TYPE Original Research

PUBLISHED 19 November 2024

DOI 10.3389/fpsyg.2024.1439401

The effects of a dual task on gaze

OPEN ACCESS behavior examined during a
simulated flight in low-time pilots
EDITED BY
Ion Juvina,
Wright State University, United States

REVIEWED BY
Thomas Sanocki, Naila Ayala 1,2*, Suzanne Kearns 2,3, Elizabeth Irving 2,4, Shi Cao 2,5
University of South Florida, United States and Ewa Niechwiej-Szwedo 1,2
Allison Jing,
University of South Australia, Australia 1
Department of Kinesiology and Health Sciences, University of Waterloo, Waterloo, ON, Canada,
*CORRESPONDENCE
2
Waterloo Institute for Sustainable Aeronautics, University of Waterloo, Waterloo, ON, Canada,
Naila Ayala
3
Department of Geography and Aviation, University of Waterloo, Waterloo, ON, Canada, 4 School of
nayala@uwaterloo.ca Optometry and Vision Science, University of Waterloo, Waterloo, ON, Canada, 5 Department of
Systems Design Engineering, University of Waterloo, Waterloo, ON, Canada
RECEIVED 27 May 2024
ACCEPTED 23 October 2024
PUBLISHED 19 November 2024
Cognitive load can impair an operator’s ability to optimally scan and process relevant
CITATION information that is critical to the safe and successful operation of an aircraft. Since
Ayala N, Kearns S, Irving E, Cao S and ​
the cognitive demands experienced by pilots fluctuate throughout a given flight
Niechwiej-Szwedo E (2024) The effects of a
dual task on gaze behavior examined during a due to changes in task demands that range from high to low cognitive load, it has
simulated flight in low-time pilots. become increasingly important to objectively track and quantify these changes
Front. Psychol. 15:1439401.
accordingly. The analysis of eye movements has been shown to be a promising
doi: 10.3389/fpsyg.2024.1439401
method to understand information acquisition, processing efficiency, and how
COPYRIGHT
© 2024 Ayala, Kearns, Irving, Cao and
these aspects of cognition impact pilot performance. Therefore, the aim of the
Niechwiej-Szwedo. This is an open-access current study was to assess the impact of a dual task paradigm on low-time pilot
article distributed under the terms of the flight performance and gaze behavior during two phases of flight with varying
Creative Commons Attribution License
(CC BY). The use, distribution or reproduction
levels of cognitive load. Twenty-two licensed pilots (<350 h) completed simulated
in other forums is permitted, provided the flight circuits alongside an auditory oddball task under visual flight rules conditions.
original author(s) and the copyright owner(s) Self-reported situation awareness scores and auditory task performance revealed
are credited and that the original publication
in this journal is cited, in accordance with
the dual task was more demanding than the single tasks. Flight performance and
accepted academic practice. No use, gaze behavior indicated that primary task performance and information processing
distribution or reproduction is permitted remained unaffected. These results suggest that the recruited pilots attained a level
which does not comply with these terms.
of skill proficiency that enabled the efficient deployment of cognitive resources
to successfully complete the flying task under states of increased cognitive load.
Combined with previous research findings, the results suggest that the effect of
secondary tasks depends on the type of tasks used (i.e., simple/choice response
tasks, memory recall, etc.). The utility of using a dual task and gaze behavior
to probe flight proficiency and information processing efficiency throughout
training are discussed.

KEYWORDS

gaze behavior, cognitive load, dual task, auditory oddball paradigm, eye movements
and visual attention, aviation

Introduction
Today’s intelligent aircraft cockpit provides multisource information via an array of
instrument panels and controls that are associated with different procedures that a pilot must
attend to at any given moment (i.e., pilot monitoring) (Federal Aviation Administration, 2021).
The capacity limitations associated with human information processing make the task of
acquiring information from these various areas of interest (AOIs) challenging, particularly
during states of high cognitive load (i.e., high amounts of cognitive resources demanded from
the operator by competing activities) (Engström et al., 2017). Therefore, it is of no surprise that
ineffective pilot monitoring was cited as being a major contributor to human factor errors

Frontiers in Psychology 01 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

during cognitively demanding phases of flight (i.e., take-off, approach contrast, cognitive tunneling is a dynamic gaze metric that identifies
and landing) (Boeing, 2021; National Transportation Safety Board, when individuals’ fall into a gaze pattern of hyper fixation toward a
1994). Evidently, a pilot’s ability to select and monitor relevant singular AOI for an extended period of time. The examination of
information in such a dynamic environment are critical to flight hyper fixation is of interest because it represents a lack of visual
success and safety. Therefore, being able to objectively track and scanning to other sources of information in the environment that may
quantify changes in information processing under varying cognitive hold pertinent information crucial to pilot SA and successful task
load conditions is an important line of inquiry as it provides a completion (Ayala et al., 2023; Ayala et al., 2024). Although more
framework to understand how operators effectively monitor task- research is required to fully understand how changes in cognitive load
critical sources of information necessary for successful task influence gaze entropy and cognitive tunneling measures, these
performance during states of high cognitive load. metrics have been shown to be useful in providing additional
Since vision plays a critical role in how we interact with our information in parallel with traditional gaze metrics and behavioral
environment, eye tracking serves as a non-invasive method that performance about the spatiotemporal properties of gaze behaviors
reveals discrete cognitive processes and strategies used to facilitate associated with information processing and goal-directed actions
skill performance that are not readily observable through overt (Ayala et al., 2022).
behaviors alone (Ayala et al., 2022; Vickers and Williams, 2017). Previous research has examined the question of cognitive load
Indeed, where we look (i.e., fixate) is governed by ongoing perceptual primarily by employing a dual task paradigm wherein the difficulty of
(i.e., bottom-up, lower-level stimulus futures) and cognitive (i.e., performing a primary task (i.e., flying) is made more challenging by
top-down, higher-level mental hierarchies and goals) processes that the addition of a simultaneous secondary task (i.e., visual search,
facilitate the selection and processing of relevant information that mental arithmetic, auditory tone, etc.); thereby, introducing additional
support skill performance (Engström et al., 2017; Hudspeth et al., rules and goals to be maintained/manipulated in working memory
2013; Land and Hayhoe, 2001). For instance, gaze (i.e., coordinated (Engström et al., 2017). As a result of the limited human brain
head-eye movements) patterns have been suggested to capacity, increased cognitive load associated with multitasking has
be predominantly driven by top-down information hierarchies (i.e., often resulted in impaired behavioral performance (Engström et al.,
scripts) rather than the ‘intrinsic salience’ of objects, particular in 2017). Indeed, numerous studies involving machine operated
populations that have experience with the task being performed (Land simulation tasks (i.e., driving, flying) have used dual-task paradigms
and Hayhoe, 2001). Traditional approaches to analyzing gaze behavior to shown that operator control, problem solving, decision making and,
have supported the claim that top-down priorities guide visual most importantly, gaze behavior are significantly impacted when the
scanning based on discrete measures of total proportion of fixation demand for cognitive resources is high (Allsop and Gray, 2014;
time (i.e., dwell time), fixation frequency or average fixation durations Engström et al., 2017; Kanaan and Moacdieh, 2022; Van de Merwe
toward specific areas of interest (AOIs). Specifically, these traditional et al., 2012; Yang et al., 2018). Specifically, high cognitive load has been
metrics demonstrate what sources of information are crucial to the shown to be associated with a narrowing of attention toward fewer
task at hand based on where gaze is allocated for larger proportions of areas of interest (AOIs), reduced SGE, more frequent fixations and
total dwell time as well as higher fixation frequencies (Brams et al., longer fixation durations (Allsop and Gray, 2014; Di Nocera et al.,
2018; Land and Hayhoe, 2001). Average fixation duration can also 2007; Engström et al., 2017; Kanaan and Moacdieh, 2022; Lu et al.,
serve as an indicator of information processing efficiency, where 2020; Lutnyk et al., 2023; Reimer, 2009; Wright et al., 2014). These
longer durations are associated with more time being required to changes in gaze behavior were reported in parallel with increased
process the information being fixated and shorter durations are misjudgment (i.e., spatial disorientation, increased time and error
associated with less time processing objects of interest (i.e., typically rates associated with hazard object/event detection), reduced aircraft
associated with quickly verifying information or highly efficient control (i.e., increased variability in aircraft speed, pitch and turn
scanning) (Brams et al., 2018; Land and Hayhoe, 2001). Other coordinator), and unstable approaches and landings (Crosby and
computationally complex measures of gaze behavior have also been Parkinson, 1979; Engström et al., 2017; Ziv, 2016). Collectively, these
applied to identify global patterns of behavior that emerge over the findings suggest that increases in cognitive load due to multitasking
course of task performance such as stationary gaze entropy (SGE), are associated with increased processing time toward fewer sources of
gaze transition entropy (GTE), and cognitive tunneling (i.e., the information because of increasing demands placed on cognitive
frequency and duration of poor internal cockpit monitoring events) resources, which increases the risk of human error and unsafe
(Ayala et al., 2024). SGE and GTE have emerged more recently in the machine operation.
literature to provide an objective indices of gaze dispersion and A more extensive review of dual-task paradigm literature revealed
fixation sequence complexity, respectively (Shiferaw et al., 2019). that the effects of cognitive load on complex tasks like driving a car are
While SGE values are modulated by the extent of fixation dispersion strongly selective and task dependent (Engström et al., 2017). For
(i.e., low entropy = concentrated fixation distribution across very few instance, several driving studies demonstrated no evidence for
AOIs; high entropy = dispersed fixation distribution across many increased crash/near-crash risk associated with primarily cognitive
AOIs), GTE is more reflective of visual attention deployment. An tasks, such as talking on the phone or using a radio (Simmons et al.,
exploratory (i.e., salience-based) mode of attentional deployment is 2016). In accounting for these apparent inconsistent and
associated with higher GTE (i.e., less attentional bias in fixating several counterintuitive findings regarding the effects of cognitive load on
areas that are relevant and/or irrelevant to the task at hand), while a driving performance, the Cognitive Control Hypothesis was proposed.
reduction in GTE is reflective of a focal mode of visual attention (i.e., The hypothesis states that cognitive load will selectively impair
structured, and biased attention toward fewer AOIs), which may performance on tasks that require controlled processing (i.e., novel,
be more indicative of task-engagement (Shiferaw et al., 2019). In uncertain, nonroutine tasks that require top-down executive functions

Frontiers in Psychology 02 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

like working memory), while leaving tasks that require efficient increased cognitive load will be associated with a reduction in
processing (i.e., practiced, effortless, generally unconscious tasks that attention allocation across a smaller set of task-relevant AOIs
require little to no cognitive control oversight) (also known as manifested as an increase in dwell time and fixation durations toward
automatic processing) relatively unchanged (Engström et al., 2017). the external environment (i.e., front window) (Allsop and Gray, 2014;
Notably, the transition of a learned skill from controlled processing to Allsop et al., 2017; Kanaan and Moacdieh, 2022; Yang et al., 2018; Yang
efficient processing represents a progression from low to high et al., 2022). Although more dynamic measures of gaze behavior have
efficiency in the cognitive processes needed to plan, execute, and only been examined in the context of increasing visuomotor task
update skilled movements (Immink et al., 2020). Since the majority of difficulty (i.e., wind manipulations) (Ayala et al., 2023), it was expected
pilot monitoring errors occur during cognitively demanding phases that an increase in cognitive load would similarly be associated with
of fight (i.e., take-off, approach and landing) (Boeing, 2021; National a reduction in the dispersion of fixations (SGE), a reduction in visual
Transportation Safety Board, 1994), the Cognitive Control Hypothesis scan pattern complexity (GTE), and an increase in cognitive tunneling.
suggests that these phases of flight are associated with controlled These gaze behavior changes are expected to be specifically evident
performance (i.e., require cognitive resources to manage the respective during the approach and landing rather than the cruise phase due to
sub-tasks, while maintaining aircraft control) and are susceptible to the increased cognitive load induced by processing task-specific
increases in cognitive load, more so than any other flight phase (i.e., information to ensure safe landing, which may overlap with the
cruise). Accordingly, a dual-task paradigm could serve as a useful processing demands of the secondary auditory task.
method to understand how changes in cognitive load impact
information processing via changes in gaze behavior across the
various phases of flight. Methods
This paper aimed to build on previous aviation-related work
employing a dual task paradigm to examine the utility of gaze metrics Participants
in assessing the impact of cognitive load on pilot’s performance and
information processing using a more ecological approach. There are Twenty-two participants (14 males, 8 females; age range
two major aspects through which this study differs from previous 18–24 years, mean = 21 years old, SD = 2) were recruited from the
work. First, the current work sought to use a dual task paradigm to student populations at the University of Waterloo. All participants
objectively probe the impact of a secondary task on cognitive load were current aviation students or individuals who had obtained at
across two phases of a flight circuit (i.e., cruise [low cognitive load], least their private pilot’s license (PPL) (number of flight hours range:
approach and landing [high cognitive load]). Since it has been well 48–340 h, mean = 199 h, SD = 73; PPL = 16; CPL = 6). All participants
established that the two stages of flight (cruise, approach and landing) had normal or corrected-to-normal vision and had not been
vary in the extent to which they require cognitive resources to monitor previously diagnosed with a neuropsychiatric/neurological disorder
and complete the required sub-tasks (Di Nocera et al., 2007; Federal or learning disability. Participation was voluntary, and participants
Aviation Administration, 2021; Wilson, 2002), it is important to received $25/h as remuneration. The University of Waterloo Research
examine how the imposed change in cognitive load impacts the phase- Ethics Board Committee (#43564) approved the study protocol, which
specific spatiotemporal aspects of gaze behavior. Crucially, accurately was performed in accordance with the 2008 Declaration of Helsinki.
monitoring pilot’s cognitive load through continuous, objective, gaze- Consent was obtained prior to beginning the protocol. Note that one
based metrics is of great significance for the development of real-time participant was excluded from analysis as a result of corrupt data files.
metrics to assess pilot’s information processing capability and may
also be an effective method to track pilot’s progression in the
development of efficient information processing throughout training. Experimental setup and apparatus
Second, the current work was designed to have high fidelity to increase
the generalizability of the findings to low-time pilots. Specifically, the Flight simulator
dual task involved monitoring auditory inputs to simulate the An AL250 ALSIM flight simulator (ALSIM, France) configured as
modality of Air Traffic Control [ATC] calls, while simulating the flight a generic single engine aircraft that is representative of a Cessna 172
environment with a fully immersive flight simulator. This is distinct was used with the necessary instrument panel (steam gauge
from previous work that used video recordings for the primary task configuration), an avionics/GPS system, an audio/lights panel, a
(van de Merwe et al., 2012), introduced secondary tasks that did not breaker panel, and a Flight Control Unit (FCU) (see Figure 1). The
resemble general aviation-like procedures (i.e., mental arithmetic, field of view covered by the simulator was 250° by 49° via panoramic
working memory tasks, visual search, military specific shooting tasks, VFR-VR-HD projectors. The participants sat in a height-adjustable
etc.) (Abel, 2009; Allsop et al., 2017; Babu et al., 2019), and recruited seat and controlled the aircraft with a yoke, throttle lever, and rudder
participants with no aviation experience (Allsop and Gray, 2014; pedals. Stimuli presentation, and behavioral data collection and
Lutnyk et al., 2023). acquisition were controlled from the Instructor Station and
The current study used a secondary task as a probe to evaluate the Engineering pack (ALSIM, France).
impact of cognitive load on flight performance and gaze behavior
across two phases of flight: cruise and landing. It was hypothesized Eye tracker
that performance on the auditory oddball will be significantly less MindLink eye-tracking glasses (AdHawk Microsystems Inc.,
accurate and slower during the approach and landing phase of flight Waterloo, ON, Canada) were used to track the participants’ gaze (i.e.,
compared to the cruise phase. Moreover, based on previous work eye and head) movements. MindLink is a non-camera-based eye
using a VFR (Visual Flight Rules) paradigm, it was hypothesized that tracker embedded in a frame of eyeglasses that uses an ultra-compact

Frontiers in Psychology 03 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 1
Illustration of the visual stimuli employed in the AL250 flight simulator environment. The participants point of view of the cockpit replicated that of a
pilot flying a Cessna 172, pre-set for a cruise and landing approach to Waterloo International Airport, Breslau, Ontario, CA. The orange boxes represent
the ten main areas of interest used in the gaze analyses. These include the airspeed (1), attitude (2), altimeter (3), turn coordinator (4), heading (5),
vertical speed (6) and power (7) indicators, as well as the front (8), left (9), and right windows (10).

micro-electromechanical system (MEMS) to track the eye and gaze For the flight task, all participants received the exact same
movements (Zafar et al., 2023). The eye tracker was operating at environmental configurations in Visual Flight Rules (VFR) conditions
250 Hz, transmitting the gaze data and the video of its front-facing (i.e., high visibility [>20 miles] and calm [0 kts, 0°]) with a pre-set start
camera (82° field of view, 1080p, 30 Hz) via the AdHawk eye tracking at the beginning of the cruise stage of the flight circuit to the airport
software to a laptop (60 Hz refresh rate, 1920 × 1,080 pixels, Microsoft at an altitude of 2017 ft., with flaps and trim set to zero, and a starting
11) visible only to the experimenter. speed and RPM of 110 kts and 2000 rpm, respectively. The flying task
was segmented into two cognitive load phases. Cruise was the first
Scenario and task phase of flight wherein the goal was to maintain aircraft control (i.e.,
Participants were tested in a single experimental session aircraft altitude [2017 ft], airspeed [100–110 kts], and vertical speed
(~90 min) which started with a visual screening that included a visual [0 fpm]) by monitoring the gauges for any slight deviations in speed
acuity test (Bailey-Lovie chart) and a stereoacuity test (Randot Stereo or altitude and adjusting the flight controls as necessary. The second
test, Stereo Optical Company Inc.). A pilot briefing (led by an phase consisted of landing the plane wherein the goal was to land the
instructor pilot) and practice trial were completed to familiarize the plane as smoothly and accurately as possible relative to the center of
participants with the simulator environment (Al 250, ALSIM, France), the 500 ft. markers near the start of runway 26. Note that this phase
flight path, and flight parameters (cruise airspeed [110 kts] and consists of several aircraft configuration changes that required specific,
altitude [2017 ft], landing approach airspeed [65 kts] and touchdown time-sensitive maneuvers that allowed the plane to descend and
reference point [center of 500-foot marker]). The experimental become aligned with the runway in a safe and stable manner. The trial
scenarios were programmed in the flight simulator environment, was terminated after the participant brought the plane to a complete
flying into the Region of Waterloo International Airport (CYKF; stop, or if the landing was deemed unsuccessful (i.e., plane crash or
Runway 26), Breslau, Ontario, Canada. Participants were then asked plane landed off the runway). Note that pilots were required to
to complete a total of 9 customized trials while their gaze movements communicate with Air Traffic Control (ATC) during both phases of
were recorded. The trials were pseudo-randomized (i.e., ABC, BCA, flight, as they normally would in a real aircraft. Figure 2 illustrates the
CBA, BAC) into three types of scenarios: (1) single-task auditory experimental overview including the three task conditions, and the
oddball task (control condition), (2) single-task flying circuit (cruise simulated scenario along with the flight path and circuit parameters.
[low cognitive load] and landing [high cognitive load]), and (3) dual Auditory oddball task (control task) used auditory tones that
task paradigm (single task circuit with auditory oddball task). Prior were created using VPixx 3.2.1 software and were presented
to beginning the experimental trials, a 9-point eye-tracking biaurally at 80 decibels using Apple computer speakers. Participants
calibration and validation procedures were completed by the remained in the left pilot seat of the cockpit with the aircraft parked
examiner (average gaze error < 2°). on the runway. Before testing, participants were presented with an

Frontiers in Psychology 04 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 2
Illustration of the simulated flight path and circuit parameters for cruise (easy [blue]) and landing (difficult[pink]) phases of flight (A). Summary of the
task conditions, scenario design, and the dependent variables assessed for each type of trial (B). Three trials of each condition were completed for a
total of 9 trials, which were pseudorandomized. Single task audio trials assessed response time and accuracy in distinguishing target (30%) tones
among the more commonly presented non-target (70%) tones. The exact timing of auditory stimuli presentation varied across and within trials with an
inter-stimulus interval range of 1900 msec to 2,100 msec. The single task flight trials assessed flight performance and gaze behavior as participants flew
the cruise and landing scenario. The dual task condition involved participants completing the flight and auditory tasks simultaneously. The dual task
trials assessed auditory and flight performance as well as gaze behavior. Each trial lasted approximately 10 min for a total session duration of 90 min.
Each trial was also followed by the Situational Awareness Rating Technique (SART) questionnaire (i.e., clipboard represents the online questionnaire
participants completed).

iteration of high (1,000 Hz) and low (375 Hz) tones and asked if condition; therefore, if the task becomes too demanding you should
they could discriminate between the two tones. Participants were focus on flying the aircraft more than attending to the auditory task.”
then instructed to respond by pressing on the pilot push-to-talk After each experimental trial (i.e., single flying circuit, single
button located on the yoke with their left hand as quickly as possible auditory oddball task, dual task paradigm) the participant was then
when hearing a high “target” tone (probability = 30%), and to asked to complete the Situational Awareness Rating Technique
withhold a response when hearing a low “non-target” tone (SART) questionnaire to gauge their subjective opinion on various
(probability = 70%). Participants were notified when a trial began domains related to task difficulty, and the supply and demand of
by the instructor followed by an iteration of high and low tones attentional resources required during task performance (Taylor and
randomly presented at inter-stimulus intervals ranging from Selcon, 1990).
1900 msec to 2,100 msec.
The dual task paradigm condition involved performing both the
flying circuit and auditory oddball tasks simultaneously. Participants Data processing and analysis
were informed by the instructor when the trial began. The auditory
tones were presented and responded to just as they were during the Based on the various flight performance measures collected,
auditory only condition. Every participant was given the same analysis was split across the two phases of flight: (1) Cruise
instructions to “primarily focus on flying the aircraft in the dual task performance- includes airspeed and vertical speed mean and

Frontiers in Psychology 05 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

variability- and (2) Landing performance- includes completion time work, fixations found outside these AOIs were defined as non-areas of
(sec; duration of time from the start of the flying scenario to the plane interest and were excluded from analysis (<5%). Trials with missing
coming to a complete stop on the runway), landing accuracy (degrees; data (i.e., loss of signal >30% found in <8% of trials) and outliers from
the difference between the center of the plane and the center of the each of the dependent variables (i.e., >1.5 the interquartile range
500 ft. runway marker at point of touchdown). Notably, the distinct around the first and third quartile found in <1% of trials) were removed.
cruise and landing performance metrics were not continuous across Gaze behavior was examined across each phase of flight (cruise
phases. For instance, vertical speed would only be an appropriate [low load], landing [high load]) using traditional and advanced (i.e.,
measure for landing performance during the final phase of landing but static and dynamic) gaze measures (Ayala et al., 2022; Ayala et al.,
not necessarily all other parts of the landing phase (i.e., downwind, 2023; Ayala et al., 2024; Glaholt, 2014). Traditional gaze-based analysis
base). Additionally, the required mean airspeed was expected to included the calculation of dwell time (%; total dwell duration spent
constantly decrease throughout the landing phase whereas during within a given AOI as a function of total flight time) and average dwell
cruise, there is an expected consistent mean airspeed that participants duration (msec; average duration of uninterrupted dwells within a
are expected to maintain. All flight metrics were examined within given AOI). Static entropy-based analyses were completed using the
their respective phase of flight and subjected to a repeated measures ten AOIs outlined in Figure 1 (Ayala et al., 2023). Custom scripts
ANOVA with a single within-subject factor (task condition: single, (Ayala et al., 2023) were used to compute both SGE (Equation 2;
dual). Participants were instructed to prioritize the primary flight task Shannon, 1984) and GTE (Equation 3; Ciuperca and Girardin, 2007),
when the secondary task became too challenging to manage. As such, which were then normalized (Equation 4; Shannon, 1984).
it was expected that flight performance would remain similar across Specifically, SGE was computed by first producing a vector, V, of
the single flight task and dual task conditions for both phases of flight length 10 (for each AOI), where Vi was the total number of fixations
(i.e., cruise, landing). at AOI i. V was then divided by the total number of fixations in the
Auditory oddball task performance was assessed using auditory sequence, so that Vi was the probability of a fixation landing at AOI i.
response time (msec; response time to accurate target tones) and The probability vector V was then applied to Equation 2.
accuracy (percent; percentage of correct responses to total number of
tones presented). Auditory oddball performance was examined during H SGE (V ) = − ∑ v·log ( v ) (2)
single auditory task conditions, as well as the dual task condition. v∈V

Since the dual task condition was split into a cruise (low load) and
landing (high load) phase of flight, raw auditory response data were GTE was computed by first creating a 10×10 (AOI) transition
subjected to a repeated measures ANOVA (within-group factor: matrix, M, where Mi,j was the total number of transitions from AOI
cognitive load [single auditory task, dual cruise, dual landing]). To i to AOI j. Each row, Mi,∗, was divided by the sum of row i, so that Mi,∗
specifically examine the impact the dual task had on auditory represented the probability of fixation transition from AOI i to any of
responses during the different flight phases, the dual task ‘cost’ (DTC) the ten AOIs. Finally, GTE was computed using Equation 3 (Ciuperca
(Equation 1) for auditory response metrics were calculated by and Girardin, 2007), applying the transition matrix M and the
expressing participants’ changes in the respective measures during probability vector V.
dual task performance relative to their control task (i.e., auditory only)
10 10
performance according to the formula (Zhou et al., 2014; Manor et al.,
H GTE ( M ) = −∑Vi ∑M i , j ·log ( M i , j ) (3)
2010; Schwenk et al., 2010): =i 1 =j 1

 ( dual task performance − control task performance )

DTC   × 100 (1)
 control task performance  H NORMAL = H / H MAX  (4)

Dual task cost comparisons for auditory response were conducted The dynamic entropy-based analysis was completed using the 10 s
with a repeated measures ANOVA (within-group factor: phase of average sliding window developed by Ayala et al. (2023). This cognitive
flight [cruise, landing]). Previous work has shown that DTC increases tunneling analysis was used to examine the reduction in gaze transitions
with increasing cognitive load. Therefore, DTC for response time and from the external environment to the internal cockpit environment.
accuracy is expected to increase significantly more in the landing This analysis specifically quantified the number of cognitive tunneling
phase of flight compared to the cruise phase of flight. bouts (i.e., number of instances in which a dwell remained entirely
Gaze data were post-processed offline using a custom-made script outside of the cockpit for at least 10 s), bout duration (sec; the average
that used the 3D gaze vectors provided by AdHawk software for duration of all bouts that occurred in a trial), and total bout time (sec;
saccade and fixation detection, similar to Ayala et al. (2024). the sum of all individual cognitive tunneling bout durations within a
Eye-movement traces were visualized by the experimenter and played trial). Gaze behavior metrics were collected during both phases of flight
back at a slowed speed superimposed over the video displaying the (cruise, landing) and under single flight task and dual task conditions.
simulator environment. The task environment was discretized using a Therefore, these data were subjected to a 2×2 repeated measures
custom code that partitioned the simulator environment into ten areas ANOVA (within-subject factors: Task Condition [single, dual], Phase
of interest (AOIs) (Figure 1). The AOIs were manually defined to of Flight [cruise, landing]) to determine the impact of cognitive load on
represent airspeed (1), attitude (2), altimeter (3), turn coordinator (4), gaze behavior. If cognitive load influences gaze behavior during aircraft
heading (5), vertical speed (6), and power (7) cockpit indicators, along operation, it is expected to be demonstrated by a reduction in SGE,
with the front (8), left (9), and right (10) windows. In line with previous GTE, an increase in cognitive tunneling and a specific increase in dwell

Frontiers in Psychology 06 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

TABLE 1 List of dependent measures and their associated ANOVAs and TABLE 2 Flight performance values for single and dual task conditions.
independent variables.
Dependent variable Single Dual
Dependent ANOVA Independent
Completion time (sec) 494.1 (22.2) 491.2 (25.35)
variable variables
Landing accuracy (°) 0.055 (0.059) 0.054 (0.049)
Flight performance Task Condition (single
metrics 1×2 flying, dual task) Landing hardness (fpm) −95.56 (46.3)** −117.1 (51.35)**

Cognitive Load (single Cruise airspeed (kts) 104.4 (0.85) 104.6 (0.83)
Raw auditory auditory, dual cruise, dual Cruise airspeed variability 0.50 (0.44) 0.46 (0.37)
performance metrics 1×3 landing)
Cruise vertical speed (fpm) −3.77 (15.53) −3.14 (14.64)
Dual-task cost auditory Phase of Flight (cruise,
Cruise vertical speed
scores 1×2 landing)
variability 2.21 (1.90) 2.92 (2.16)
Task Condition (single
Mean (standard deviation) flight performance values are reported for single and dual task
flying, dual task) conditions. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.
Phase of Flight (cruise,
Gaze behavior metrics 2×2 landing)
Cruise aircraft performance results demonstrated no significant
Task Condition (single differences between the single and dual task for cruise airspeed mean
SART scores 1×2 flying, dual task) and variability, F(1, 20) = 1.594 and 0.091, p = 0.224 and 0.767,
The listed dependent variables represent the larger clusters of actual dependent variables. 𝜂p2 = 0.087 and 0.005, or cruise vertical speed mean and variability, F(1,
Flight performance metrics include landing performance metrics (total completion time, 20) = 0.613 and 1.407, p = 0.444 and 0.252, 𝜂p2 = 0.035 and 0.076,
landing error, landing hardness), as well as cruise performance metrics (airspeed and vertical
speed mean and variability). Raw auditory performance metrics include response time and
respectively.
response accuracy. Dual task cost auditory scores include auditory response cost and Landing performance results demonstrated no significant
auditory accuracy cost. Gaze behavior metrics include dwell time, fixation duration, SGE, differences between the single and dual task for completion time, F(1,
GTE, cognitive tunneling bout number, duration, and total time. SART scores include overall
SA, SA supply, SA demand, and SA understanding scores.
20) = 1.308, p = 0.266, 𝜂p2 = 0.061, and landing accuracy, F(1,
20) = 0.126, p = 0.727, 𝜂p2 = 0.006. However, landing hardness was
significantly higher (i.e., more negative) in the dual task, F(1,
time toward fewer, task-relevant AOIs as cognitive load increases. These 20) = 8.562, p = 0.008, 𝜂p2 = 0.300. All means and standard deviations
findings are expected to be more significant in the landing phase of for cruise and landing flight performance measures are reported in
flight compared to the cruise phase of flight. Table 2.
Situation awareness (SA) was assessed using a subjective
questionnaire. The SART questionnaire (Taylor and Selcon, 1990) is a
post-trial self-report questionnaire that uses a 7-point Likert scale
(1 = Low; 7 = High) across 10 dimensions of SA. This is collapsed into Auditory task performance
three larger dimensions of attentional demands, attentional supply
and situation understanding. These ratings are then combined to Raw auditory performance scores were subjected to a repeated
calculate an overall measure of SA. measures ANOVA with a single within-subject factor (cognitive load:
single, dual cruise, dual landing). Dual task cost scores were subjected
SA Understanding − ( Demand − Supply )
= (5) to a repeated measures ANOVA with phase of flight (cruise, landing)
as the only within-subject factor.
Auditory oddball task performance results demonstrated a
SART scores were post-trial surveys that were completed at the significant main effect of cognitive load for response time and
end of a trial and thus could not be partitioned into cruise and landing response accuracy, F(2, 40) = 30.914 and 23.648, p < 0.001 and < 0.001,
flight phases. As such, SART scores were subjected to a repeated 𝜂p2 = 0.607 and 0.542, respectively. Specifically, response time increased
measures ANOVA (within-group factor: task condition [single, dual]). as a function of cognitive load (single = 752.17 msec, SD = 393.16; dual
All ANOVAs were performed with an alpha level set at 0.05. The cruise = 1079.26 msec, SD = 487.44; dual landing = 1450.71 msec,
Bonferroni post hoc correction for multiple comparisons was also SD = 574.79). Response accuracy decreased as a function of cognitive
applied for all post hoc analyses following the repeated measure load (single = 99.6%, SD = 0.8; dual cruise = 98.3%, SD = 2.9, dual
ANOVAs to determine significant differences between variables. landing = 90.9%, SD = 7.5).
Table 1 provides a summary of all the ANOVAs conducted and their Figures 3A,B illustrate the dual task costs associated with auditory
respective independent variables. response time and response accuracy, which demonstrated a main
effect of phase of flight, F(1,20) = 6.106 and 23.039, p = 0.027
and < 0.001, 𝜂p2 = 0.234 and 0.535, respectively. Specifically, response
Results time cost increased significantly from cruise (mean = 71.2%,
SD = 102.7) to landing (mean = 166.3%, SD = 205.3). This is consistent
Flight task performance with the response accuracy cost (i.e., this is reflected as the inverse of
the actual accuracy difference to express increased ‘cost’ to
Flight performance measures were subjected to a repeated measures performance), which is also shown to increase significantly from
ANOVA (within-group factor: task condition [single, dual task]). cruise (mean = 1.3%, SD = 3.1) to landing (mean = 8.7%, SD = 7.6).

Frontiers in Psychology 07 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 3
Individual data points and their respective group means for auditory response time cost (%) (A) and response error cost (%) are demonstrated for cruise
and landing phases of flight. (B) Error bars represent SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

TABLE 3 Dwell time (%) descriptive and group statistics. toward the turn coordinator demonstrated an interaction between
task and flight phase, F(1,20) = 6.689, p = 0.018, 𝜂p2 = 0.251. Post-hoc
Area of Cruise Landing F P Eta
interest statistic value squared decomposition of this interaction demonstrated that dwell time
(AOI) toward the turn coordinator was significantly higher for the cruise
phase during the flying only (i.e., single) task (mean = 8.58%,
6.77
SD = 4.34) compared to the dual task (mean = 6.87%, SD = 2.86), while
Airspeed (5.45) 16.43 (7.06) 77.293 <0.001 0.794
this remained relatively consistent across tasks during the landing
12.72
flight phase (single: mean = 6.61%, SD = 3.47; dual: mean = 6.29%,
Altimeter (6.14) 6.12 (2.47) 44.94 <0.001 0.692
SD = 3.01). No other effects of flight phase, task condition, or
4.99 interactions between the two reached statistical significance
Heading (4.39) 2.52 (1.48) 9.547 0.006 0.323 (p > 0.05). The lack of main effect or interaction involving task
Front 40.88 condition for dwell time are illustrated in Figure 4A. As such the
window (15.31) 35.37 (7.47) 6.241 0.021 0.238 mean (SD) and group statistics were collapsed across task conditions
Left 4.61
and reported in Table 3 for each phase of flight.
window (2.38) 16.42 (5.59) 119.21 <0.001 0.856
Results for average dwell duration are shown in Table 4. There
was a main effect of flight phase across several AOIs. This included
Mean (standard deviation) dwell time (%) values across significant areas of interest (AOI) are
reported across cruise and landing phases of flight, which were collapsed across single and significant increases in fixation duration across airspeed, turn
dual task conditions since no main effect of task condition was reported. Statistical test coordinator, and front window AOIs, along with a significant
values (F, p, 𝜂p2) for main effects of phase of flight (i.e., cruise, landing) are reported for only decrease in fixation duration toward the altimeter AOI. Airspeed also
significant AOIs. Df (1,20).
demonstrated a main effect of task, F(1,20) = 5.983, p = 0.024,
𝜂p2 = 0.024, which specifically demonstrated a reduction in fixation
duration during the dual task (mean = 480.72 msec, SD = 73.42)
Gaze behavior compared to the flying only condition (mean = 500.67 msec,
SD = 87.39). All other AOIs were not significantly modulated by task
Gaze metrics were assessed via a 2×2 repeated measures ANOVA condition or phase of flight (p > 0.05). The lack of main effect or
(within-group factors: task condition [single, dual task], phase of interaction involving task condition for dwell duration are illustrated
flight [cruise, landing]). Dwell time (%) demonstrated a main effect in Figure 4B. As such the mean (SD) and group statistics were
of flight phase across several AOIs (Table 3). This included significant collapsed across task conditions and reported in Table 4 for each
increases in dwell time toward airspeed and left window AOIs, as well phase of flight.
as significant reductions in dwell time across altimeter, heading, and SGE (bits) showed a main effect of flight phase, F(1,20) = 5.311,
front window AOIs from the cruise phase to the landing phase of p = 0.0332, 𝜂p2 = 0.210, demonstrating that fixation dispersion was
flight. All dwell time means, standard deviations (SD), and statistical significantly greater during landing (mean = 2.74 bits, SD = 0.16)
test values for the main effect of phase of flight are presented in compared to the cruise phase (mean = 2.66 bits, SD = 0.24) (Figure 5A).
Table 3. Notably, dwell time toward the power gauge demonstrated a However, SGE and GTE were not significantly impacted by task, and
main effect of task, F(1,20) = 7.417, p = 0.013, 𝜂p2 = 0.271. Specifically, there was no interaction with flight phase (p > 0.05). Similarly, there
dwell time was higher during the dual task (mean = 6.67%, SD = 0.73) were no significant main effects of task, flight phase or interaction for
compared to the single task (mean = 5.41%, SD = 0.49). Dwell time cognitive tunneling bout metrics (p > 0.05) (Figure 5).

Frontiers in Psychology 08 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 4
Group means for dwell time difference (%) (A) and dwell duration (msec) difference (B) (Δ difference: dual-single) across all areas of interest (AOIs)
during cruise (open circle) and landing (open triangle) phases of flight. Error bars represent SEM. Significant interactions between task and flight phase
are indicated accordingly *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

TABLE 4 Average dwell duration (msec) descriptive and group statistics No significant differences were reported for task understanding
(SEM). (p = 0.054).
Area of Cruise Landing F p Eta
interest statistic value squared
(AOI) Discussion
428.88 552.51
Airspeed (85.73) (107.18) 24.657 <0.001 0.552 Pilot monitoring behavior is a critical aspect of safe aircraft
operation. Understanding where and when pilots allocate their
502.67 457.99
attention are important aspects of information processing. Specifically,
Altimeter (101.32) (104.67) 4.985 0.037 0.2
gaze analysis could help to understand a pilot’s ability to use task-
Turn 431.69 496.93
relevant information in the environment to assess, plan and generate
coordinator (146.78) (146.77) 5.88 0.025 0.227
the necessary sequence of actions required at any given moment. The
Front 657.22 888.88 current study examined the effect of a dual task on flight performance
window (172.90) (284.48) 12.572 0.002 0.386 and gaze behavior during two phases of flight with differing levels of
Mean (standard deviation) average dwell duration (msec) values across significant areas of cognitive load. Participants reported the dual task was more
interest (AOI) are reported across cruise and landing phases of flight. Statistical test values challenging as indicated by the SART questionnaire. Notably, the
(F, p, 𝜂p2) for main effects of phase of flight (i.e., cruise, landing) are reported for only
significant AOIs. Df (1,20).
auditory task revealed that the landing phase imposed a higher
cognitive load compared to the cruise phase as demonstrated by
reduced accuracy and longer response times to auditory tones. In
Subjective situation awareness ratings contrast to our hypotheses, flight performance and most of the gaze
behavior indicators were not significantly affected by the dual task
SART scores were examined with a repeated measures ANOVA suggesting that the type of secondary task used and the way it is
with a single within-subject factor (task condition: single, dual). introduced plays a role in the impact it has on the primary task.
Participant self-ratings of situation awareness demonstrated a main Therefore, this finding suggests that the pilot cohort tested in this
effect of task, F(1,20) = 31.199, p < 0.001, 𝜂p2 = 0.609. Participants study attained a level of proficiency that allowed them to effectively
reported higher subjective situation awareness in the single flying deploy their cognitive resources to complete the flying task across all
condition (mean = 21.5, SD = 5.0) than in the dual task (mean = 18.0, task manipulations. These findings are discussed in the context of
SD = 4.6) (Figure 6). The breakdown of situation awareness scoring evaluating gaze behavior as a marker of information
also demonstrated main effects of task for attentional demand and processing efficiency.
attentional supply, F(1,20) = 49.021 and 7.827, p < 0.001 and 0.012, Although SA and auditory task performance confirmed that the
𝜂p2 = 0.710 and 0.281, respectively (Figure 6). Specifically, dual task increased cognitive load as hypothesized, responding to the
attentional demand demonstrated a significant increase from the auditory tones had very small effects on flight performance and gaze
single flying condition (mean = 6.30, SD = 3.17) to the dual task behavior metrics in both the cruise and landing flight phases. The
(mean = 10.73, SD = 2.95). This was seen in parallel with a flight performance results confirm that participants followed
significant increase in attentional supply from the single instructions to make flying the aircraft the primary task but indicate
(mean = 17.69, SD = 4.03) to the dual task (mean = 19.28, SD = 2.69). that the secondary task was not challenging enough to impair

Frontiers in Psychology 09 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 5
Individual data points and their respective group means for stationary gaze entropy (SGE) (A), gaze transition entropy (GTE) (B), number of cognitive
tunneling bouts (C), and total cognitive tunneling bout time (sec) (D) are demonstrated for both task conditions (single, dual) across cruise and landing
phases of flight. Error bars represent SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

performance on the primary task, as has been demonstrated in other task conditions, so it is not a matter of reducing dwell duration time
literature employing a dual-task paradigm (Allsop and Gray, 2014; to match the reduced time taken to complete the task. Second, dwell
Babu et al., 2019; Engström et al., 2017; Lutnyk et al., 2023; Yang et al., time was significantly higher toward the power gauge during the dual
2018; Yang et al., 2022). The only significant difference in flight task condition. This increase in power dwell time was reported
performance was an increase in landing hardness during the dual task. alongside an interaction effect for the turn coordinator gauge.
However, the approximately 22 fpm increase in landing hardness is not Specifically, it was demonstrated that dwell time was highest for the
perceptible, nor did it deem the landings unsafe given that the aircraft turn coordinator during the single task cruise flight phase. Therefore,
in question can handle landings with a landing hardness of 700 fpm. it seems as though the dual task condition caused a shifting of
With respect to the gaze behavior findings, a few significant attention from the turn coordinator to the power gauge. This shift in
differences were reported between single and dual tasks. First, average gaze, however, is only shown to be true during cruise, which is an
dwell durations were shown to decrease significantly (~20 msec) interesting finding considering that the turn coordinator is primarily
during the dual task, specifically for the airspeed gauge. According to used as a visual reference to gauge the rate of turn (or lack of turn
previous work, this finding suggests that participants required less during cruise) whereas the power gauge is primarily used to get a
time to process airspeed AOI information in this condition (Ayala sense of engine power and fuel efficiency (Federal Aviation
et al., 2022; Brams et al., 2018; Glaholt, 2014). This may be taken as Administration, 2021). As a result of no significant reductions in dwell
evidence suggesting a change in gaze strategies to optimize the time shown across any other AOIs because of task condition, it is likely
scanning of gauges (i.e., completing quick checks for task relevant that it came from a number of AOIs, which collectively resulted in an
AOIs only to ensure alignment with expected values) while allocating increase in dwell time toward the power gauge during approach and
attentional resources to the secondary auditory task. However, this landing. In any case, since these changes in dwell time were quite small
was only shown across a single AOI, which does not provide sufficient (~1–1.5%) and they were not accompanied by a significant reduction
evidence that information processing became more efficient in general in fixation distribution (SGE), there is no evidence to suggest that
during the dual task condition. Additionally, the total time taken to these changes were associated with a narrowing of attention toward
complete these scenarios was not different between single and dual fewer task critical AOIs during states of high cognitive load. Taken

Frontiers in Psychology 10 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

FIGURE 6
Individual data points and their respective group means for situation awareness (SA) (A), SA understanding (B), SA supply (C), and SA demand are
demonstrated for single flying and dual task conditions. (D) Error bars represent SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

together, the reported changes in gaze behavior may reflect minor of gaze metrics, which found that increases in task difficulty (i.e.,
changes in information processing that helped maintain flight significant crosswinds) (Ayala et al., 2023; Ayala et al., 2024) and
performance and scanning of task critical AOIs across single and dual cognitive load (i.e., dual task paradigms) (Engström et al., 2017;
task conditions during cruise. However, the significance of these Kanaan and Moacdieh, 2022; Lu et al., 2020; Yang et al., 2018; Yang
changes in gaze behavior and their impact on task performance is not et al., 2022) resulted in a narrowing of attention during high difficulty/
clear and require further research. cognitive load task conditions via reductions in SGE, GTE, saccade
In contrast to the traditional gaze metrics, the more dynamic amplitudes, visual scanning speed (scan path length per second),
measures of gaze dispersion (SGE), gaze sequence complexity (GTE), along with increases in the percentage of time looking at the center of
and cognitive tunneling (i.e., poor gaze monitoring characterized by the external environment (i.e., center of the road or the front window)
a lack of instrument scanning and long dwells outside the cockpit) and cognitive tunneling behavior. A key distinguishing feature
demonstrated no significant differences between single and dual tasks, between previous dual task work and the current work is the type of
nor did they show any interaction with phase of flight to suggest that secondary task employed. Unlike previous studies that used tasks that
the task condition had a significantly larger impact on gaze during the are not typically practiced or encountered often in traditional domain-
approach and landing phase of flight. Instead, the distribution of specific routines (i.e., mental arithmetic tasks or the n-back task)
fixations (SGE) demonstrated an increase in the allocation of attention (Crosby and Parkinson, 1979; Engström et al., 2017; Kanaan and
across a wider range of AOIs during approach and landing compared Moacdieh, 2022; Lu et al., 2020; Yang et al., 2018; Yang et al., 2022),
to cruise, which was associated with a number of changes in traditional the current auditory task closely resembled the aspect of responding
measures of gaze behavior (i.e., dwell time and fixation duration). to aircraft specific ATC calls while ignoring ATC calls to other aircraft.
Given that these changes were not associated with the increase in It is important to note that the actual ATC calls and verbal responses
cognitive load from the task condition manipulation or an interaction were replaced with target and non-target tones that required a simple
with phase of flight, we propose that the findings reflect the changes button press, mimicking aspects of auditory perception and
in task objectives and sub-goals between cruise and landing. These identification, without high working memory demands or speech
findings are in stark contrast to previous work employing a similar set production demands. Accordingly, the lack of verbal responses and

Frontiers in Psychology 11 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

memory of relevant call signs (i.e., Whiskey-Alpha-Tango-52) might gaze behavior changes, it was less detrimental to information
have resulted in the addition of a secondary task without necessarily processing compared to novice players. This finding was taken to
increasing the demands on working memory that might have been evince that expert players developed a level of information processing
necessary to see performance deficits in the primary task (Bhojwani and performance efficiency that spared the use of cognitive resources,
et al., 2022; Engström et al., 2017; Lavie, 2005). This finding was which were then used to support task performance when cognitive
similar to precious work done using a dual task driving paradigm load and anxiety increased. Notably, research examining the extent to
(Cao and Liu, 2013), where researchers found that a speech which improved processing and performance efficiency impacts gaze
comprehension task alone had minimal impact on driving behavior during states of high cognitive load is sparse and requires
performance. Evidently, the type of secondary task employed in dual further investigation.
task paradigms affects its impact on primary task performance and
gaze measures. This may be accounted for by Baddley’s model of
working memory (1986), which suggests that specialized subsystems Limitations and future directions
exist that represent particular types of information. For instance,
verbal (i.e., phonological loop) and visuospatial (i.e., visuospatial This work was constrained by two main methodological
sketchpad) types of sensory information are structurally independent limitations. First, the method through which the auditory tone task
of one another, and the integrity of information being represented in was implemented was done because it was easy to conduct and
one domain (i.e., visuospatial demands of flying and aircraft) is not at control. However, it did not fully represent the verbal perception and
risk of interference effects from information that may be received and speech production aspects of ATC communication that likely added
maintained through another domain (i.e., secondary auditory to the working memory load that participants would have had to
response task). Therefore, we propose that the cognitive load manage. Although alternative accounts have been suggested, this
manipulation was not sufficiently challenging to warrant any might have played a role in why the dual task paradigm employed was
significant deviations from pilot scan patters between the cognitive not sufficiently challenging for the recruited pilot group. Naturally,
load manipulations. this produced additional questions for future research about its utility
The Cognitive Control Hypothesis (Engström et al., 2017) in providing an objective measure of pilot monitoring behavior and
provides support for an alternative explanation for the lack of flight proficiency. Specifically, experienced pilots who have reached a
performance and gaze findings reported in the current work. Namely, level of information processing and performance efficiency with the
the apparent inconsistencies demonstrated may be reconciled if evaluated flight maneuvers will have additional cognitive control
we were to consider the flight scenario employed to be a task resources to allocate toward secondary cognitively demanding tasks.
associated with efficient performance and not controlled performance. In contrast, less experienced pilots who are still performing the same
Since the recruitment pool consisted of pilots who all obtained at least flight maneuvers at a controlled level of performance will have fewer
their PPL - thus implying a certain level of proficiency with flying a cognitive control resources to allocate toward a secondary cognitively
single engine aircraft-, perhaps their experience with flight circuits demanding task and will likely show larger task related and
(including cruise and landing phases of flight, which are normally information processing impairments. The extent to which this will
accompanied by numerous auditory-verbal ATC calls) may become evident in pilot monitoring behaviors via gaze analyses
be characterized as well-practiced maneuvers that are refined over the requires further exploration. Another direction for future research is
course of training. Thus, it is not a stretch to suggest that task to examine the effect of a dual-task paradigm using an auditory
performance is supported during challenging scenarios by improved secondary task that more realistically simulates ATC verbal
neural efficiency among low-time pilots who obtain their PPL/CPL communication. To that point, this work provides a necessary
certifications. This may fall along a gradient of improved processing empirical stepping stone from which auditory ATC call outs can
and performance efficiencies that are associated with practice- be used in place of auditory tones to not only increase cognitive load
dependent neuroplastic changes. These changes may not necessarily sufficiently, but also to increase fidelity (i.e., mimicking a busy airspace
reflect diminished neural involvement in the performed actions, but with frequent ATC calls being conducted to several aircraft).
rather reflect the development of neural connections that facilitate Furthermore, a more significant limitation associated with the current
performance speed and accuracy while reducing the cognitive work is connected to the way in which auditory response time data
demands of skilled performance (Immink et al., 2020). Thus, highly were collected. The auditory tones were created and recorded using an
experienced pilots would be more efficient than those examined in the external device VPixx, while the hardware used to collect the
current work. Since the Cognitive Control Hypothesis states that responses was implemented in the ALSIM simulator. Therefore, the
manipulations to cognitive load selectively impairs controlled start and stop of trials across these various streams of stimulus
performance, the current study likely demonstrates a case wherein presentation and data collection were tightly coordinated between two
increasing cognitive load through a secondary task left the primary researchers to ensure button presses (to start and stop trial) were
flying subtasks and pilot gaze behavior mostly unaffected. Indeed, occurring at the same time. Nevertheless, human variability exists in
work by Nibbeling et al. (2012) demonstrated similar findings when simple button presses (Stergiou and Decker, 2011) and the
examining the effects of anxiety and a dual task paradigm on a dart synchronization of the auditory tones with responses was likely not
throwing task across novice and expert players. Specifically, primary perfectly synchronized. Future work should incorporate stimulus
task performance, perceived effort and gaze behavior were negatively presentation and response recording in the simulator environment to
impacted by anxiety and cognitive load in novice players whereas get a more accurate recording of auditory tone responses. Despite this
expert players experienced little-to-no changes in performance and limitation, it is important to note that the auditory response accuracy
perceived effort. Although, expert players also demonstrated negative data reported in the current study echoes previous work in

Frontiers in Psychology 12 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

demonstrating the presence a dual task cost that is further exaggerated with the local legislation and institutional requirements. The
under more demanding task conditions. This is a finding that was not participants provided their written informed consent to participate in
contaminated by the synchronization limitation and thus provides this study.
support in the auditory task findings reported here (Allsop and Gray,
2014; Allsop et al., 2017; Crosby and Parkinson, 1979; Lutnyk et al.,
2023; Yang et al., 2022).
Author contributions
Conclusion NA: Writing – review & editing, Writing – original draft,
Visualization, Validation, Methodology, Investigation, Formal
The present paper investigated the utility of gaze behavior in analysis, Data curation, Conceptualization. SK: Writing – review &
assessing changes in cognitive load within a simulated flying task editing, Resources. EI: Writing – review & editing, Resources, Funding
using a dual task paradigm. Auditory task performance demonstrated acquisition. SC: Writing – review & editing. EN-S: Writing – review
that increases in cognitive load differentially impacted the cruise and & editing, Validation, Supervision.
landing phases of flight. Specifically, dual task cost in response time
and response errors was higher in the approach and landing phase of
flight compared to cruise. The increase in task demands across flight Funding
phases was associated with an increase in average dwell duration
across AOIs that were crucial to the phase specific sub-tasks. However, The author(s) declare that financial support was received for the
since most gaze behavior metrics and flight performance remained research, authorship, and/or publication of this article. This research
unchanged when the dual task was introduced it is likely that the was supported in part by grant 00753 from the New Frontiers in
recruited pilot group was not sufficiently challenged. These findings Research Fund.
provide guidance for future research in this area with regard to the
types of tasks that should be used for creating measurable levels of
cognitive load at the low-time pilot level, while also shedding light on Conflict of interest
the efficacy of utilizing dual task paradigms to assess pilot information
processing and flight proficiency. The authors declare that the research was conducted in the
absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Data availability statement
The raw data supporting the conclusions of this article will Publisher’s note
be made available by the authors, without undue reservation.
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated
Ethics statement organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
The studies involving humans were approved by the University of claim that may be made by its manufacturer, is not guaranteed or
Waterloo Ethics Board. The studies were conducted in accordance endorsed by the publisher.

References
Abel, S. M. (2009). Stimulus complexity and dual tasking effects on sustained Babu, M. D., JeevithaShree, D. V., Prabhakar, G., Saluja, K. P. S., Pashilkar, A., and
auditory attention in noise. Aviat. Space Environ. Med. 80, 448–453. doi: 10.3357/ Biswas, P. (2019). Estimating pilots’ cognitive load from ocular parameters through
ASEM.2383.2009 simulation and in-flight studies. J. Eye Mov. Res. 12:3. doi: 10.16910/jemr.12.3.3
Allsop, J., and Gray, R. (2014). Flying under pressure: effects of anxiety on attention Bhojwani, T. M., Lynch, S. D., Bühler, M. A., and Lamontagne, A. (2022). Impact of
and gaze behavior in aviation. J. Appl. Res. Mem. Cogn. 3, 63–71. doi: 10.1016/j. dual tasking on gaze behaviour and locomotor strategies adopted while circumventing
jarmac.2014.04.010 virtual pedestrians during a collision avoidance task. Exp. Brain Res. 240, 2633–2645.
doi: 10.1007/s00221-022-06427-2
Allsop, J., Gray, R., Bülthoff, H. H., and Chuang, L. (2017). Eye movement planning
on single-sensor-single-Indicator displays is vulnerable to user anxiety and cognitive Boeing. (2021). Statistical summary of commercial jet airplane accidents-worldwide
load. J. Eye Mov. Res. 10, 8–1. doi: 10.16910/jemr.10.5.8 operations 1959-2021. Available online at: https://www.boeing.com/resources/
boeingdotcom/company/about_bca/pdf/statsum.pdf
Ayala, N., Mardanbegi, D., Zafar, A., Niechwiej-Szwedo, E., Cao, S., Kearns, S., et al.
(2024). Does fiducial marker visibility impact task performance and information Brams, S., Hooge, I. T., Ziv, G., Dauwe, S., Evens, K., De Wolf, T., et al. (2018). Does
processing in novice and low-time pilots? Comput. Graph. 119:103889. doi: 10.1016/j. effective gaze behavior lead to enhanced performance in a complex error-detection
cag.2024.103889 cockpit task? PLoS One 13:e0207439. doi: 10.1371/journal.pone.0207439
Cao, S., and Liu, Y. (2013). Concurrent processing of vehicle lane keeping and speech
Ayala, N., Zafar, A., Kearns, S., Irving, E., Cao, S., and Niechwiej-Szwedo, E. (2023).
comprehension tasks. Accid. Anal. Prev. 59, 46–54. doi: 10.1016/j.aap.2013.04.038
The effects of task difficulty on gaze behaviour during landing with visual flight rules in
low-time pilots. J. Eye Mov. Res. 16:3. doi: 10.16910/jemr.16.1.3 Ciuperca, G., and Girardin, V. (2007). Estimation of the entropy rate of a countable
Markov chain. Commun. Stat. 36, 2543–2557. doi: 10.1080/03610920701270964
Ayala, N., Zafar, A., and Niechwiej-Szwedo, E. (2022). Gaze behaviour: a window into
distinct cognitive processes revealed by the tower of London test. Vis. Res. 199:108072. Crosby, J. V., and Parkinson, S. R. (1979). A dual task investigation of pilots’ skill level.
doi: 10.1016/j.visres.2022.108072 Ergonomics 22, 1301–1313. doi: 10.1080/00140137908924704

Frontiers in Psychology 13 frontiersin.org

Ayala et al. 10.3389/fpsyg.2024.1439401

Di Nocera, F., Camilli, M., and Terenzi, M. (2007). A random glance at the flight deck: Shannon, C. E. (1984). A mathematical theory of communication. Bell Syst. Tech. J.
Pilots' scanning strategies and the real-time assessment of mental workload. J. Cogn. 27, 379–423.
Eng. Decis. 1, 271–285. doi: 10.1518/155534307X255627
Shiferaw, B., Downey, L., and Crewther, D. (2019). A review of gaze entropy as a
Engström, J., Markkula, G., Victor, T., and Merat, N. (2017). Effects of cognitive load measure of visual scanning efficiency. Neuroscience and biobehavioral reviews. 96,
on driving performance: the cognitive control hypothesis. Hum. Factors 59, 734–764. 353–366. doi: 10.1016/j.neubiorev.2018.12.007
doi: 10.1177/0018720817690639
Simmons, S. M., Hicks, A., and Caird, J. K. (2016). Safety-critical event risk associated
Federal Aviation Administration (2021). Airplane flying handbook. Washington, DC: with cell phone tasks as measured in naturalistic driving studies: a systematic review and
Federal Aviation Administration (FAA-H-8083-3C). meta-analysis. Accid. Anal. Prev. 87, 161–169. doi: 10.1016/j.aap.2015.11.015
Glaholt, M. G. (2014). Eye tracking in the cockpit: a review of the relationships Stergiou, N., and Decker, L. M. (2011). Human movement variability, nonlinear
between eye movements and the aviators cognitive state. Available online at: https:// dynamics, and pathology: is there a connection? Hum. Mov. Sci. 30, 869–888. doi:
apps.dtic.mil/sti/pdfs/AD1000097.pdf 10.1016/j.humov.2011.06.002
Hudspeth, A. J., Jessell, T. M., Kandel, E. R., Schwartz, J. H., and Siegelbaum, S. A. Taylor, R. M., and Selcon, S. J. (1990). Cognitive quality and situational awareness with
(Eds.).(2013). Principles of neural science. McGraw-Hill, Health Professions Division. advanced aircraft attitude displays. Proc. Human Factors Soc. Ann. Meet. 34, 26–30. doi:
Immink, M. A., Verwey, W. B., and Wright, D. L. (2020). The neural basis of cognitive 10.1177/154193129003400107
efficiency in motor skill performance from early learning to automatic stages. van De Merwe, K., Van Dijk, H., and Zon, R. (2012). Eye movements as an indicator
Neuroergonomics, 221–249. doi: 10.1007/978-3-030-34784-0_12 of situation awareness in a flight simulator experiment. Int. J. Aviat. Psychol. 22, 78–95.
Kanaan, D., and Moacdieh, N. M. (2022). Eye tracking to evaluate the effects of doi: 10.1080/10508414.2012.635129
interruptions and workload in a complex task. Hum. Factors 64, 1168–1180. doi: Vickers, J. N., and Williams, A. M. (2017). “The role of mental processes in elite sports
10.1177/0018720821990487 performance” in Oxford Research Encyclopedia of Psychology (Oxford: Oxford
Land, M. F., and Hayhoe, M. (2001). In what ways do eye movements contribute to University Press).
everyday activities? Vis. Res. 41, 3559–3565. doi: 10.1016/S0042-6989(01)00102-X Wilson, G. F. (2002). An analysis of mental workload in pilots during flight using
Lavie, N. (2005). Distracted and confused?: selective attention under load. Trends multiple psychophysiological measures. Int. J. Aviat. Psychol. 12, 3–18. doi: 10.1207/
Cogn. Sci. 9, 75–82. doi: 10.1016/j.tics.2004.12.004 S15327108IJAP1201_2
Lu, T., Lou, Z., Shao, F., Li, Y., and You, X. (2020). Attention and entropy in simulated Wright, J. L., Quinn, S. A., Chen, J. Y. C., and Barnes, M. J. (2014). Individual
flight with varying cognitive loads. Aerospace Med. Human Perform. 91, 489–495. doi: differences in human-agent teaming: an analysis of workload and situation awareness
10.3357/AMHP.5504.2020 through eye movements. Proc. Human Factors Ergonom. Soc. Ann. Meet. 58, 1410–1414.
doi: 10.1177/1541931214581294
Lutnyk, L., Rudi, D., Schinazi, V. R., Kiefer, P., and Raubal, M. (2023). The effect of
flight phase on electrodermal activity and gaze behavior: a simulator study. Appl. Ergon. Yang, S., Kuo, J., and Lenné, M. G. (2018). Analysis of gaze behaviour to measure
109:103989. doi: 10.1016/j.apergo.2023.103989 cognitive distraction in real-world driving. Proc. Human Fact. Ergonom. Soc. Ann. Meet.
62, 1944–1948. doi: 10.1177/1541931218621441
Manor, B., Costa, M. D., Hu, K., Newton, E., Starobinets, O., Kang, H. G., et al. (2010).
Physiological complexity and system adaptability: evidence from postural control dynamics Yang, S., Wilson, K. M., Roady, T., Kuo, J., and Lenné, M. G. (2022). Evaluating driver
of older adults. J. Appl. Physiol. 109, 1786–1791. doi: 10.1152/japplphysiol.00390.2010 features for cognitive distraction detection and validation in manual and level 2
automated driving. Hum. Factors 64, 746–759. doi: 10.1177/001872082096
National Transportation Safety Board (1994). A review of Flightcrew-involved, major 4149
accidents of US air carriers, 1978 through 1990. Available online at: https://reports.
Zafar, A., Martin Calderon, C., Yeboah, A. M., Dalton, K., Irving, E., and
aviation-safety.net/1994/19940618-1_LJ25_XA-BBA.pdf
Niechwiej-Szwedo, E. (2023). Investigation of camera-free eye-tracking
Nibbeling, N., Oudejans, R. R., and Daanen, H. A. (2012). Effects of anxiety, a glasses compared to a video-based system. Sensors 23:7753. doi: 10.3390/s231
cognitive secondary task, and expertise on gaze behavior and performance in a far 87753
aiming task. Psychol. Sport Exerc. 13, 427–435. doi: 10.1016/j.psychsport.2012.02.002
Zhou, J., Hao, Y., Wang, Y., Jor'dan, A., Pascual-Leone, A., Zhang, J., et al. (2014).
Reimer, B. (2009). Impact of cognitive task complexity on drivers’ visual tunneling. Transcranial direct current stimulation reduces the cost of performing a cognitive
Transport. Res. Record 2138, 13–19. doi: 10.3141/2138-03 task on gait and postural control. Eur. J. Neurosci. 39, 1343–1348. doi: 10.1111/
ejn.12492
Schwenk, M., Zieschang, T., Oster, P., and Hauer, K. (2010). Dual-task performances
can be improved in patients with dementia: a randomized controlled trial. Neurology 74, Ziv, G. (2016). Gaze behavior and visual attention: a review of eye tracking studies in
1961–1968. doi: 10.1212/WNL.0b013e3181e39696 aviation. Int. J. Aviat. Psychol. 26, 75–104. doi: 10.1080/10508414.2017.1313096

Frontiers in Psychology 14 frontiersin.org