sector or approaches the
airport, it would turn black to indicate that it is no longer
under the jurisdiction of that controller.
Response-system realism. The second important requirement for achieving realism was to ensure that participants performing control tasks would be able to interact
with the ATC-labAdvanced system as similarly as possible
to how controllers would interact with ATC systems in the
field (Schiff et al.,1994). ATC-labAdvanced can be customized to provide simulations of the major control tasks previously identified in cognitive task analyses of ATC (Cox,
1994; Rodgers & Drechsler, 1993). These control tasks


120

FOTHERGILL, LOFT, AND NEAL

Figure 1. A generic example of a display used in an ATC-labAdvanced simulation. The screen shot
displays one active and six (four adjacent, two approach) nonactive sectors, various route structures,
and aircraft in their different phases of flight. All the aircraft have probe minute vectors to indicate
their position in 1 min’s time; the route for SIA16 is displayed; a scale marker is available in the
top left corner; and a bearing and range line has been attached to VOZ555. To resolve the potential
conflict between VOZ555 and VOZ892, VOZ555 is being vectored away from its planned route. The
clock is paused, and the mode display shows that a vector solution is being used.


include accepting and handing off aircraft from adjacent
sectors or airports, assigning boundary and cruise altitudes, monitoring air traffic to detect potential conflicts,
resolving conflicts, and traffic sequencing.
The intervention methods participants use to modify aircraft trajectories in ATC-labAdvanced, the way participants
accept and hand off aircraft, and how they use prediction
tools were designed on the basis of structured interviews
with controllers (Fothergill & Neal, 2005, 2006) and analyses of the ATC literature (Callantine, 2002; Späth & Eyferth, 2001). Examples of aircraft intervention methods
include changing flight levels, speeds, or headings and assigning flight-level requirements (e.g., reaching a flight

level by a certain distance). Flight levels or speeds can be
altered by clicking on the data block label where these values are displayed and then choosing new values from dropdown menus. An example of how to change a flight level is
illustrated in Figure 2. Similarly, heading changes can be
chosen from drop-down menus. Headings of aircraft can
be changed by selecting a predetermined heading function on the keyboard, clicking on the aircraft, and dragging
a line to a new destination point. Level requirements can
be issued by pressing designated keys and entering into
text boxes the distances by which aircraft are required to
reach certain flight levels. Participants can accept aircraft
by pressing designated keys and clicking on aircraft track

Figure 2. Changing the cleared flight level of an aircraft. By clicking on the
current cleared flight level, a new cleared flight level can be selected from the
menu. The new level will be displayed in the aircraft’s label in the next 5-sec
update.


ATC-LABADVANCED

Figure 3. The bearing and range line tool. This shows the distance between the aircraft and the selected end point (in nautical

miles), the bearing (in degrees), and the time that it would take
the aircraft to reach the selected end point (in minutes) based on
its indicated speed.

symbols. Similar to ATC operational environments, handoffs can be designed to occur automatically at a set distance (e.g., 5 nm) beyond the sector boundary.
Prediction tools in ATC-labAdvanced include scale markers, bearing and range lines, probe vectors, route displays,
and history dots. These tools are regularly used by controllers in the field. Scale markers are moved around the screen
to measure distance. Bearing and range lines indicate distance (in nautical miles), bearing (in degrees), and the time
(in minutes) to a future waypoint or another aircraft. An
example of how to use the bearing and range line function is
illustrated in Figure 3. Route displays indicate the planned
routes of aircraft, punctuated by the times at which the aircraft are predicted to reach waypoints, on the basis of their
current nominal trajectory. History dots are displayed behind aircraft and represent the routes that aircraft have traveled. Probe vectors display the predicted position of aircraft
(in a specified number of minutes) in the horizontal plane,
on the basis of their current nominal trajectory.
Realism: Comparison with existing ATC simulators. A significant limitation of existing low- and
medium-fidelity ATC simulators is that they lack display
realism and response system realism. One prototypical
example is our medium-/low-fidelity predecessor to ATClabAdvanced, which we called ATC-lab (Loft, Hill, Neal,
Humphreys, & Yeo, 2004). ATC-lab simulations are realistic for participants to the extent that they involve and affect participants and to the extent that participants take the
simulations seriously (DiFonzo et al., 1998). However, a
major limitation of ATC-lab is that it simulates very selective aspects of ATC. ATC-lab has low display realism because it does not simulate features such as aircraft altitude,
does not use real aircraft performance profiles, does not
present adjacent/approach sectors, and does not provide
any notification system for denoting aircraft transition
states. In addition, ATC-lab has low response system realism because it simulates very few control tasks (conflict
detection/resolution only), provides a very limited number
of intervention methods for modifying aircraft trajectory
(speed change only), and provides no prediction tools. The
medium-fidelity ATC simulators used by Metzger and

Parasuraman (2001) and Remington, Johnston, Ruthruff,
Gold, and Romera (2000; also see Stone et al., 2001) also

121

generally have low display realism and low system response realism. For example, these simulators do not all
have the capability to simulate changes in aircraft altitude,
do not allow participants to interact with the ATC system
in order to modify aircraft trajectory only in limited ways,
and do not provide access to prediction tools.
This lack of realism does not present a problem when
the intent of the research is to test a theoretical idea by
mapping the functional relations between variables in a
simulation, rather than generalizing to a specific domain.
ATC-lab, for example, has been successfully used to develop general theories (Loft, Humphreys, & Neal, 2004;
Loft, Neal, & Humphreys, 2007; Yeo & Neal, 2004) and
computational models (Kwantes, Neal, & Loft, 2004) of
the processes by which individuals make decisions about
the movement of objects on radar displays. However, the
lack of realism is problematic when one is building theories and models of performance that apply directly to ATC
operations (Kopardekar & Magyarits, 2003; Laudeman,
Shelden, Branstrom, & Brasil, 1998), since a lack of realism poses a substantial threat to the external validity of
results. For example, a researcher may be interested in
examining the processes underlying ATC conflict detection. Here, it would be essential that aircraft performance
is accurately simulated so that aircraft transit through sectors as they would in the field. Controllers must also have
access to their regular prediction tools, so they are able to
make aircraft trajectory predictions in a way that is similar
to how they would make them in the field.
There are many high-fidelity ATC simulators that can
provide levels of display realism and response system realism that are similar to (or better than) those in ATClabAdvanced. These include but are not limited to the FAA

Academy Training Simulator (Jones & Endsley, 2000), the
EUROCONTROL Simulation Capability and Platform for
Experimentation, FIRSTplus (Raytheon, 2005), and the
Total Airport and Airspace Modeler (TAAM) (Jeppesen,
2007). For example, TAAM runs real gate-to-gate traffic extracted from the Australian Air Traffic Management
System, and FIRSTplus replicates all the features of modern ATC radar situation displays and can even emulate
future operational ATC display types. However, as will
be discussed in the sections below, many of these highfidelity simulators are not made freely available for research, nor do they necessarily provide experimental control or programming control.
Experimental and Programming
Control in ATC-labAdvanced
ATC-labAdvanced provides the experimental control required to make definitive conclusions regarding the effects of independent variables on dependent variables.
Standardized air traffic scenarios can be presented that
control extraneous variables and separate confounding variables. Programming control refers to the extent
to which the researcher can control what is presented in
simulations. ATC-labAdvanced provides high programming
control over a wide range of task features. These task features include display realism, response system realism,
trial presentation, and presentation of rating scales. This


122

FOTHERGILL, LOFT, AND NEAL

programming control of ATC-labAdvanced is an important
advance, since it allows simulations to be adapted quickly
and cost effectively to serve different research purposes
without the need for technical support.
Standardized air traffic scenarios. ATC-labAdvanced
experimental scripts are used to specify aircraft events
that occur during experimental trials. An example is illustrated in Figure 4. These scripts are written using the Extensible Markup Language (XML) Version 1.0. This is a

free-to-use general purpose markup language, which can
be used as a generic framework for storing any amount
of text or any data whose structure can be represented as
a tree. In contrast to the text files used in ATC-lab, XML
scripts can be screened for errors before they are used in
experiments. Aircraft details specified in the scripts include call sign, type, minimum and maximum speed and
flight level, current speed, current flight level, starting
x- and y-coordinates, planned route, position (if any) for
automatic start of climb or descent, and climb and descent
rate. The values for aircraft call sign, aircraft type, ground
speed, current flight level (altitude), and cleared flight
level are derived from these scripts and are displayed on
aircraft data blocks. When participants intervene during
trials, these values are updated.
ATC-labAdvanced provides a set of mathematical spreadsheets to control the spatial (e.g., minimum separation,
angle of convergence) and temporal (e.g., time to minimum separation) characteristics of aircraft events. These
spreadsheets were developed to replace the script developer provided in the ATC-lab simulation package (Loft,
Hill, et al., 2004). The script developer represented a
substantial improvement over existing medium- and lowfidelity simulators because it improved the degree to which
air traffic scenarios could be standardized (see Loft, Hill,
et al., 2004, for a detailed description), and eliminated
the need for manual calculation or trial-by-error scripting. However, the ATC-lab script developer had two major
limitations. First, it was time consuming to use because

researchers were required to wait for the developer (for
up to 10 min) to generate starting x- and y-coordinates.
Second, it did not allow the calculation of vertical distance, which is essential for ATC-labAdvanced. With the
mathematical spreadsheets, researchers enter the desired
spatial and temporal characteristics of aircraft events, and
hard-coded formulae contained in these spreadsheets provide starting x- and y-coordinates for aircraft in the lateral

plane. These spreadsheets are accompanied by a report
documenting the underlying formulae. A scenario tester
is also included in the ATC-labAdvanced simulation package, which enables researchers to view (at a faster speed)
the air traffic scenarios that are being developed.
Programming control over task features. Due to
high levels of display realism and system response realism, ATC-labAdvanced simulates a much wider array of
potential task features than do many existing simulators.
Furthermore, a significant advantage of ATC-labAdvanced
is that the XML scripting language and code base architecture provide the researcher with programming control
over task features. First, researchers can control the realism of the display, which includes specifying the type of
sector (e.g., approach, en route, tower), active and inactive
sectors, route structures, position of waypoints, position
of airports, and weather patterns. Trials can be constructed
so that different sector maps with different traffic patterns
can be presented within the same experiment. Researchers
can control settings of the aircraft transition notification
system, such as the specific color used to denote aircraft
transitional states and the positions in sectors where aircraft automatically begin climbing or descending. Aircraft
performance can also be modified. Second, researchers
can control response system realism features, such as the
type of prediction tools available to participants and the
manner in which they are used, the type of methods that
participants can use to modify aircraft trajectory, and the
timing/content of instructions and questionnaire items
(e.g., workload ratings, motivation ratings). Third, re-

Figure 4. Specifications for an aircraft using XML scripting language. The
aircraft’s type, call sign, starting altitude, starting velocity, starting coordinates,
cleared flight level, and flight path are scripted.



ATC-LABADVANCED
searchers can control general features, such as the order of
presentation of trials, the timing and length of task breaks,
and when scenarios are paused.
Control: Comparison with existing ATC simulators. There are a handful of ATC simulators that provide
some level of experimental control. For example, both
ATC-lab (Loft, Hill, et al., 2004) and TRACON (Ackerman, 1992) can present standardized air traffic scenarios.
However, in comparison with ATC-labAdvanced, they provide little programming control. Ackerman noted that in
order to adapt TRACON to the study of skill acquisition,
the features of TRACON simulations needed to be considerably modified, which resulted in high programming
costs. This is the case with the original ATC-lab (Loft,
Hill, et al., 2004) as well. As a result, researchers using
simulators such as ATC-lab or TRACON would need to
hire a technical specialist to implement changes to simulation features. In addition, as was discussed previously,
many of these simulators have low realism.
Despite high realism, a significant limitation of many
existing high-fidelity ATC simulators (such as the FAA
Academy Training Simulator) is that they lack the experimental control required to make definitive conclusions regarding the effects of independent variables on dependent
variables (see Loft, Hill, et al., 2004). Furthermore, many
of these simulators and other high-fidelity simulators that
do provide better experimental control are not made freely
available for research (e.g., EUROCONTROL Simulation
Capability and Platform for Experimentation; ATCoach).
There are at least two ways in which experimental control is restricted in some high-fidelity simulators. First,
although general task conditions, such as the number of
aircraft, type and mix of aircraft, and flight paths, can be
controlled, little control is provided over the spatial and
temporal properties of aircraft events. A lack of standardization in air traffic scenarios makes it difficult to control
extraneous variables or to separate confounding variables.

This can present a problem, such as when the effects of
task demands on the time taken to complete specific control tasks are assessed. Task demands may include average
distance between aircraft, number of aircraft in altitude
transition, and number of potential conflicts. Without
control, researchers would be forced to extract values for
task demands from historic flight data in ATC simulations and correlate those values with performance on a
post hoc basis (e.g., Laudeman et al., 1998). This method
would make it difficult to determine how unique factors
and combinations of factors influence performance differentially (Loft, Sanderson, et al., 2007).
Second, the programming architecture underlying
many existing high-fidelity simulators is typically based
on an all-or-none philosophy, in that it does not provide
substantial experimental control over what is displayed
(e.g., altitude, maps), what specific ATC control tasks are
conducted (e.g., accepting aircraft, conflict detection), or
the manner in which participants interact with the ATC
system (e.g., intervention methods, prediction tools). A
consequence of this is that it is difficult to test predictions
about processing mechanisms underlying performance on
control tasks or to test specific theoretical questions.

123

In the next section of this article, we will provide examples of applied and basic research programs in which
ATC-labAdvanced simulations have been used. The degree
of realism and control used in the three research programs
were specifically tailored to the research question(s)
under investigation, demonstrating the flexibility of ATClabAdvanced as a tool for cognition research.
Illustrative Examples of
ATC-labAdvanced Simulations

The three main studies that have used ATC-labAdvanced
simulations to date are summarized in Table 1. Fothergill
and Neal (2008) used ATC-labAdvanced to examine the effect of workload on the selection of conflict resolution
strategies. Participating controllers managed traffic in
their sector and resolved potential conflicts as efficiently
as possible. The purpose was to inform the development
of a computational model that could simulate how controllers resolve conflicts in the field (Bolland, Fothergill,
& Humphreys, 2007). The key finding was that controllers were less likely to implement optimal conflict resolution strategies under a high workload than under a low
workload, but only in situations in which these strategies
were more difficult to calculate (see Table 1). To obtain
applicable results, the simulations were required to be
representative of ATC, especially in terms of (1) aircraft
performance, (2) sector structure, (3) aircraft transition
notification, (4) controller intervention methods, and
(5) prediction tool use. In order to systematically manipulate independent variables, a high degree of experimental
control was also required to vary configurations of air
traffic. For example, high-workload scenarios contained
configurations that produced more tasks (e.g., conflicts,
acceptances and handoffs, aircraft sequencing) than did
lower workload scenarios.
A recent issue raised in the experimental literature concerns how to capture expert performance across different
task domains (Ericsson & Williams, 2007). Loft, Bolland, and Humphreys (2007) recently developed a theory
of expertise for ATC conflict detection. ATC-labAdvanced
simulations were then used to test a series of predictions
from this theory that concerned the factors that affect the
likelihood of controllers intervening to ensure separation
between aircraft. In addition, data were used to test the
development of a computational model that simulates
how controllers detect conflicts in the field (Loft, Bolland, & Humphreys, 2007). Thus, it was essential for
ATC-labAdvanced to simulate the environmental context in

which controllers make conflict detection decisions. In
particular, it was critical that controllers have access to
their regular prediction tools, such as range and bearing
lines, in order to ensure that they acquire aircraft trajectory information in a realistic manner.
However, in contrast to Fothergill and Neal (2008), ATClabAdvanced was programmed in such a way that controllers
performed only conflict detection. By using the programming control available in ATC-labAdvanced to remove other
ATC control tasks, Loft, Bolland, and Humphreys (2007)
isolated conflict detection by eliminating visual search
requirements and competing demands on attention (see


124

FOTHERGILL, LOFT, AND NEAL
Table 1
Summary of the Three Main Studies That Have Used ATC-labAdvanced Simulations
Research
Questions
1. What is the effect of
workload on conflict resolution decisions?
2. Can we computationally model conflict
resolution heuristics as a
function of workload?

Independent
Variables
1. Workload level
of scenario (high
vs. low)
2. Difficulty of

calculating the optimal solution (difficult vs. easy)

Dependent
Variables
1. Conflict resolution strategy

13 endorsed air
traffic controllers
and 7 trainee controllers (1 year
training)

1. What aircraft geometry
factors affect the probability that controllers will
intervene to ensure separation between aircraft?
2. Will intervention decisions differ as a function
of controller experience?
3. Can the psychological processes underlying
these intervention decisions be captured by a
computational model?

1. Distance of
minimum lateral separation
(0 nm–20 nm)
2. Controller experience (experts
vs. trainees)

1. Probability
of controller
intervention


32 undergraduate psychology
students

1. Will participants find it
more difficult to remember to deviate from strong
routines, as compared
with weak routines?
2. Will ongoing task
performance decrease
when participants have to
remember to deviate from
strong routines, as compared with weak routines?

Routine strength

1. Probability of
performing a routine action instead
of an intended
action
2. Ongoing task
performance;
aircraft acceptance and conflict
detection

Authors
Fothergill
& Neal
(2008)

Participants

16 endorsed air
traffic controllers

Loft, Bolland,
& Humphreys
(2007)

Loft, Campbell,
& Remington
(2008)

Results/
Conclusions
1. When the optimal solution
was difficult to calculate,
controllers were less likely
to select the optimal solution
under high workload than
under low workload.*
2. When the optimal solution
was easy to calculate, controllers were likely to select the
optimal solution under both
levels of workload.*
3. These results can be incorporated into the development
of a computational model that
simulates how controllers resolve conflicts in the field.
1. Controllers were more
likely to intervene with increases in minimum lateral
separation.
2. Experts were more likely to

intervene than trainees.
3. A computational model
that assumes controllers place
safety margins around the
projected trajectory of aircraft
can account for both expert
and trainee intervention
decisions.

1. Participants were more
likely to forget to deviate
from strong routines, as compared with weak routines.
2. No effect of routine
strength on ongoing task
performance.

*Since

the dependent variable in this study was qualitative (solution type), categorical difference tests (McNemar tests) were used to determine
whether participants switched their conflict resolution strategy preferences under different levels of workload and as a function of the difficulty of
calculating the optimal solution.

Remington et al., 2000). Experimental control was also
required in order to systematically vary factors such as
(1) the minimum separation of aircraft pairs, (2) the angles
of intersection, and (3) the times to minimum separation.
For example, for vertical problems, one aircraft was cruising and the other climbing, with lateral separation set at
0 nm. On the basis of current speeds and climb rates, the
vertical separation distance when the aircraft violated lateral separation (
5 nm) varied from 0 ft to 4,000 ft. As

is illustrated in Figure 5, one of the key findings was that
the probability of controller intervention decreased with
increases in the minimum lateral separation of the aircraft.
Furthermore, expert controllers were significantly more
likely to intervene than were trainees. A computational
model that assumed that controllers place different safety

margins around the projected trajectory of aircraft as a
function of experience could closely predict these intervention decisions (see Table 1).
In addition to these applied research programs, ATClabAdvanced has been used when more basic research has
been conducted. Prospective memory refers to remembering to perform an action in the future and is traditionally
studied using verbal task paradigms (Einstein & McDaniel, 1990). In the real world, highly practiced tasks make
up much of the work of experts, meaning that in order
to execute intentions, people must remember to deviate
from routine (Dismukes, 2008). In addition, prospective memory demands often occur in visuospatial, rather
than verbal, contexts. Exploring prospective memory in
the context of routine visuospatial tasks is thus of both


ATC-LABADVANCED
1

Probability of Intervention

Experts
Trainees
.8

.6


.4

.2

0
0

1

2

4

6

8

10

16

14

16

18

20

Minimum Lateral Separation (nm)

Figure 5. The probability of intervention by controllers across
the minimum lateral separation of aircraft pairs.

practical and theoretical importance, and ATC-labAdvanced
provides a useful platform for conducting such investigations. Loft, Campbell, and Remington (2008) used ATClabAdvanced to investigate individuals’ ability to remember
to deviate from routine. Participants accepted aircraft into
their sector and intervened to prevent the occurrence of
conflicts by changing the flight levels of aircraft. Routine
strength was manipulated by varying the number of times
the participants performed a specific sequence of actions
when accepting aircraft. At test, prospective memory instructions asked the participants to substitute a different
key for the standard key when accepting aircraft. The participants were more likely to forget to deviate from their
strong routines (M  .17), as compared with weak ones
(M  .08). Although experimental control was required
to present standardized air traffic scenarios, the realism
of the simulation was minimized in order to allow participating first-year psychology students to quickly become
highly practiced on a small number of ATC control tasks.
ATC-labAdvanced also has the potential to be more broadly
used in basic and applied experimental research contexts.
For example, we are currently using ATC-labAdvanced to examine the motivational processes responsible for the regulation of task-directed effort, using a variety of behavioral,
physiological, and self-report measures. The simulation is
suited to the analysis of psychological phenomena at both
the within- and between-persons levels of analysis, using
both experimental and correlational methods (e.g., growth
curve modeling; Bliese & Ployhart, 2002). Other types of
phenomena that can be examined include the effects of
fatigue, alcohol, and caffeine on attention, reaction time,
and decision-making processes.
Training Manual, Data Logging,
and System Requirements

The ATC-labAdvanced simulation package includes a
training manual and practice scenarios. There are six mod-

125

ules to the training program. The amount of emphasis on
each training module will depend on the realism of the
simulation and the expertise of the participants (e.g., controllers, university students). The first module provides
a general overview of the task. The second module describes the human–machine interface, which includes the
general display, maps, and aircraft flight strips. For the
third module, participants are instructed on and practice
how to use the prediction tools. For the fourth module,
participants are instructed on and practice how to accept
and hand off aircraft, how to assign cruise or boundary
levels, and where the top of descent points are on sector
maps. The fifth module instructs participants on how to
answer questions that may appear during the experiment.
For the sixth and final training module, participants are
instructed on and practice how to intervene to modify
aircraft trajectories. The duration of the ATC-labAdvanced
training is approximately 30 min, although there is some
variance with respect to how long it takes participants to
familiarize themselves with the intervention methods and
prediction tools.
The contents of data log files recorded at the end of experimental sessions vary according to the type of experiment. Nevertheless, these files generally collect two types
of data. The first type consists of the details of the air traffic scenarios that were presented on each trial, including
the type, timings, and durations of aircraft events. Participants’ actions are the second source of data. These actions
include the timing of interventions to aircraft trajectories,
subjective ratings, timing of aircraft acceptances and
handoffs, and the timing and type of prediction tool use.

ATC-labAdvanced also records all mouse movements made
by participants in x-, y-coordinates, allowing researchers
to make inferences regarding participant attention. Log
files generated for each participant can be imported into
statistical packages such as Microsoft Excel and SPSS.
ATC-labAdvanced was written using Qt Widget Library,
owned by Troltech. Microsoft Visual C6 compiler was
used to build the source code. ATC-labAdvanced can be run
on desktops or laptop computers that run Microsoft Windows. No additional software or hardware is required. The
program will update and display each aircraft’s position,
speed, and level in the sector once every 5 sec, on the
basis of the aircraft’s current speed, average climb/descent
rates, and heading. These values are preset in a simulation
script that specifies the series of x-, y-coordinates through
which the aircraft will travel at various flight levels and
speeds. In simulations in which participants are asked to
resolve potential conflicts and assign boundary and cruise
altitudes, participants may change these parameters during a trial.
Conclusions
ATC simulators are frequently used in a variety of applied and basic research programs. Existing ATC simulators typically compromise between the extent to which
they can mimic field experience (realism) and the experimental control that they can provide. In addition, very few
ATC simulators are made publicly available to research-


126

FOTHERGILL, LOFT, AND NEAL

ers who wish to use or adapt them. The present article
has presented a new, publicly available ATC simulation

package called ATC-labAdvanced.1 The realism and experimental control provided by ATC-labAdvanced represents an
advance over many currently available simulators. In addition, ATC-labAdvanced has the programming control to
allow systematic variation of realism and control in order
to investigate specific research questions of interest in a
cost-effective manner.
AUTHOR NOTE
This research was supported in part by Linkage Grant LP0453978 from
the Australian Research Council. The authors thank Phillip Waller for his
C programming of the ATC-labAdvanced program. Thanks also go to
Peter Lindsay for his contribution to the formulae that underlie the 2-D
(lateral) dynamics of ATC-labAdvanced. Please contact Peter (p.lindsay@
uq.edu.au) for further information regarding how these formulae were
derived. Correspondence concerning this article should be addressed to
S. Fothergill, School of Psychology, University of Queensland, Brisbane
4072, QSLD, Australia (e-mail: ).
REFERENCES
Ackerman, P. L. (1992). Predicting individual differences in complex
skill acquisition: Dynamics of ability determinants. Journal of Applied Psychology, 77, 598-614.
Athenes, S., Averty, P., Puechmorel, S., Delahaye, D., & Collet, C. (2002). Complexity and controller workload: Trying to bridge
the gap. In Proceedings of the 2002 International Conference on
Human–Computer Interaction in Aeronautics (HCI-Aero 2002)
(pp. 56-60). Cambridge, MA: MIT.
Berkowitz, L., & Donnerstein, E. (1982). External validity is more
than skin deep: Some answers to criticisms of laboratory experiments.
American Psychologist, 37, 245-257.
Bliese, P. D., & Ployhart, R. E. (2002). Growth modeling using random coefficient models: Model building, testing, and illustration. Organizational Research Methods, 5, 362-387.
Bolland, S., Fothergill, S., & Humphreys, M. (2007). Modelling
the human operator: Part II. Emulating controller intervention. In
R. Jensen (Ed.), Proceedings of the 14th International Symposium on
Aviation Psychology (pp. 57-62). Dayton, OH: Association for Aviation Psychology.

Boring, E. G. (1954). The nature and history of experimental control.
American Journal of Psychology, 67, 573-589.
Brehmer, B., & Dorner, D. (1993). Experiments with computersimulated microworlds: Escaping both the narrow straights of the laboratory and the deep blue sea of the field study. Computers in Human
Behavior, 9, 171-184.
Brunswick, E. (1956). Perception and the representative design of psychological experiments. Berkeley: University of California Press.
Callantine, T. J. (2002). CATS-based air traffic controller agents
(NASA Technical Memorandum 211856). Moffett Field, CA: NASA
Ames Research Center.
Cox, M. (1994). Task analysis of selected operating positions within UK
air traffic control (Rep. No. 749). Farnborough, U.K.: DRA/Institute
of Aviation Medicine.
DiFonzo, N., Hantula, D. A., & Bordia, P. (1998). Microworlds for
experimental research: Having your (control and collection) cake, and
realism too. Behavior Research Methods, Instruments, & Computers,
30, 278-286.
Dismukes, R. K. (2008). Prospective memory in aviation and everyday
settings. In M. Kliegel, M. A. McDaniel, & G. O. Einstein (Eds.),
Prospective memory: Cognitive, neuroscience, developmental, and
applied perspectives (pp. 411-428). Mahwah, NJ: Erlbaum.
Ehret, B. D., Gray, W. D., & Kirschenbaum, S. S. (2000). Contending with complexity: Developing and using a scaled world in applied
cognitive research. Human Factors, 42, 8-23.
Einstein, G. O., & McDaniel, M. A. (1990). Normal aging and prospective memory. Journal of Experimental Psychology: Learning,
Memory, & Cognition, 16, 717-726.

Ericsson, K. A., & Williams, A. M. (2007). Capturing naturally occurring superior performance in the laboratory: Translational research on
expert performance. Journal of Experimental Psychology: Applied,
13, 115-123.
Fothergill, S., & Neal, A. (2005). Managing the airspace: A task analysis of Australian air traffic control. Australian Journal of Psychology
Supplement, 57, 109-110.
Fothergill, S., & Neal, A. (2006). Decision making in air traffic

control: How contextual factors influence conflict resolution choices.
Australian Journal of Psychology Supplement, 59, 3.
Fothergill, S., & Neal, A. (2008). An evaluation of the effect of
workload on conflict decision making in air traffic control. Australian
Journal of Psychology Supplement, 60, 4.
Gray, W. D. (2002). Simulated task environments: The role of highfidelity simulations, scaled worlds, synthetic environments, and microworlds in basic and applied cognitive research. Cognitive Science
Quarterly, 2, 205-227.
Gronlund, S. D., Ohrt, D. D., Dougherty, M. R. P., Perry, J. L., &
Manning, C. A. (1998). Role of memory in air traffic control. Journal of Experimental Psychology: Applied, 4, 263-280.
Hendy, K. C., Liao, J., & Milgram, P. (1997). Combining time and
intensity effects in assessing operator information processing load.
Human Factors, 39, 30-47.
Jeppesen (2007). TAAM solutions. Retrieved April 8, 2008, from www
.preston.net/products/TAAM.htm.
Jones, D. G., & Endsley, M. R. (2000). Overcoming representational
errors in complex environments. Human Factors, 42, 367-378.
Kopardekar, P., & Magyarits, S. (2003, June). Measurement and prediction of dynamic density. Paper presented at the 5th USA/Europe
ATM Research and Development Seminar, Budapest.
Kwantes, P. J., Neal, A., & Loft, S. (2004). Developing a formal
model of human memory in a simulated air traffic control conflict
detection task. In Proceedings of the Human Factors and Ergonomics Society 48th Annual Meeting (pp. 391-395). Santa Monica, CA:
Human Factors & Ergonomics Society.
Lamoureux, T. (1999). The influence of aircraft proximity data on the
subjective mental workload of controllers in the air traffic control
task. Ergonomics, 42, 1482-1491.
Laudeman, I., Shelden, S., Branstrom, R., & Brasil, C. (1998).
Dynamic density: An air traffic management metric (NASA-TM1988-11226). Moffett Field, CA: NASA Ames Research Center.
Loft, S., Bolland, S., & Humphreys, M. (2007). Modelling the human
air traffic controller. Expert–trainee differences in conflict detection.
In R. Jensen (Ed.), Proceedings of the 14th International Symposium

on Aviation Psychology (pp. 409-414). Dayton, OH: Association for
Aviation Psychology.
Loft, S., Campbell, L. & Remington, R. W. (2008, March). Failure to
deviate from routine and task interference in an air traffic control task.
Paper presented at the 34th Australasian Experimental Psychology
Conference, Perth, Australia.
Loft, S., Hill, A., Neal, A., Humphreys, M., & Yeo, G. (2004). ATClab: An air traffic control simulator for the laboratory. Behavior Research Methods, Instruments, & Computers, 36, 331-338.
Loft, S., Humphreys, M., & Neal, A. (2004). The influence of memory
for prior instances on performance in a conflict detection task. Journal
of Experimental Psychology: Applied, 10, 173-187.
Loft, S., Neal, A., & Humphreys, M. (2007). The development of a
general associative learning account of skill acquisition in a conflict
detection task. Journal of Experimental Psychology: Human Perception & Performance, 33, 938-959.
Loft, S., Sanderson, P., Neal, A., & Mooij, M. (2007). Modeling and
predicting mental workload in en route air traffic control: Critical
review and broader implications. Human Factors, 49, 376-399.
Metzger, U., & Parasuraman, R. (2001). The role of the air traffic
controller in future air traffic management: An empirical study of active control versus passive monitoring. Human Factors, 43, 519-528.
Mook, D. G. (1983). In defense of external validity. American Psychologist, 38, 379-387.
Rantanen, E. M., & Nunes, A. (2005). Hierarchical conflict detection
in air traffic control. International Journal of Aviation Psychology,
15, 339-362.
Raytheon (2005). FIRSTplus tower and radar simulator. Retrieved


ATC-LABADVANCED
April 8, 2008, from www.ray.ca/external/home.nsf /(Webpages)/
Products_FIRSTplus? OpenDocument.
Remington, R. W., Johnston, J. C., Ruthruff, E., Gold, M., &
Romera, M. (2000). Visual search in complex displays: Factors affecting conflict detection by air traffic controllers. Human Factors,

42, 349-366.
Rodgers, M. D., & Drechsler, G. K. (1993). Conversion of the CTA
Inc, En Route operations concepts database into a formal sentence
outline job task taxonomy (FAA Rep. DOT/FAA/AM-93/1). Washington, DC: FAA Office of Aviation Medicine.
Schiff, W., Arnone, W., & Cross, S. (1994). Driving assessment with
computer-video scenarios: More is sometimes better. Behavior Research Methods, Instruments, & Computers, 26, 192-194.
Simon, H. A. (1956). Rational choice and the structure of environments.
Psychological Review, 63, 129-138.
Späth, O., & Eyferth, K. (2001). Conflict resolution in en route traffic:
A draft concept for an assistance system compatible with solutions of
air traffic controllers. MMI-Interaktiv, 5, 1-11.
Stone, M., Dismukes, K., & Remington, R. (2001). Prospective memory in dynamic environments: Effects of load, delay, and phonological
rehearsal. Memory, 9, 165-176.

127

UFA Inc. (n.d.). The experts in air traffic control simulation: Products.
Retrieved on April 8, 2008, from www.atcoach.com/products1.html.
Yeo, G., & Neal, A. (2004). A multilevel analysis of effort, practice and
performance: Effects of ability, conscientiousness and goal orientation. Journal of Applied Psychology, 89, 231-247.
NOTE
1. Research groups interested in using ATC-labAdvanced for noncommercial purposes can download the program from www.psy.uq.edu.au/
directory/index.html?id=25. The following materials will be available
for download: the ATC-labAdvanced base code; an example XML script
based on a representative sample of the published studies; instructions on
how to use the programming control features of the XML scripts; mathematical formulae, spreadsheets, and instructions; the training modules
and instructions; and the practice scenarios. Questions regarding any of
these materials can be directed to S. Fothergill ()
at the University of Queensland.
(Manuscript received January 14, 2008;

revision accepted for publication September 26, 2008.)