PreprintPDF Available

Abstract and Figures

Speeding because of time pressure is a leading contributor to traffic accidents. Previous research indicates that people respond to time pressure through increased physiological activity and by adapting their task strategy in order to mitigate task demands. In the present driving simulator study, we investigated effects of time pressure on measures of eye movement, pupil diameter, cardiovascular and respiratory activity, driving performance, vehicle control, limb movement, head position, and self-reported state. Based on existing theories of human behavior under time pressure, we distinguished three categories of results: (1) driving speed, (2) physiological measures, and (3) driving strategies. Fifty-four participants drove a 6.9-km urban track with overtaking, car following, and intersection scenarios, first with no time pressure (NTP) and subsequently with time pressure (TP). The results showed that under TP in comparison to NTP, participants (1) drove significantly faster, an effect that was also reflected in auxiliary measures such as maximum brake position, throttle activity, and lane keeping precision, (2) exhibited increased physiological activity, such as increased heart rate, increased respiration rate, increased pupil diameter, and reduced blink rate, and (3) adopted scenario-specific strategies for effective task completion, such as driving to the left of the lane during car following, and early visual lookout when approaching intersections. The effects of TP relative to NTP were generally large and statistically significant. However, individual differences in absolute values were large. Hence, we recommend that real-time driver feedback technologies use relative instead of absolute criteria for assessing the driver’s state.
Content may be subject to copyright.
The effects of time pressure on driver performance
and physiological activity: A driving simulator study
E. Rendon-Velez
a,1
, P. M . van Leeuwen
b,
,1
, R. Happee
b
, I. Horváth
a
, W.F. van der Vegte
a
,
J.C.F. de Winter
b
a
Delft University of Technology, Faculty of Industrial Design Engineering, The Netherlands
b
Delft University of Technology, Biomechanical Engineering, Faculty of Mechanical, Maritime and Materials Engineering, The Netherlands
article info
Article history:
Received 21 March 2015
Received in revised form 25 February 2016
Accepted 13 June 2016
Keywords:
Simulation
Virtual reality
Workload
Psychophysiology
abstract
Speeding because of time pressure is a leading contributor to traffic accidents. Previous
research indicates that people respond to time pressure through increased physiological
activity and by adapting their task strategy in order to mitigate task demands. In the
present driving simulator study, we investigated effects of time pressure on measures of
eye movement, pupil diameter, cardiovascular and respiratory activity, driving
performance, vehicle control, limb movement, head position, and self-reported state.
Based on existing theories of human behavior under time pressure, we distinguished three
categories of results: (1) driving speed, (2) physiological measures, and (3) driving
strategies. Fifty-four participants drove a 6.9-km urban track with overtaking, car follow-
ing, and intersection scenarios, first with no time pressure (NTP) and subsequently with
time pressure (TP) induced by a time constraint and a virtual passenger urging to hurry
up. The results showed that under TP in comparison to NTP, participants (1) drove signif-
icantly faster, an effect that was also reflected in auxiliary measures such as maximum
brake position, throttle activity, and lane keeping precision, (2) exhibited increased phys-
iological activity, such as increased heart rate, increased respiration rate, increased pupil
diameter, and reduced blink rate, and (3) adopted scenario-specific strategies for effective
task completion, such as driving to the left of the lane during car following, and early visual
lookout when approaching intersections. The effects of TP relative to NTP were generally
large and statistically significant. However, individual differences in absolute values were
large. Hence, we recommend that real-time driver feedback technologies use relative
instead of absolute criteria for assessing the driver’s state.
Ó2016 Elsevier Ltd. All rights reserved.
1. Introduction
1.1. The dangers of ‘time pressure’
A large portion of road traffic crashes occurs because drivers have been speeding or committing other types of traffic
violations, such as tailgating and dangerous overtaking (Elander, West, & French, 1993; Elvik, Christensen, & Amundsen,
http://dx.doi.org/10.1016/j.trf.2016.06.013
1369-8478/Ó2016 Elsevier Ltd. All rights reserved.
Corresponding author at: Delft University of Technology, Biomechanical Engineering, Faculty of Mechanical, Maritime and Materials Engineering,
Mekelweg 2, 2628 CD Delft, The Netherlands.
E-mail address: P.M.vanLeeuwen@TUDelft.nl (P. M . van Leeuwen).
1
Joint first authors.
Transportation Research Part F 41 (2016) 150–169
Contents lists available at ScienceDirect
Transportation Research Part F
journal homepage: www.elsevier.com/locate/trf
2004; Evans & Wasielewski, 1982; Organisation for Economic Co-operation & Development, 2006; Parker, Reason, Manstead,
& Stradling, 1995). Speeding is a factor in between 10% and 40% of accidents on European roads (Organisation for Economic
Co-operation & Development, 2006; Treat et al., 1979). McKenna (2005) reported that 33% of 9470 surveyed speeding
offenders indicated that they were in a hurry at the time of their speeding offence. Another survey study by Collinson
(2014) indicated that 77% of 150 people who were caught speeding did so because they lacked time to make the journey.
Several factors may explain why drivers decide to speed and violate the traffic rules. This includes personality factors,
such as thrill seeking, pleasure in fast driving, and aggressiveness, as well as environmental factors, such as peer pressure,
and perhaps most importantly, a shortage of available time (e.g., Beck, Daughters, & Ali, 2013; Beck, Wang, & Yan, 2012;
Cœugnet, Miller, Anceaux, & Naveteur, 2013; Cœugnet, Naveteur, Antoine, & Anceaux, 2013; Matthews, 2002; Rendon-
Velez, Horvath, & Van der Vegte, 2012; Rothengatter, 1988). Note, however, that a time constraint alone is not a necessary
condition for speeding; the driver also has to believe it is important to complete the task in time (Benson & Beach, 1996;
Cœugnet et al., 2013).
1.2. Models that describe how time pressure influences (driver) performance
A model of Wickens, Lee, Liu, and Gordon-Becker (2004) describes how (1) information input, (2) information-processing
efficiency, and (3) task performance are influenced by external ‘stressors’ (such as pressure to complete a task in time).
Specifically, Wickens et al.’s model illustrates that external stressors have direct influences on the quality of the information
input and task performance (e.g., through increased levels of noise, lighting, or vibrations). The direct consequence of driving
faster is that a higher amount of information has to be processed per unit of time. Thus, driving speed has a direct influence
on the information input rate. Stress also has indirect psychological influences. For example, having to complete a task in a
short amount of time could lead to high mental workload, anxiety, frustration, and anger, which in turn reduces information
processing efficiency.
Maule and Hockey (1993) describe the effects of time pressure by means of a two-level control model. According to this
model, the human cognitive system is self-regulatory. On the lower control level, small discrepancies between the current
and target mental state are regulated by subconscious corrective actions (e.g., changes in speed, memory use, timing). When
the discrepancy between the current and target state is large and subconscious control strategies are inadequate, control
temporarily shifts to a higher level of cognitive (conscious) control (Maule & Hockey, 1993; see also Robert & Hockey,
1997). At this higher level, four modes are available to cope with high task demands: (1) increasing effort (trying harder)
and accelerating control actions, (2) adopting a strategy that requires less effort, (3) changing the environment by removing
stressors (e.g., re-negotiating the time deadline), or (4) doing nothing to reduce task demands (for further studies, see Edland
& Svenson, 1993; Miller, 1960; Wright, 1974). When a driver adopts mode 1, this will be reflected in measures of speed as
well as physiological measures associated with the activity of the sympathetic nervous system (Maule & Hockey, 1993).
Modes 3 and 4 are usually not feasible when having to drive to a destination in a fixed amount of time, as the driver can
control the state of his own vehicle in the environment but can hardly modify the environment itself. In this paper, our focus
is on modes 1 and 2. That is, in the present study, we evaluated whether drivers modify their lateral/longitudinal driving
behavior, posture, and gaze patterns by increasing their effort (mode 1) or by modifying their behavior in such a way that
the driving task becomes easier to carry out (mode 2) while maintaining a high average driving speed in order to arrive at the
destination in time.
1.3. Previous research that investigated the effects of time pressure on driving performance
Several previous studies have demonstrated the effects of time pressure on measures of driving performance. Van der
Hulst, Rothengatter, and Meijman (1998) studied car following behavior in fog conditions using a driving simulator. Partic-
ipants who were instructed to drive on a fixed time schedule showed less variability in their time headway due to deceler-
ations of lead vehicles compared to a control group instructed to drive as they would normally do. The improved precision in
the control of the vehicle suggests that the drivers adapted to the time constraint by increasing their level of alertness (Van
der Hulst et al., 1998), an effect that corresponds to mode 1 (trying harder) in the model of Maule and Hockey (1993).
Cnossen, Rothengatter, and Meijman (2000) instructed drivers to drive as fast as possible in a simulated environment.
The results showed that participants had poorer lane keeping accuracy when they drove as fast as possible compared to
when asked to adhere to the speed limits as if they were taking a driving test. In another driving simulator study, Zhai,
Accot, and Woltjer (2004) found that drivers slowed down when they were required to maintain lane position accurately.
Conversely, when the lane width increased, drivers were able to drive faster. These latter two studies suggest that the effects
of time pressure can be described as a speed-accuracy tradeoff (see also Szalma, Hancock, & Quinn, 2008, for a meta-analysis
on the effects of time pressure on measures of speed and accuracy).
Performance measures of speed and accuracy are advantageous for driver assessment applications because they represent
an objective and observable state of the vehicle in its environment. Another advantage of these measures is that they are
closely related to safety and accidents (Aarts & Van Schagen, 2006; Cooper, 1997; Lajunen, Karola, & Summala, 1997). A
disadvantage of performance measures of speed and accuracy is that they cannot readily be used to identify whether a driver
is subjected to time pressure or not, because these measures are highly situation-dependent (e.g., Cantin, Lavallière,
Simoneau, & Teasdale, 2009). For example, a driver under time pressure may be stuck in a traffic jam, as a result of which
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 151
speed/accuracy measures of driving performance are not informative at all. Similarly, a measure of lane keeping accuracy
will be meaningless when a time-pressurized driver is frequently overtaking other road users.
1.4. The potential of psychophysiology for studying the effects of time pressure on car driving
When humans are subjected to stressors (such as time pressure), they tend to show a variety of physiological responses
such as pupil dilation, increased heart rate, slowed digestion, and a constriction of blood vessels, mechanisms that are col-
lectively known as the ‘fight-or-flight’ response (e.g., Cain, 2007; Kramer, 1991; Wickens et al., 2004). Furthermore, visual
and cognitive tunneling occurs, referring to the fact that a stressed person stops carrying out secondary tasks and processes
the cues that are most immediate and familiar (Hancock, 1989; Hancock & Szalma, 2008).
Various experimental studies in flight/driving simulators and real vehicles (e.g., Backs, Lenneman, Wetzel, & Green, 2003;
Brookhuis & De Waard, 2010; Veltman & Gaillard, 1996) have measured physiological responses as a function of task
demands. Examples include physiological measurements during the presence/absence of a secondary task (Mehler,
Reimer, Coughlin, & Dusek, 2009), as a function of road infrastructure (Dijksterhuis, Brookhuis, & De Waard, 2011), or for
different levels of automated driving (De Winter, Happee, Martens, & Stanton, 2014). An experiment by Cnossen et al.
(2000) showed increased heart rates when participants drove as fast as possible compared to driving as accurately as
possible.
Car driving is predominantly a visual task (e.g., Sivak, 1996), and a large body of research has evaluated the effects of task
demands on drivers’ visual scanning behavior (e.g., Crundall & Underwood, 1998; Recarte & Nunes, 2000; Reimer, 2009;
Wikman, Nieminen, & Summala, 1998). In a driving simulator study, Rogers, Kadar, and Costall (2005) increased the task
demands by increasing the driving speed during a straight-lane driving task. Their findings showed that participants, regard-
less of their level of driving experience, narrowed their gaze distribution when the driving speed was increased. Recently,
remote eye trackers have shown to be promising tools for measuring the pupil dilation response as a function of cognitive
task demands in low-cost measurement setups (Klingner, Kumar, & Hanrahan, 2008; Marquart & De Winter, 2015) as well as
in driving simulators (Palinko, Kun, Shyrokov, & Heeman, 2010).
In addition to the human physiological response, time pressure also influences bodily posture and kinetics (Birch, Juul-
Kristensen, Jensen, Finsen, & Christensen, 2000; Bongers, De Winter, Kompier, & Hildebrandt, 1993; Van Galen & Van
Huygevoort, 2000). Using pressure sensors in the seat, Riener, Ferscha, and Matscheko (2008) found that drivers adjusted
their posture in curves as a function of curve radius and driving speed. Tran and Trivedi (2010) showed using a vision-
based motion tracking system that relaxed drivers took a more ‘leaned back’ posture, whereas concentrated drivers showed
a more ‘forward leaning’ posture during a highway-driving task in a simulator.
1.5. The present study
In a driving simulator experiment, we evaluated two levels of time pressure: a baseline condition with no time pressure
(NTP) and a time pressure (TP) condition in which drivers drove with a time constraint imposed on their driving task. Par-
ticipants drove along an urban road in which various scenarios occurred: car following, overtaking an obstacle, and crossing
intersections. We evaluated the effects of time pressure on a large number of dependent measures (including measures of
eye movement, pupil diameter, cardiovascular and respiratory activity, driving performance, vehicle control, limb move-
ment, head position, and self-reported status) to explore which of these measures are indicative of driving under time
pressure.
Based on the models of Wickens et al. (2004) and Maule and Hockey (1993), we derived three broad hypotheses. Our first
hypothesis was that drivers under time pressure drive at a higher speed and execute their tasks at a higher rate. This first
hypothesis provides what is essentially a manipulation check as to whether, and to what extent, the task instructions cause
participants to arrive at the destination in a shorter amount of time compared to driving without time pressure. We also
investigated auxiliary measures of driving speed, such as braking and throttling activity, as well as lane keeping accuracy
(accuracy was expected to decrease when driving faster, as predicted by the speed-accuracy tradeoff). The second hypothesis
was that drivers react physiologically to the presence of the time pressure stressor. Although it is well established that stress
causes signs of sympathetic arousal, what is less well known is which of the physiological measures are most sensitive to
time pressure instructions in a car driving task. Furthermore, the present study exhibits several features that allowed us
to test this hypothesis with a high level of spatiotemporal detail. Specifically, we synchronized the driving performance
and physiological signals, allowing us to explore which of the measures are indicative of driving under time pressure at
the different scenarios along the route. The third hypothesis was that drivers adapt their behavior by means of adjusting
their driving strategy. As described above, we defined a change in strategy as a change in driving or visual behavior (other
than simply driving faster) that allowed the driver to achieve the goal of arriving at the destination with greater
effectiveness.
In the analysis, we put special emphasis on physiological data, because physiological data can provide a real-time assess-
ment of the driver’s state without requiring an overt reaction from the driver (De Waard, 1996; Kramer, 1991). For example,
it might be possible to detect an altered physiological state of a driver when the driving speed is restricted or when the driver
does not physically move the wheel or pedals. Compared to vehicle-centered performance measures, measures based on
152 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
human physiology can provide person-centered indicators of time pressure that may be of value in the development of dri-
ver monitoring and feedback applications (cf. Mehler et al., 2009; Reimer, 2009).
2. Methods
2.1. Participants
Fifty-six participants (48 males and 8 females) were recruited from the Delft University of Technology student and
employee community. Participants were in possession of a valid driver’s license and had normal or corrected-to-normal
eye sight. Prior to the experiment participants filled out a 18-item intake questionnaire consisting of general items (age, gen-
der, wearing glasses or contact lenses, medication, educational qualification, occupation), simulation-related items (playing
computer games, prior experience in driving simulation, number of participated simulator experiments in the past), and
driving experience items (e.g., driving frequency and mileage in the past 12 months, and accident involvement and traffic
violations in the past 36 months). Some of these items were derived from the Driving Habits Questionnaire (Owsley,
Stalvey, Wells, & Sloane, 1999).
Of the 54 participants who completed the experiment, there were 46 males (mean age = 28.5, SD = 4.3) and 8 females
(mean age = 27.0, SD = 2.9). On average participants had their driving license for 9.1 (SD = 4.5) years, with a mean annual
mileage of 6350 (SD = 8116) km. Three participants reported the use of medication (insulin, Aerius and folate, and paraceta-
mol, respectively) and 18 participants wore contact lenses or glasses during driving. Twenty participants reported prior
experience in a driving simulator, with a mean of 0.59 (SD = 1.12, N= 54) experiments per participant. For an overview of
the results of the intake questionnaire, see Table 1. Before commencing the experiment, all participants provided written
informed consent. The research was approved by the Human Research Ethics Committee of the Delft University of
Technology.
2.2. Apparatus
A fixed-base driving simulator (Green Dino, Wageningen, the Netherlands) was used in this experiment. The simulator
cabin was equipped with the following components: steering wheel, ignition key, gear lever, single seat, and pedals. The
steering wheel, pedals, gear lever, and indicators were obtained from a regular passenger car, and the dashboard, interior,
and mirrors were integrated in the projected visuals, as shown in Fig. 1. Steering wheel force feedback was provided by a
passive spring system. Surround sound was used to provide auditory wind, tire, and engine feedback. The simulator provided
a horizontal field of view of 180 degrees by means of three projectors. The front view projection (front projector: NEC VT676)
had a resolution of 1024 768 pixels, and the side views (side projectors: NEC VT470) featured a resolution of
800 600 pixels. The simulation ran at a frequency of 100 Hz, and the frame rate of the visual projection was estimated
to be greater than 25 Hz (i.e., high enough to guarantee a smooth visual experience throughout the experiment).
Eye and head movements were recorded using a Smart Eye eye-tracking system (software version 5.9), consisting of
three remote mounted cameras (Sony XC-HR50) and two infrared illuminators. The data from the simulator and eye
tracker were sampled and stored synchronously at 60 Hz. The participant’s electrocardiogram (ECG) was obtained using
a lead II configuration with three disposable snap electrodes and was recorded on a portable Mobi8 device (Twente
Medical Systems International). The expansion of the thorax during inhalation and exhalation was measured using an
inductive effort belt (Sleep Sense) worn around the chest. This belt was connected to a respiration effort sensor
(RespiV6) which in turn was connected to the Mobi8 device. Both ECG and respiration data were received wireless
and stored at 256 Hz. Limb movements were measured using four wireless inertial 3D motion trackers (Xsens MTw)
placed at the ankles and wrists. The limb movement data were received wirelessly and stored at 75 Hz.
A trigger signal was sent when the clutch was pressed as the participant started the driving session. Using
this trigger signal, data from the peripheral hardware were synchronized with the driving simulator data during
post-processing.
Table 1
Distribution of participants (N= 54), for frequency of playing computer games, driving frequency, and educational qualification.
On average how often did you play
computer or video games in the last 12
months?
On average, how often did you drive a car in
the last 12 months?
What is your highest educational
qualification?
Every day 0 Every day 7 Primary/elementary school 1
4–6 days/week 3 4–6 days/week 5 Secondary/high school 0
1–3 days/week 3 1–3 days/week 14 Bachelor degree 15
About once a week 12 About once every two weeks 13 Postgraduate degree 38
Less than once a month 14 About once a month 7
Never 22 Less than once a month 5
Never 3
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 153
2.3. Independent variable
The independent variable was the time constraint imposed on the driving task. In the no time pressure (NTP) session, the
participant had sufficient time to complete the driving task. In the second session, time pressure (TP) was imposed by
requesting the participants to complete the driving task within 80% of their NTP completion time, with a minimum of
7 min 20 s (defined as an absolute minimum according to pilot tests by the authors). Thus, the time constraint was different
for each individual. In both sessions, the elapsed time was displayed on the virtual dashboard. Furthermore, during the NTP
and TP sessions, auditory information was provided: the voice of a previously recorded fictitious ‘passenger’ was played back
during both sessions. In the NTP session, the passenger was talking about casual things, while in the TP session, the
passenger was complaining about being late and was motivating the participant to hurry up. In both sessions, the passenger
sentences were uttered every 15 s.
2.4. Procedures
Prior to the simulator drives, participants received a paper handout explaining the experiment and procedures, and filled
out the 18-item intake questionnaire. Additionally, participants filled out the Mini Driver Behavior Questionnaire
(Mini-DBQ) to measure aberrant driving behaviors (Martinussen, Lajunen, Møller, & Özkan, 2013) and the Multidimensional
Driving Style Inventory (MDSI) for assessing driving style (Taubman-Ben-Ari, Mikulincer, & Gillath, 2004). Next, participants
watched a 5-min instruction video, explaining the driving simulator operation, the sensor instrumentation procedures, and
the instructions for the training and NTP sessions. The video informed the participants only about the upcoming training and
the NTP sessions, in order to ensure that participants were naïve to the specific instructions of the TP session while driving
the training and NTP sessions. After watching the instruction video, the inertial motion trackers were attached to the ankles
and wrists of the participants, and the three ECG electrodes were placed below the left and right clavicle and below the left
pectoral muscle in a lead II configuration. The respiration belt was placed at the diaphragm level of the sternum and tight-
ened sufficiently without causing discomfort.
Participants then seated themselves inside the driving simulator. Next, participants carried out a series of head move-
ments and eye movements to calibrate the eye tracker. Participants completed three sessions in the following order: training
session (T), no time pressure session (NTP), and time pressure session (TP). Before commencing with the training session,
participants were told to relax, and the instructions regarding the training and NTP sessions were repeated orally by the
experimenter. After having completed the NTP session, participants received a tablet showing the video instructions for
the TP session. After each session, a 5-min break took place during which participants remained seated in the driving sim-
ulator. During these breaks, participants filled out the NASA task load index (TLX) for measuring workload (Hart & Staveland,
1988). Furthermore, participants filled out a questionnaire measuring their perceived time pressure, as well as a 6-item con-
fidence questionnaire measuring their confidence in the driving task (see Section 2.8.5). The order of the driving sessions was
not counterbalanced, to facilitate the individually adapted time constraint in the TP condition.
2.5. Driving task
Prior to the training session, participants received video instructions to drive straight ahead, to cross the intersections,
and to overtake the obstacles when required. Participants were instructed to obey traffic rules and were informed that their
Fig. 1. One of the experimenters in the driving simulator, with inertial sensors and eye tracker.
154 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
lane was not a priority lane. For the NTP session, participants received instructions to drive safely and in a relaxed manner, as
if they drove a fictitious friend to the airport without any time constraints. After the NTP session participants received oral
and video instructions to drive to the airport with a time constraint. All sessions started with the vehicle from standstill at
the center of the lane, and participants were requested to start the vehicle by pressing the clutch pedal. Participants were
required to accelerate, brake, steer, and use the clutch and gear lever to operate the manual gearbox.
2.6. Driving environment
The driving environment consisted of an urban area with regular traffic conditions, identical in both the NTP and the TP
sessions. The two-way road had a length of 6970 m and a lane width of 4 m. The road consisted of 17 segments and 16 inter-
sections with stop signs and without traffic lights. Several traffic situations were triggered on passing specific positions in the
virtual scenery. The traffic situations included: (1) free driving, (2) car following with traffic in the opposing lane, (3) obstacle
overtaking with and without traffic in the opposing lane, and (4) intersections with and without approaching traffic. During
the car following scenarios, traffic in the opposing lane prevented participants from overtaking the lead car. Traffic in the
opposing lane during the obstacle overtaking events required participants to decelerate before the obstacle until the traffic
in the opposing lane had passed the obstacle. During the intersection scenarios with traffic, participants were unable to cross
the intersection until the traffic had cleared from the intersection. During the training session the environment was identical
to the TP and NTP sessions, but included three additional obstacle overtaking scenarios. See Table 2 for an overview of the
traffic scenarios during the three sessions and Fig. 2 for screenshots of the four scenarios.
2.7. Data processing
The driving simulator, eye-tracker, physiological, inertial, and force data were synchronized and re-sampled to 100 Hz
prior to post-processing. Data were analyzed from the start of each session to the point where participants were 380 m past
the final intersection (i.e., after having traversed 6835 m). At this location the end of the road was visible. The following post-
processing was performed on the recorded signals:
2.7.1. Steering signal
Steering signal data were low-pass filtered with a 3 Hz cut-off frequency, to remove the high frequency noise.
2.7.2. Eye movements and head movements
Eye movements and head movements were low-pass filtered with cut-off frequencies of 10 Hz and 5 Hz, respectively.
Data loss with remote mounted eye trackers occurs when the system is unable to detect a participant’s facial features, pupil,
or corneal reflections due to an obstruction of the eye-tracker cameras or due to large head movements (e.g., Prado Vega, Van
Leeuwen, Vélez, Lemij, & De Winter, 2013; Van Leeuwen, Gòmez i Subils, Ramon Jimenez, Happee, & De Winter, 2015; Van
Leeuwen, Happee, & De Winter, 2014). Eye closures were classified as a blink when the eye opening was smaller than 50% of
the participant’s median eye opening. Gaze data during blinks as well as data from 0.2 s before to 0.2 after segments of miss-
ing data were removed. When more than 60% of data had to be removed, all the eye-tracker data of the respective session
were removed from the analysis.
2.7.3. The pupil diameter
The pupil diameter is highly sensitive to illumination (Watson & Yellott, 2012). During the simulation, the illumination
intensity was a function of the virtual environment and varied with the participant’s location in the virtual world. The pupil
diameter measurements were corrected for illuminance intensity at each traveled distance in the virtual scenery using mea-
sured illumination intensity data (see supplementary material).
2.7.4. Physiological data
Physiological data were filtered before further processing. Specifically, the ECG signal was high-pass filtered at 10 Hz, to
remove low frequency drift from the signal. The resulting QRS complex of the ECG signal was de-noised using wavelets
Table 2
Overview of traffic scenarios.
Scenario Traffic condition Starting positions (m)
Free driving Without traffic in opposing lane 1710, 2875
Intersection crossing Without traffic in intersection lane 810, 2635, 3775, 4100, 4935, 5260, 6100
With traffic in intersection lane 180, 1135, 1680, 2005, 2845, 3150, 4305, 5470, 6425
Car following With traffic in opposing lane 210, 2035, 3850, 4965, 5500
Obstacle overtaking Without traffic in opposing lane 1495, 3320
a
, 3980
a
, 4665, 6635
With traffic in opposing lane 960, 2785, 3510, 4250, 4500
a
, 5410, 6250
a
Only during the training session.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 155
(Addison, 2005), and inter-beat intervals were extracted from the clean R-peak signal. The respiration rate signal was band-
pass filtered (0.05–1 Hz) to remove the low frequency drift and high frequency noise from the signal. The resulting signal
was used to calculate the inter-breath frequency from the time between two subsequent inhalation peaks.
2.7.5. Inertial sensor
Inertial sensor data were low-pass filtered with a 10 Hz cut-off frequency to remove the high frequency noise component.
2.8. Dependent measures
A number of dependent measures were calculated per session and per participant. The dependent measures were divided
into the following categories:
2.8.1. Driving performance
Lane keeping accuracy and precision were defined as the mean lateral position (m) (left = positive) and the standard
deviation of the lateral position (SDLP) (m), respectively (cf. Van Leeuwen et al., 2014). Obstacle overtaking maneuvers
were excluded from these measures. Measures of vehicle speed (mean speed and maximum speed) (m/s) were used to
capture driving style and task performance. During car following situations, the time headway (s) was determined
(for headways smaller than 300 m with respect to the lead car), a measure which is indicative of tailgating behavior
(Vogel, 2003).
2.8.2. Vehicle control
Mean absolute steering speed (deg/s) and throttle variance (minimum possible = 0, maximum possible = 0.25) were cal-
culated as measures of steering and throttle activity. Furthermore, the mean number of gear changes and the mean number
of brake applications were determined to represent the amount of control actions performed during the session. Finally, the
maximum brake pedal position, on a scale of 0 (minimum) to 1 (maximum), was determined as a measure of braking per-
formance (De Groot, De Winter, Wieringa, & Mulder, 2009). Mean limb accelerations (m/s
2
) were determined by taking the
mean of the square root of the sum of the squared x,y, and zcomponents of the measured wrist/ankle accelerations. The limb
acceleration measure is indicative of the driver’s control activity.
2.8.3. Eye movements and head movements
Gaze road center (GRC) (%) was calculated as the percentage of time that participants gazed within an approximately
8 deg radius from the road center. This measure is representative of the amount of gaze tunneling and has been
Fig. 2. Screenshots of the four traffic scenarios; free driving (top left), car following with traffic in the opposing lane (top right), obstacle overtaking without
traffic in the opposing lane (bottom left), and intersection crossing with traffic in the intersection lane (bottom right). Note. The route guidance arrow was
shown at each intersection.
156 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
demonstrated to be sensitive to secondary task demands (Van Leeuwen, Happee, & De Winter, 2013; Victor, Harbluk, &
Engström, 2005). Additionally, we calculated the horizontal gaze variance (HGV; deg
2
), representing the spread of visual
search. The percentages of time that the participants were glancing at the dials and clock were calculated from the gaze
vector with respect to predefined regions on the screen. These measures were used to verify the use of the simulated
dashboard instruments and the clock showing the elapsed time in the session. The mean head position (m) was defined
as the longitudinal component of the distance from the participants head to the top of the steering wheel (as determined
by the eye-tracker system), and was regarded as a measure of driver posture.
2.8.4. Physiological responses
The mean eye blink frequency (Hz) and the mean pupil diameter (mm) were extracted from the eye-tracker data, as these
measures are known to be sensitive to task demands (Beatty, 1982; Recarte, Pérez, Conchillo, & Nunes, 2008). From the
respiratory measurements, the mean respiration rate (1/min) and the respiration amplitude (mm) were calculated. These
measures have also been shown to be sensitive to emotions and task demands (Boiten, Frijda, & Wientjes, 1994;
Wientjes, Grossman, & Gaillard, 1998). The mean heart rate (1/min) and the mean heart rate variability (HRV) were deter-
mined from the inter-beat intervals in the ECG signal. The HRV was calculated by dividing the standard deviation of the
inter-beat interval by the mean inter-beat interval (De Waard, 1996). Measures of cardiac response are indicative of task
demands (Backs et al., 2003; Cain, 2007).
2.8.5. Self-report measures
2.8.5.1. NASA TLX (0–100). The participants’ self-reported workload was assessed with the NASA TLX questionnaire (Hart &
Staveland, 1988) consisting of the following six items: mental demand, physical demand, temporal demand, performance,
effort, and frustration. The response scale for each of the six items consisted of 21 checkboxes with anchors on the left
(low), center (med), and right (high). For the performance item, the anchors good,med, and poor were used from the left
to right.
2.8.5.2. Confidence (0–100). The participants’ confidence was assessed using a confidence questionnaire consisting of the fol-
lowing six items: (1) ‘‘I understood how to negotiate the driving situations presented in the simulation”, (2) ‘‘Driving in this
environment was easy”, (3) ‘‘I performed well on driving the car (I was confident about my driving skills)”, (4) ‘‘I think I per-
formed better than the average participant in driving to the airport”, (5) ‘‘I had a feeling of risk during driving”, and (6) ‘‘I feel
confident to drive in similar conditions in the real world”. These items were inspired from previous questionnaires assessing
driver’s confidence (De Craen, 2010; De Groot, De Winter, López-García, Mulder, & Wieringa, 2011; Ivancic & Hesketh, 2000;
Wells, Tong, Genderton, Grayson, & Jones, 2008). The corresponding response scale consisted of 21 checkboxes with anchors
on the left (strongly disagree), center (neither agree nor disagree), and right (strongly agree).
2.8.5.3. Simulator discomfort and time pressure. The simulator discomfort experienced by the participants was assessed by the
following question: ‘‘I have experienced motion sickness in this experiment (general discomfort felt, in cars or boats, during
long trips)” on a five-point scale (1 = never,2=little,3=somewhat,4=much,5=very much). Furthermore, the sensation of
time pressure during the experiment was assessed with three questions: (1) ‘‘During driving I felt there WAS NOT enough
time to drive and arrive to the airport”, (2) ‘‘During driving I felt that I have to hurry up”, and (3) ‘‘How much time pressure
did you feel when driving?” on a five-point scale (1 = no pressure at all,2=a little pressure,3=moderate pressure,4=high
pressure,5=very high pressure). Finally, to assess the participant’s self-reported driving speed, participants were asked the
following question: ‘‘How fast did you drive in order to arrive at the airport?” (1 = not at all fast,2=a little fast,3=moderately
fast,4=fast,5=very fast).
2.9. Statistical analysis
Means and standard deviations were computed over the complete session, as well as for individual scenarios (e.g., car
following). Differences between sessions were statistically analyzed with paired ttests. The questionnaire results were frac-
tionally ranked (Conover & Iman, 1981) over all sessions and participants, because of their skewed distributions. Results
were declared significant if p< 0.001. This conservative alpha value was used to reduce the probability of Type I error, in light
of the large number of dependent measures. Correlations between the NTP and TP sessions were determined with the Pear-
son’s correlation coefficient. Additionally, because data may be sensitive to outliers, Spearman’s rank correlation coefficients
were calculated.
3. Results
Two of the 56 participants aborted the experiment because of simulator discomfort; these participants were excluded
from the analyses. For the remaining 54 participants (i.e., 54 NTP sessions and 54 TP sessions), 28.6% of the eye-tracking data
were removed because of data loss. For 7 of these 108 sessions (3 NTP sessions and 4 TP sessions), the data loss exceeded
60%. Therefore, the eye-tracking data for these 7 sessions were removed entirely.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 157
3.1. The effects of time pressure on the dependent measures
Table 3 shows the results for the training, NTP, and TP sessions for all driving performance, vehicle control, physiology,
gaze, and self-report measures. Furthermore, pvalues and effect sizes are tabulated for comparisons between T and NTP, T
and TP, and NTP and TP. Statistically significant differences between the NTP and TP sessions can be observed for all driving
performance measures. That is, consistent with Hypothesis 1 (i.e., the manipulation check of the effects of time pressure),
participants increased their speed during the TP session compared to the NTP session. Furthermore, drivers in the TP session
drove significantly closer to the lead car during car following, had lower driving precision (i.e., a higher SDLP), and had faster
control actions (i.e., an increase of steering speed, throttle variance, and number of brake operations). Specifically, in the TP
Table 3
Means and standard deviations of the dependent measures for the training (T), no time pressure (NTP), and time pressure (TP) sessions. pvalues and effect sizes
(in parentheses) are shown. The Pearson (r) and Spearman (
q
) correlation coefficients are shown for the NTP vs TP session (N= 54, but N= 50 for the gaze and
pupil diameter measures). Effect sizes were determined as Cohen’s d
z
=t/N
0.5
.
Dependent measures Session mean (SD)pvalue (d
z
) Correlation
Training (T) NTP TP T – NTP T – TP NTP – TP r
q
Driving performance
Completion time (s) 551.7 (51.9) 548.1 (38.3) 462.8 (34.6) n/a n/a <0.001 (2.83) 0.656 0.628
SDLP (m) 0.26 (0.09) 0.24 (0.08) 0.30 (0.14) 0.033 (0.30) 0.021 (0.33) <0.001 (0.49) 0.527 0.465
Mean lateral position (m) 0.22 (0.2) 0.16 (0.21) 0.05 (0.25) <0.001 (0.53) <0.001 (0.90) <0.001 (0.70) 0.764 0.763
Mean speed (m/s) 12.3 (1.2) 12.3 (0.9) 14.7 (1.1) 0.782 (0.04) <0.001 (1.98) <0.001 (2.74) 0.669 0.631
Max speed (m/s) 26.6 (3.3) 24.7 (2.4) 30.5 (3.6) <0.001 (0.65) <0.001 (1.07) <0.001 (1.83) 0.490 0.500
Minimum time headway (s) n/a 4.1 (2.9) 1.0 (0.6) n/a n/a <0.001 (1.09) 0.215 0.293
Vehicle control
Mean steering speed (deg/s) 8.02 (1.71) 7.63 (0.99) 8.36 (1.26) 0.070 (0.25) 0.116 (0.22) <0.001 (0.80) 0.684 0.755
Throttle variance (0–0.25) 0.051 (0.03) 0.043 (0.03) 0.107 (0.04) 0.009 (0.37) <0.001 (1.91) <0.001 (1.97) 0.669 0.664
Mean number of gear shifts (#) 65.3 (11.9) 63.9 (11.5) 59.4 (11.5) 0.142 (0.20) <0.001 (0.53) 0.002 (0.44) 0.602 0.617
Mean brake operations (#) 27.7 (5.2) 24.3 (5.4) 26.4 (5.2) <0.001 (0.59) 0.153 (0.20) 0.001 (0.47) 0.645 0.588
Max brake (0–1) 0.90 (0.04) 0.88 (0.07) 0.91 (0.02) 0.011 (0.36) 0.024 (0.32) <0.001 (0.57) 0.505 0.530
Mean acc. right hand (m/s
2
) 0.059 (0.07) 0.041 (0.07) 0.068 (0.08) 0.005 (0.42) 0.037 (0.30) <0.001 (0.63) 0.850 0.807
Mean acc. right foot (m/s
2
) 0.133 (0.04) 0.130 (0.04) 0.149 (0.04) 0.215 (0.18) <0.001 (0.76) <0.001 (0.91) 0.880 0.842
Mean acc. left hand (m/s
2
) 0.065 (0.05) 0.061 (0.05) 0.066 (0.05) 0.374 (0.13) 0.533 (0.09) 0.071 (0.25) 0.947 0.919
Mean acc. left foot (m/s
2
) 0.154 (0.06) 0.148 (0.06) 0.163 (0.07) 0.249 (0.17) 0.004 (0.42) 0.002 (0.45) 0.881 0.845
Gaze
Gaze road center (%) 55.4 (12.5) 52.5 (12.5) 55.8 (11.2) <0.001 (0.52) 0.738 (0.05) 0.056 (0.28) 0.696 0.777
Horizontal gaze variance (deg
2
) 60.8 (16.7) 68.8 (23.6) 76.9 (23.6) <0.001 (0.67) <0.001 (0.85) 0.003 (0.44) 0.723 0.778
Percentage dials (%) 11.0 (11.2) 11.5 (10.9) 10.1 (11.5) 0.428 (0.11) 0.143 (0.20) 0.035 (0.30) 0.906 0.742
Percentage clock (%) 1.48 (1.01) 1.31 (1.00) 2.25 (1.06) 0.120 (0.22) <0.001 (0.67) <0.001 (0.86) 0.423 0.346
Missing eye-tracker data (%) 26.7 (16.4) 29.1 (15.3) 28.4 (16.7) 0.039 (0.29) 0.152 (0.20) 0.674 (0.06) 0.849 0.787
Physiology
Mean blink rate (Hz) 0.24 (0.15) 0.29 (0.15) 0.24 (0.11) <0.001 (1.09) 0.077 (0.26) 0.001 (0.50) 0.852 0.833
Mean pupil diameter (mm) 5.38 (0.69) 5.19 (0.68) 5.37 (0.66) <0.001 (1.47) 0.565 (0.08) <0.001 (0.94) 0.956 0.951
Mean respiration rate (1/min) 20.6 (2.07) 19.0 (2.65) 20.3 (2.70) <0.001 (1.08) 0.097 (0.23) <0.001 (0.81) 0.840 0.847
Mean respiration amplitude (mV) 12.7 (3.18) 12.0 (2.73) 12.2 (2.71) 0.014 (0.35) 0.243 (0.16) 0.292 (0.15) 0.876 0.856
Mean heart rate (1/min) 82.1 (12.8) 79.2 (12.3) 82.1 (12.9) <0.001 (1.01) 0.924 (0.01) <0.001 (0.78) 0.956 0.954
Mean HRV (–) 0.069 (0.02) 0.068 (0.02) 0.071 (0.02) 0.391 (0.12) 0.249 (0.16) 0.021 (0.33) 0.844 0.803
Mean head position (m) 0.764 (0.05) 0.767 (0.05) 0.764 (0.05) 0.023 (0.33) 0.918 (0.01) <0.001 (0.50) 0.993 0.987
Subjective measures
Motion sickness (1–5) 1.89 (0.98) 1.70 (0.96) 1.70 (0.98) 0.032 (0.30) 0.077 (0.25) 1.000 (0.00) 0.921 0.894
Not enough time (1–5) 1.76 (0.87) 1.37 (0.56) 3.98 (0.79) <0.001 (0.50) <0.001 (1.88) <0.001 (2.42) 0.283 0.245
Feeling of hurry (1–5) 2.09 (0.92) 1.61 (0.74) 4.28 (0.76) <0.001 (0.56) <0.001 (1.84) <0.001 (2.41) 0.106 0.019
Fast arrival (1–5) 1.67 (0.87) 1.70 (0.74) 3.74 (0.85) 0.766 (0.04) <0.001 (1.69) <0.001 (1.93) 0.115 0.159
Time pressure (1–5) 1.50 (0.64) 1.35 (0.55) 3.69 (0.91) 0.146 (0.20) <0.001 (2.36) <0.001 (2.53) 0.262 0.234
Understanding (0–100) 84 (15) 87 (12) 78 (20) 0.115 (0.22) 0.054 (0.27) <0.001 (0.49) 0.466 0.604
Was easy (0–100) 66 (26) 78 (19) 63 (23) <0.001 (0.60) 0.320 (0.14) <0.001 (0.70) 0.417 0.477
Performed well (0–100) 61 (25) 72 (20) 61 (26) <0.001 (0.57) 0.936 (0.01) 0.001 (0.47) 0.478 0.441
Performed above average
(0–100)
46 (17) 53 (18) 46 (23) <0.001 (0.53) 0.943 (0.01) 0.005 (0.41) 0.656 0.603
Feeling of risk (0–100) 36 (27) 23 (21) 53 (28) <0.001 (0.67) <0.001 (0.64) <0.001 (1.08) 0.370 0.366
Confident (0–100) 84 (21) 85 (18) 65 (26) 0.817 (0.03) <0.001 (0.80) <0.001 (0.94) 0.580 0.626
Mental demand (0–100) 52 (21) 39 (23) 59 (20) <0.001 (0.71) 0.008 (0.38) <0.001 (1.22) 0.690 0.650
Physical demand (0–100) 27 (18) 23 (16) 38 (22) 0.033 (0.30) <0.001 (0.76) <0.001 (0.89) 0.613 0.672
Temporal demand (0–100) 28 (22) 20 (16) 71 (16) 0.012 (0.36) <0.001 (2.17) <0.001 (2.68) 0.290 0.220
Performance (0–100) 62 (26) 68 (25) 47 (25) 0.036 (0.30) 0.002 (0.45) <0.001 (0.62) 0.154 0.183
Effort (0–100) 44 (25) 33 (19) 64 (16) <0.001 (0.64) <0.001 (0.87) <0.001 (1.54) 0.341 0.304
Frustration (0–100) 30 (23) 21 (20) 49 (27) <0.001 (0.59) <0.001 (0.75) <0.001 (1.16) 0.503 0.474
Statistically significant effects (p< .001) are denoted in boldface.
158 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
session, drivers moved their limbs more rapidly, especially their right foot (which is used for operating the throttle) and their
right hand (which is used for changing gears). No statistically significant differences were observed regarding the activity of
the left hand, which is interpretable because the left hand serves no specific purpose on a road without curves.
The results show a significantly higher workload on each of the six items of the NASA TLX for the TP versus the NTP ses-
sion. Among the 17 self-report measures, the largest effects (|d
z
| > 2.0) of NTP versus TP were observed for the time-related
measures (i.e., not enough time, feeling of hurry, time pressure, and temporal demand). Thus, our time pressure manipula-
tion was successful in the sense that participants in the TP condition drove faster and experienced a greater feeling of hurry,
time pressure, and temporal demand than in the NTP condition.
Consistent with Hypothesis 2, participants exhibited physiological reactions that represent an increase of sympathetic
arousal. Statistically significant differences were observed for each of the physiological measures, except for the mean res-
piration amplitude and the mean HRV. The mean blink rate decreased, while the mean pupil diameter, mean respiration rate,
and mean heart rate increased from the NTP to the TP session. Additionally, drivers sat slightly (but statistically significantly)
closer to the steering wheel in the TP session compared to the NTP session. Table 3 further shows that the session-averaged
horizontal gaze variance and percentage road center were not significantly different for the TP session compared to the NTP
session. The time spent gazing at the in-vehicle clock increased when participants drove in the TP session compared to the
NTP session, most likely because the clock contained task-relevant information in TP condition.
The comparisons of physiological responses between the NTP and TP session had a medium effect size (d
z
= 0.5) for the
mean blink rate and a large effect size for the mean pupil diameter (d
z
= 0.9). These effect sizes were comparable to the effect
sizes of the vehicle control measures shown in Table 3.
Correlation coefficients between the NTP and TP sessions are shown in Table 3. Correlations were about 0.5–0.6 for the
vehicle control measures and about 0.8–0.9 for the physiological measures. Fig. 3 illustrates the correlation coefficient
between the NTP and TP sessions for the mean speed (left), mean blink rate (center), and mean pupil diameter (right). Sta-
tistically significant effects of the time pressure manipulation are visible in all three figures. Furthermore, Fig. 3 signifies that
the differences between individuals are substantially larger than the effects within individuals due to time pressure.
3.2. The relative validities of the physiological measures
A correlation matrix for the within-subject difference between the NTP and TP sessions is shown in Table 4 and Table S.1.
This correlation matrix shows a positive manifold among the mean speed, maximum speed, SDLP, mean steering speed,
throttle variance, mean pupil diameter, and the mental demands item from the NASA TLX. Thus, the mean pupil diameter
exhibits relative validity with respect to driving performance measures and self-reported mental workload.
Several of the correlations listed in Table 4 are illustrated in Fig. 4. The correlations of the pupil diameter and mean speed
(left), pupil diameter and mental demands (NASA TLX) (center), and pupil diameter and heart rate (right) are depicted. The
figures show that an increase in pupil diameter was moderately associated with an increase in driving speed, heart rate, and
mental demands (NASA TLX). For example, Fig. 4 (right) shows that people who showed a large increase in mean heart rate
generally also showed a large increase in mean pupil diameter.
3.3. The effects of time pressure during traffic scenarios
3.3.1. Physiological signals versus traversed distance along the route
Fig. 5 shows an overview of 11 selected measures as a function of traveled distance in the NTP and TP sessions. This figure
illustrates the difference between the NTP and TP session for the various types of scenarios along the route (see Table 2, for
an overview of the traffic scenarios).
Consistent with Table 3 and Hypothesis 1, drivers in the TP session drove with higher average speeds and throttle posi-
tions than they did in the NTP session. However, this was not the case during the car following scenarios, where the partic-
ipants were held up by a lead car that was driving at constant speed. It can also be seen that participants braked harder
10 12 14 16 18 20
10
12
14
16
18
20
Mean speed (m/s) - NTP Session
Mean speed (m/s) - TP
Session
0.1 0.2 0.3 0.4 0.5
0.1
0.2
0.3
0.4
0.5
Mean blink rate (Hz) - NTP Session
Mean blink rate (Hz) - TP
Session
44.5 55.5 66.5
4
4.5
5
5.5
6
6.5
Mean pupil diam. (mm) - NTP Session
Mean pupil diam. (mm) - TP
Session
Fig. 3. Correlation between mean speed (left, N= 54), blink rate (center, N= 54), and pupil diameter (right, N= 50) between the No time pressure (NTP) and
Time pressure (TP) sessions. The correlation coefficients are shown in Table 3.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 159
before intersections during the TP session than during the NTP session, which can be explained by their higher approach
speed and their attempt to brake late in order to prevent time loss. Fig. 5 also shows that limb movement occurred partic-
ularly when approaching and leaving intersections, associated with accelerating, braking, and gear changing.
Consistent with Hypothesis 2 (effects on physiological measures), Fig. 5 shows an overall increase in pupil diameter, res-
piration rate, and heart rate during the TP session compared to the NTP session. Fig. 5 also shows a slightly more forward
posture (indicated by a lower longitudinal head position). It can also be seen that the participants moved the head forward,
on average about 1 cm, when approaching an intersection.
Regarding Hypothesis 3, several strategies can be observed. First, participants showed a higher (i.e., more to the left of the
road) lateral position during car following for the TP session compared to the NTP session. This might represent a useful
strategy to be able to change lanes quickly as soon as the left lane is free from traffic, or a previously learned strategy to
signal to other road users that one is in a rush and wants to overtake the lead car (see e.g., Portouli, Nathanael, &
Marmaras, 2014, for the communicative strategies that drivers use in traffic). Second, the increase of horizontal gaze vari-
ance (HGV) when approaching the intersections, which was most pronounced during the TP session, indicates that partici-
pants widened or accelerated their visual search. This behavior might represent an increased lookout, similar to the fact that
participants adopted a more forward posture when approaching intersections.
In the following sections, we zoom in and describe the distance-based effects for three scenarios: obstacle overtaking,
intersection crossing, and car following.
3.3.2. Obstacle overtaking
Fig. 6 shows the pupil diameter, respiration rate, and lateral position for both the NTP and TP sessions during the
obstacle overtaking scenarios. Consistent with Hypothesis 2, participants showed an increased mean pupil diameter when
Table 4
Pearson correlation matrix (N= 54) for driving performance, physiological, and self-reported workload measures. Correlations per measure were determined by
taking the difference between the time pressure (TP) and no time pressure (NTP) sessions.
123456789101112131415
1. Mean speed (m/s)
2. Max speed (m/s) .39
3. SDLP (m) .43 .13
4. Mean steering speed (deg/s) .56 .39 .66
5. Throttle variance (0–0.25) .60 .57 .32 .51
6. Mean acc. right hand (m/s
2
).29 .33 .31 .48 .35
7. Mean acc. right foot (m/s
2
).45 .36 .35 .48 .55 .48
8. Mean pupil diameter (mm) .36 .26 .37 .39 .29 .14 .20
9. Mean blink rate (1/s) .06 .07 .21 .06 .05 .09 .01 .34
10. Percentage clock (%) .15 .02 .04 .10 .13 .06 .06 .00 .23
11. Percentage dials (%) .23 .47 .22 .28 .36 .11 .27 .13 .49 .26
12. Hor. gaze variance (deg
2
) .04 .01 .37 .34 .02 .15 .06 .05 .30 .17 .10
13. Mean heart rate (1/min) .04 .34 .06 .18 .26 .15 .18 .41 .01 .01 .13 .10
14. Mean respiration rate (1/min) .21 .28 .03 .15 .29 .10 .04 .40 .11 .02 .13 .05 .21
15. TLX mental demand (0–100) .23 .19 .28 .40 .28 .11 .15 .49 .09 .00 .05 .10 .12 .18
Note: Correlations of magnitude greater than .27 correspond to p< 0.05 and are in boldface.
-3 -2 -1 0 1 2 3
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Delta mean pupil diameter (mm)
Delta mean speed (m/s)
-50 -25 025 50
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Delta mean pupil diameter (mm)
Delta TLX - ment al demand (-)
-10 -5 0 5 10
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Delta mean pupil diameter (mm)
Delta mean heart rate (1/min)
Fig. 4. Correlation between session differences of pupil diameter and mean speed (left, N= 50), pupil diameter and mental demand (NASA TLX) (center,
N= 50), and pupil diameter and heart rate (right, N= 50). Session differences between the Training (T) and No time pressure (NTP) sessions are indicated by
unfilled markers and session differences between the Time pressure (TP) and NTP session are indicated by filled markers. Linear fits calculated from session
differences between the TP and NTP session are shown as gray lines.
160 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
0 1000 2000 3000 4000 5000 6000
0
20
Speed
(m/s)
0 1000 2000 3000 4000 5000 6000
0
0.5
1
Throttle
(-)
0 1000 2000 3000 4000 5000 6000
0
2
4
Lateral
position (m)
0 1000 2000 3000 4000 5000 6000
0
0.5
1
Brake
(-)
0 1000 2000 3000 4000 5000 6000
0
100
200
300
HGV
(deg
2
)
0 1000 2000 3000 4000 5000 6000
5
5.5
6
Pupil diameter
(mm)
0 1000 2000 3000 4000 5000 6000
15
20
25
Respiration
rate (1/min)
0 1000 2000 3000 4000 5000 6000
80
90
Heartrate
(1/min)
0 1000 2000 3000 4000 5000 6000
0.76
0.77
0.78
Longitudinal head
position (m)
0 1000 2000 3000 4000 5000 6000
-0.1
0
0.1
0.2
Acceleration
RH (m/s
2
)
0 1000 2000 3000 4000 5000 6000
0.1
0.15
0.2
Acceleration
RF (m/s
2
)
X-world (m)
Fig. 5. Means of 11 signals as a function of traveled distance for the No time pressure (NTP; black) and Time pressure (TP; red) sessions. The speed, lateral
position, throttle position, and brake position were determined using a spatial sliding window of 4 m. The horizontal gaze variance (HGV), pupil diameter,
respiration rate, heart rate, head position, and limb accelerations were determined using a temporal sliding window of 3 s. The intersections with and
without traffic are indicated by green and red shading, respectively. Car following situations are indicated by gray shading and the overtaking maneuvers
can be identified by lateral positions exceeding 2 m. (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.)
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 161
approaching the obstacle, both in the overtaking scenario with traffic in the opposing lane (top left figure) as well as in the
scenario without traffic in the opposing lane (bottom left figure). The respiration rate shows an increase prior the overtaking
maneuver and a decrease thereafter (middle two figures).
The lateral position shows that when traffic was present in the opposing lane, participants in the TP session initiated their
overtaking maneuver later compared to the NTP session (right top figure). However, when no traffic was present in the
opposing lane, the participants in the TP session initiated their overtake maneuver earlier compared to the NTP session (right
bottom figure). In the context of Hypothesis 3, this can be interpreted as a strategy to complete the task as quickly as possible
in a safe manner. That is, without traffic, it makes sense to change lanes early in order to minimize the traveled distance and
to maximize the smoothness of travel.
3.3.3. Intersection crossing
Fig. 7 shows the mean speed, mean brake position, mean horizontal gaze variance, and mean respiration rate for both the
NTP and TP sessions during the intersection-crossing scenarios. Consistent with Hypothesis 1 (manipulation check), TP
resulted in an increased driving speed, increased intersection approach speed, and faster acceleration after crossing the
intersection compared to the NTP session, both for the intersections with and without traffic. Consistent with Hypothesis
2 (physiological effects of time pressure), for both intersection types, the respiration rate was higher for the TP session than
for the NTP session. A distinct pattern can be observed here, with the respiration rate rising upon approaching the intersec-
tion (see also Fig. 5, demonstrating a distance-based synchrony of the TP and NTP sessions for several of the physiological
measures).
Drivers in the TP session braked later when there was no traffic on the intersection, and earlier when traffic was present
on the intersection, compared to the NTP session. This strategy can be explained as follows: If the approach speed is higher
and there is crossing traffic at the intersection, it makes sense to brake early, because one has to stop before the crossing
traffic. However, if there is no traffic, then braking is not required and deceleration has a negative effect on the overall mean
speed. The horizontal gaze variance increases in both intersection types during both the NTP and TP sessions, which indicates
that participants scanned the intersection before crossing the intersection. Furthermore, participants in the TP session ini-
tiated their visual search earlier while approaching the intersection compared to the NTP session, as can be seen in the
increasing horizontal gaze variance before the intersection for both intersection types. This altered visual scanning behavior
when approaching intersections may be a strategy (Hypothesis 3) to acquire a maximal amount of visual information, in an
attempt to minimize risk when crossing intersections with high speed.
Traffic in opposing laneNo traffic in opposing lane
-100 0 100
5
5.2
5.4
5.6
5.8
6
6.2
Distance from obstacle (m)
Pupil diameter (mm)
-100 0 100
18
19
20
21
22
23
Distance from obstacle (m)
Respiration rate (1/min)
-100 0 100
-2
0
2
4
6
Distance from obstacle (m)
Lateral position (m)
NTP
TP
-100 0 100
5
5.2
5.4
5.6
5.8
6
6.2
Distance from obstacle (m)
Pupil diameter (mm)
-100 0 100
18
19
20
21
22
23
Distance from obstacle (m)
Respiration rate (1/min)
-100 0100
-2
0
2
4
6
Distance from obstacle (m)
Lateral position (m)
NTP
TP
Fig. 6. Mean pupil diameter (left), mean respiration rate (center), and mean lateral position (right) for NTP (black) and TP (red) during overtaking
maneuvers with traffic approaching in the opposing lane (top) and without traffic in the opposing lane (bottom). In the right two figures, the black lines
indicate the lane boundaries and the dashed gray line indicates the road center line. The measures were averaged across all obstacle overtake maneuvers
and participants. The measures were determined using a spatial sliding window of 0.5 m. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)
162 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
3.3.4. Car following
Fig. 8 shows the distribution of the lateral position, time headway, and throttle position for the NTP and TP sessions dur-
ing the car following scenarios. During these scenarios, participants followed a lead car that had a constant speed. Fig. 8 (left)
shows a probability distribution function indicating that, during the TP session, participants drove more toward the left of
the lane than during the NTP session, possibly representing a strategy that prepares for overtaking or that signals to other
road users that he/she is in rush (see also above). Fig. 8 (center) illustrates the smaller time headway adopted by participants
in the TP session compared to participants in the NTP session. Fig. 8 (right) shows the throttle position for participants in
both sessions. Participants in the TP session more often applied full throttle than participants in the NTP session. At first
sight, this behavior seems to serve no functional purpose as the lead car’s speed was constant, but it may be a preparatory
strategy allowing participants to overtake as soon as the traffic in the adjacent lane is free.
In Fig. 9, a heat map showing the gaze distribution during car following illustrates the increased gaze tunneling of drivers’
gaze. A significant difference in percentage road center (t(49) = 4.25, p< 0.001) was found during the car following scenario
between the NTP session (M= 57.9%, SD = 14.3%) and the TP session (M= 66.7%, SD = 16.9%). The reduced gaze variance dur-
ing the car following scenarios is indicative of increased gaze tunneling during the TP session compared to the NTP session.
-100 0 100
0
5
10
15
20
25
Distance before intersection (m)
Speed (m/s)
-100 0 100
0
0.2
0.4
0.6
0.8
Distance before intersection (m)
Brake (-)
-100 0 100
0
50
100
150
200
Distance before intersection (m)
HGV (deg2)
-100 0 100
17
18
19
20
21
22
23
Distance before intersection (m)
Respiration rate (1/min)
NTP
TP
-100 0 100
0
5
10
15
20
25
Distance before intersection (m)
Speed (m/s)
-100 0 100
0
0.2
0.4
0.6
0.8
1
Distance before intersection (m)
Brake (-)
-100 0 100
0
50
100
150
200
Distance before intersection (m)
HGV (deg2)
-100 0 100
17
18
19
20
21
22
23
Distance before intersection (m)
Respiration rate (1/min)
NTP
TP
Fig. 7. Speed (far left), brake (left), horizontal gaze variance (right), and respiration rate (far right) before intersections without (top) and with traffic
(bottom) on the intersection. Color shaded regions indicate the intersecting lane, with green and red corresponding to intersections without and with
traffic, respectively. The measures were determined using a spatial sliding window of 0.5 m. Additionally, for each 0.5 m, the horizontal gaze variance was
determined using a temporal window of 3 s. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)
1 0.5 -0 -0.5 -1
0
0.05
0.1
0.15
0.2
0.25
Lateral error (m)
Probability density (-)
0 5 10 15 20 25
0
0.05
0.1
0.15
0.2
0.25
Time headway (s)
Probability density (-)
0 0.2 0.4 0.6 0.8 1
10
-3
10
-2
10
-1
Throttle (-)
Probability density (-)
NTP
TP
Fig. 8. Distributions of the lateral position (left), time headway (center), and throttle position (right, displayed on a logarithmic scale) during the car
following scenarios. The distributions were determined by grouping the data across all participants in 0.1 m, 0.5 s, and 0.01 bins for the lateral position, time
headway, and throttle, respectively.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 163
4. Discussion
This study explored the effects of time pressure on measures of driver physiology, driving performance, and vehicle con-
trol. We formulated three broad hypotheses: (1) When under time pressure, drivers show an increase of speed and an accel-
eration of control actions, (2) When under time pressure, drivers show increased signs of sympathetic arousal, that is,
increased physiological activity, and (3) When under time pressure, drivers demonstrate various strategic behaviors that
allow them to complete the driving task more effectively while minimizing the risk of crashing.
4.1. Hypothesis 1: Effects of time pressure on speed
Regarding the first hypothesis, it is concluded that the time pressure instructions clearly had the expected effect. Looking
at Table 3, the four largest effect sizes between NTP and TP (|d
z
|> 2.5) among the 44 dependent variables were observed for
(1) the task completion time itself, (2) the mean speed (which is highly correlated with the reciprocal of task completion
time), (3) the self-reported time pressure, and (4) the self-reported temporal demand. These observations indicate that a
driving simulator setup can elicit strong behavioral effects when drivers are exposed to a temporal constraint.
Various measures that are causally related to driving speed, such as throttle variance, activity of the right foot, maximum
brake position, and mean absolute steering speed, were also higher for TP compared to NTP. These effects can be explained
through classical mechanics. For example, when approaching an intersection with high speed and having to come to a stand-
still, a greater brake pedal pressure is required compared to when approaching with low speed. Similarly, when accelerating
to high speed, a high throttle position is a prerequisite.
Another expected finding was that the lane keeping precision was poorer in the TP session compared to the NTP session.
This indicates that a speed-accuracy tradeoff existed (cf. Szalma et al., 2008; Zhai et al., 2004). A reduction of lane keeping
precision is also consistent with results from, for example, Engström, Johansson, and Östlund (2005), who found increased
SDLP values when visual demands were increased by a secondary visual task.
4.2. Hypothesis 2: Effects of time pressure on physiological measures
Consistent with Hypothesis 2 and the stress model of Wickens et al. (2004), the time pressure ‘stressor’ resulted in
increased physiological activity such as increased heart rate, increased respiration rate, increased pupil diameter, and
decreased blink rate for participants in the TP session versus the NTP session. The strongest effects were observed for the
pupil diameter, the respiration rate, and the heart rate. Our findings of heart rate and respiratory rate are in line with pre-
vious transportation research on the effects of secondary tasks during driving. For example, Mehler et al. (2009) found
increased heart rates and respiration rates in 121 participants when performing an n-back mental task in a driving simulator
compared to a control condition without secondary task. Our results regarding the pupillary response and blink rates are
similar to the literature as well. For example, Recarte et al. (2008) found an increase in pupil diameter and a reduction of
blink rate when drivers performed a secondary visual task.
Participants directed their gaze more at the lead car during the TP session than during the NTP session. Such gaze tun-
neling occurs when participants attend to their primary task and disregard secondary tasks (Crundall, Shenton, &
Underwood, 2004; Hancock, 1989; Recarte & Nunes, 2003; Williams, 1988). In the present experiment, gaze tunnelling
might signal a strategy in which the participant focuses acutely on the car in front, explained by the closer following distance
which required drivers to be more alert on the behavior of the lead vehicle.
Although the effects for the different physiological measures were mostly similar to each other and in the expected direc-
tion, each of these measures has unique strengths and weaknesses. For example, consistent with previous research in mem-
ory tasks and arithmetic tasks (e.g., Beatty, 1982; Klingner, 2010; Marquart & De Winter, 2015), pupil diameter has the
advantage that it responds within a few tenths of second to changes in task demands, and that it can reach a peak dilation
10E-12
10E-10
10E-08
10E-06
10E-04
10E-02
Fig. 9. Heat map of gaze probability density during car following in the No time pressure (NTP; left) and Time pressure (TP; right) sessions, overlaid on a
screenshot of the simulator display. Gaze distributions were determined by aggregating gaze data from car following sections of all participants in one-by-
one degree bins and are shown on a logarithmic scale.
164 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
of 0.5 mm in as little as 1 s. It should be noted, however, that fluctuations in pupil diameter might be confounded by envi-
ronmental lighting and gaze direction (e.g., Beatty, 1982; Klingner, 2010). The heart rate has lower temporal sensitivity, and
therefore is less suitable for assessing the effect of scenarios. Specifically, the mean inter-beat interval is 0.75 s, and it takes at
least several beats to detect a change in heart rate (Jorna, 1992; Rowe, Sibert, & Irwin, 1998).
It is interesting that HRV, which has been said to be a valid index of time pressure (Nickel & Nachreiner, 2003), did not
decline under TP. We believe that there are two main issues with the use of HRV. First, many different operationalizations of
HRV exist (Task Force of the European Society of Cardiology & the North American Society of Pacing & Electrophysiology,
1996), such as frequency domain approaches, successive differences of inter-beat intervals, or average variability of inter-
beat intervals (as employed in the present study). Second, HRV is difficult to interpret unless the task is constant with time.
In our study, the task was dynamic, featuring various points along the route where the situation changed (e.g., car following,
coming to a standstill). Thus, using heart rate variability for a task that itself varies as a function of time creates difficulties in
interpretation.
One may argue that it is not surprising that time pressure elicited signs of sympathetic arousal. While this may be true,
what is unique in our research is that we synchronized a large number of physiological measures with measures of driving
performance and vehicle control. This allowed us to compare the differences between the NTP and TP sessions as a function
of traveled distance along the route. Furthermore, we assessed the magnitude of inter-individual and found strong correla-
tions between the two sessions (r> 0.8, see Table 3 and Fig. 3), indicating that the effects of time pressure should be inter-
preted as relative changes within individuals rather than changes on an absolute scale. Furthermore, we demonstrated the
relative validities of the physiological measures. For example, people who showed a greater increase in mean speed were
generally also those people who showed the greater increase in pupil diameter (Table 4).
4.3. Hypothesis 3: Effects of time pressure on driving strategy
Consistent with Hypothesis 3 and the model of Maule and Hockey (1993), drivers adapted their driving strategies to the
time constraint. For example, in the TP session, drivers drove more to the left of the right lane than in the NTP session, an
effect that was particularly pronounced during car following. Presumably, drivers were maintaining a lateral position closer
to the left lane in order to prepare for an overtaking maneuver. Another explanation is that participants drove to the left to
signal to the lead car that they were in a rush.
A second change in strategy was observed when participants overtook obstacles and when crossing intersections. Specif-
ically, in the TP session when no traffic was present in the opposing lane, participants made the overtaking maneuver earlier
than in the NTP session. A similar effect could be seen in the intersection scenario, where participants in the TP session
braked later compared to the NTP session when no traffic was present on the intersection, and braked earlier compared
to the NTP session when traffic was present on the intersection. Presumably, participants in the TP session used these strate-
gies to minimize the overall time to complete the session. Furthermore, at intersections, participants in the TP session
adopted a more forward seating posture and showed greater gaze variance, possibly in an attempt to scan the intersection
for oncoming traffic more rapidly before crossing it.
A third strategy was identified during car following, during which participants in the TP session showed a high throttle
activity despite the fact that the lead car speed was constant. Note that some of these behaviors are nonfunctional in the
driving simulator. For example, the other traffic did not adapt to the participant’s behavior in any way (and so gestures
or signaling did not have an effect). Collectively, these strategies under time pressure represent behaviors that may seem
irrational, but serve the higher-order purpose to complete the task quickly yet safely.
4.4. Implications of our research for driver assessment applications
The results of our experiment showed strong and statistically significant changes in physiological measures when drivers
experienced time pressure. These findings provide support for the potential implementation of physiological measurements
as estimators of driver’s task demands (see also Brookhuis & De Waard, 2010; Mehler et al., 2009). The mean pupil diameter,
mean heart rate, and mean respiration rate demonstrated a particularly strong effect as a function of time pressure. The
vehicle-derived performance measures also showed strong effects of time pressure, but exhibited a strong context depen-
dency. For example, during car following the participants’ driving speeds were about equal to the speed of the lead car. Fur-
thermore, our experiment illustrated that, on an absolute scale, within-subject effects of time pressure are small compared
to between-subject differences. This phenomenon implies that changes in driver state can only be detected at the individual
level (see also Brookhuis, De Waard, & Fairclough, 2003; Matthews, Reinerman-Jones, Barber, & Abich, 2014; Mulder,
Dijksterhuis, Stuiver, & De Waard, 2009). That is, physiological measures have to be corrected for individual differences if
they are to be used in real-time driver assessment applications.
4.5. Limitations
Our experiment is affected by several limitations. First, the order of the NTP and TP sessions was not counterbalanced
because the time constraint was determined on an individual basis. Thus, our protocol did not control for learning effects
and other types of carryover effects. However, this limitation can be countered, because the majority of the dependent
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 165
measures showed a decreasing trend from the training to the NTP session (which, indeed, can likely be attributed to learning
and acclimation) but an increasing trend from the NTP to the TP session. For example, the mean pupil diameter was 5.38 mm,
5.19 mm, and 5.37 mm in the T, NTP, and TP sessions, respectively. Such a U-shaped pattern across the three sessions is
apparent for all seven physiological measures, as illustrated in Fig. 10. This pattern of results suggests that the physiological
response is not a methodological artifact caused by time-on-task.
A second limitation is that 28.6% of eye-tracker data had to be removed from our analysis. Missing eye-tracker data of
around 30% are consistent with the literature (e.g., Ahlstrom, Victor, Wege, & Steinmetz, 2012). Because much of the data
loss occurred at random events (e.g., due to eye blinks), no systematic error is expected in our results.
A third limitation is that it is unknown which mental or physical mechanisms have caused the observed physiological
signals. It is likely that the effects in Table 3 are attributable at least partly to physical exertion. Indeed, the acceleration data
of the limbs (Table 3) unequivocally indicate that participants were more physically active in the TP session than in the NTP
session. This probably had an influence on the physiological outcomes, such as heart rate and respiratory rate, which are well
known to be a function of physical activity. Moreover, Table 3 showed that the physiological measures exhibit a U-shaped
pattern across the three sessions, while the self-reported time pressure exhibits a different pattern, with only a small effect
between the training session and the NTP session and a very large effect between the NTP and TP session. This differential
pattern suggests that the physiological measures do not capture time pressure per se. Indeed, different mental mechanisms
might be at play, such as mechanisms associated with mental workload, frustration, ‘time stress’, anxiety, and arousal. We
argue that it is impossible to uniquely identify these mechanisms and states based on the present physiological data (see also
Cacioppo & Tassinary, 1990; Cacioppo, Tassinary, & Berntson, 2000). Taking pupil diameter for example, and setting aside
issues of experimental control such as the fact that the pupil diameter is sensitive to environmental light conditions mod-
erated by biological age (Näätänen, 1992; Watson & Yellott, 2012), it has been established that ‘‘any sensory occurrence–
whether tactile, auditory, gustatory, olfactory, or noxious–evokes a pupillary reflex dilation”, and that ‘‘one should not
assume that pupillary reflex dilations occur only to external sensory events, because emotions, mental processes, increases
in intentional efforts, and motor output also produce systematic changes in pupillary diameter.” (Beatty & Lucero-Wagoner,
2000, pp. 145–146). In summary, our research showed that the time pressure manipulation (i.e., a time constraint and a vir-
tual passenger urging to hurry up) had clear physiological effects, but it is not possible to reverse this causality and identify
‘time pressure’ as a unique cognitive construct from the physiological recordings. It is likely that other types of stressors,
such as traffic complexity and secondary tasks, yield physiological effects that are indistinguishable from the effects that
were observed in this study.
A fourth limitation is related to driving simulator validity. Driving simulators have been shown to provide measures that
are strongly predictable of real world driving (Lee, Cameron, & Lee, 2003). Reimer (2009) demonstrated the relative validity
of physiological measures recorded both in a fixed base driving simulator and during a field study. However, driving simu-
lator validity remains an important limitation in our study, and more research is recommended to validate our findings on
the road.
A fifth limitation is that our participants were recruited from a technical university campus, and may be assumed to have
above average intelligence, spatial ability (Wai, Lubinski, & Benbow, 2009), and a specific interest in (driving simulator) tech-
nology. It has been argued that people in high-income societies, participants from university communities in particular, are
not representative of the general population (Henrich, Heine, & Norenzayan, 2010).
Fig. 10. Scatter plot of the Cohen’s d
z
effect size of Session 1 (Training session, T) versus Session 2 (No time pressure session, NTP) and the Cohen’s d
z
effect
size of Session 2 (NTP) versus Session 3 (Time pressure, TP). These results indicate that the measures of sympathetic arousal decreased from Session 1 to 2,
but increased from Session 2 (r=0.95, p= 0.001).
166 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
A sixth limitation is that our research focused exclusively on manual driving. In the foreseeable future, automated driving
technologies will be introduced on the roads, a development that entails new types of psychological questions (De Winter
et al., 2014; Fisher, Reed, & Savirimuthu, 2015; Young & Stanton, 2007). Specifically, the driving task is gradually changing
from manual control into supervisory control, placing pressures on drivers to monitor both the environment and in-vehicle
systems (Banks, Stanton, & Harvey, 2014; see also Warm, Parasuraman, & Matthews, 2008).
Acknowledgments
We thank Green Dino BV, the Netherlands, for their technical and software support. The research was supported by
Universidad EAFIT, the Dutch Ministry of Economic affairs through the projects ‘‘Driver Observation in Car Simulators”
(HTASI09004-DrivObs), ‘‘Advanced Driver Vehicle Interface in a Complex Environment” (partner, RAAK PRO-3-08, 2011-
2015, ADVICE), and by the Stichting voor de Technische Wetenschappen (Dutch Technology Foundation; STW), applied
science division of the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (Netherlands Organisation for Scientific
Research; NWO) and the Technology Program of the Ministry of Economic Affairs.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.trf.
2016.06.013.
References
Aarts, L., & Van Schagen, I. (2006). Driving speed and the risk of road crashes: A review. Accident Analysis and Prevention, 38, 215–224.
Addison, P. S. (2005). Wavelet transforms and the ECG: A review. Physiological Measurement, 26, R155–R199.
Ahlstrom, C., Victor, T., Wege, C., & Steinmetz, E. (2012). Processing of eye/head-tracking data in large-scale naturalistic driving data sets. IEEE Transactions
on Intelligent Transportation Systems, 13, 553–564.
Backs, R. W., Lenneman, J. K., Wetzel, J. M., & Green, P. (2003). Cardiac measures of driver workload during simulated driving with and without visual
occlusion. Human Factors: The Journal of the Human Factors and Ergonomics Society, 45, 525–538.
Banks, V. A., Stanton, N. A., & Harvey, C. (2014). Sub-systems on the road to vehicle automation: Hands and feet free but not ‘mind’ free driving. Safety
Science, 62, 505–514.
Beatty, J. (1982). Task-evoked pupillary responses, processing load, and the structure of processing resources. Psychological Bulletin, 91, 276–292.
Beatty, J., & Lucero-Wagoner, B. (2000). The pupillary system. In J. Caccioppo, L. G. Tassinary, & G. Berntson (Eds.), The handbook of psychophysiology (2nd ed.,
pp. 142–162). Hillsdale, NJ: Cambridge University Press.
Beck, K. H., Daughters, S. B., & Ali, B. (2013). Hurried driving: Relationship to distress tolerance, driver anger, aggressive and risky driving in college students.
Accident Analysis and Prevention, 51, 51–55.
Beck, K. H., Wang, M. Q., & Yan, A. F. (2012). Hurried driver dispositions: Their relationship to risky traffic behaviors. American Journal of Health Behavior, 36,
86–95.
Benson, L., III, & Beach, L. R. (1996). The effects of time constraints on the prechoice screening of decision options. Organizational Behavior and Human
Decision Processes, 67, 222–228.
Birch, L., Juul-Kristensen, B., Jensen, C., Finsen, L., & Christensen, H. (2000). Acute response to precision, time pressure and mental demand during simulated
computer work. Scandinavian Journal of Work, Environment & Health, 26, 299–305.
Boiten, F. A., Frijda, N. H., & Wientjes, C. J. (1994). Emotions and respiratory patterns: Review and critical analysis. International Journal of Psychophysiology,
17, 103–128.
Bongers, P. M., De Winter, C. R., Kompier, M. A., & Hildebrandt, V. H. (1993). Psychosocial factors at work and musculoskeletal disease. Scandinavian Journal
of Work, Environment & Health, 19, 297–312.
Brookhuis, K. A., & De Waard, D. (2010). Monitoring drivers’ mental workload in driving simulators using physiological measures. Accident Analysis and
Prevention, 42, 898–903.
Brookhuis, K. A., De Waard, D., & Fairclough, S. H. (2003). Criteria for driver impairment. Ergonomics, 46, 433–445.
Cacioppo, J. T., & Tassinary, L. G. (1990). Inferring psychological significance from physiological signals. American Psychologist, 45, 16–28.
Cacioppo, J. T., Tassinary, L. G., & Berntson, G. G. (2000). Psychophysiological science. In J. Caccioppo, L. G. Tassinary, & G. Berntson (Eds.), The handbook of
psychophysiology (2nd ed., pp. 3–26). Hillsdale, NJ: Cambridge University Press.
Cain, B. (2007). A review of the mental workload literature (Report RTO-TR-HFM-121-Part-II). Toronto, Canada: Defence Research and Development.
Cantin, V., Lavallière, M., Simoneau, M., & Teasdale, N. (2009). Mental workload when driving in a simulator: Effects of age and driving complexity. Accident
Analysis and Prevention, 41, 763–771.
Cnossen, F., Rothengatter, T., & Meijman, T. (2000). Strategic changes in task performance in simulated car driving as an adaptive response to task demands.
Transportation Research Part F: Traffic Psychology and Behaviour, 3, 123–140.
Cœugnet, S., Miller, H., Anceaux, F., & Naveteur, J. (2013). How do time pressured drivers estimate speed and time. Accident Analysis and Prevention, 55,
211–218.
Cœugnet, S., Naveteur, J., Antoine, P., & Anceaux, F. (2013). Time pressure and driving: Work, emotions and risks. Transportation Research Part F: Traffic
Psychology and Behaviour, 20, 39–51.
Collinson, J. (2014). A9 average speed cameras and HGV 50mph pilot monitoring – ‘‘beforemarket research. Retrieved from http://a9road.info/uploads/
publications/A9_Average_Speed_Cameras_and_HGV_50mph_Pilot_Monitoring_-_Before_Market_Research.pdf.
Conover, W. J., & Iman, R. L. (1981). Rank transformations as a bridge between parametric and nonparametric statistics. The American Statistician, 35,
124–133.
Cooper, P. J. (1997). The relationship between speeding behavior (as measured by violation convictions) and crash involvement. Journal of Safety Research,
28, 83–95.
Crundall, D., Shenton, C., & Underwood, G. (2004). Eye movements during intentional car following. Perception, 33, 975–986.
Crundall, D. E., & Underwood, G. (1998). Effects of experience and processing demands on visual information acquisition in drivers. Ergonomics, 41, 448–458.
De Craen, S. (2010). The X factor. A longitudinal study of calibration in young novice drivers. Doctoral dissertation : Delft University of Technology.
De Groot, S., De Winter, J. C. F., López-García, J. M., Mulder, M., & Wieringa, P. A. (2011). The effect of concurrent bandwidth feedback on learning the lane
keeping task in a driving simulator. Human Factors: The Journal of the Human Factors and Ergonomics Society, 53, 50–62.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 167
De Groot, S., De Winter, J. C. F., Wieringa, P. A., & Mulder, M. (2009). An analysis of braking measures. In Exploiting the possibilities of simulators for driver
training (pp. 4–12). The Netherlands: Delft. Retrieved from http://repository.tudelft.nl/assets/uuid:893379c9-0e8c-43d8-9f57-06149a778183/2013.11.
07_DeGroot.pdf.
De Waard, D. (1996). The measurement of drivers’ mental workload. Doctoral dissertation. Haren, The Netherlands: University of Groningen.
De Winter, J. C. F., Happee, R., Martens, M. H., & Stanton, N. A. (2014). Effects of adaptive cruise control and highly automated driving on workload and
situation awareness: A review of the empirical evidence. Transportation Research Part F: Traffic Psychology and Behaviour, 27, 196–217.
Dijksterhuis, C., Brookhuis, K. A., & De Waard, D. (2011). Effects of steering demand on lane keeping behavior, self-reports, and physiology. A simulator
study. Accident Analysis and Prevention, 43, 1074–1081.
Edland, A., & Svenson, O. (1993). Judgment and decision making under time pressure: Studies and findings. In O. Svenson & A. J. Maule (Eds.), Time pressure
and stress in human judgment and decision making (pp. 83–101). New York, NY: Plenum Press.
Elander, J., West, R., & French, D. (1993). Behavioral correlates of individual differences in road-traffic crash risk: An examination of methods and findings.
Psychological Bulletin, 113, 279–294.
Elvik, R., Christensen, P., & Amundsen, A. (2004). Speed and road accidents. An evaluation of the power model (TØI Report 740/2004). Oslo: Institute of
Transport Economics (TOI).
Engström, J., Johansson, E., & Östlund, J. (2005). Effects of visual and cognitive load in real and simulated motorway driving. Transportation Research Part F:
Traffic Psychology and Behaviour, 8, 97–120.
Evans, L., & Wasielewski, P. (1982). Do accident-involved drivers exhibit riskier everyday driving behavior? Accident Analysis and Prevention, 14, 57–64.
Fisher, M., Reed, N., & Savirimuthu, J. (2015). Misplaced trust? Engineering & Technology Reference.http://dx.doi.org/10.1049/etr.2014.0054.
Hancock, P. A. (1989). A dynamic model of stress and sustained attention. Human Factors: The Journal of the Human Factors and Ergonomics Society, 31,
519–537.
Hancock, P. A., & Szalma, J. L. (Eds.). (2008). Performance under stress. Aldershot: Ashgate.
Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. In P. A. Hancock & N.
Meshkati (Eds.), Human mental workload (pp. 139–183). North-Holland: Elsevier Science.
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33, 61–83.
Ivancic, K., & Hesketh, B. (2000). Learning from errors in a driving simulation: Effects on driving skill and self-confidence. Ergonomics, 43, 1966–1984.
Jorna, P. G. (1992). Spectral analysis of heart rate and psychological state: A review of its validity as a workload index. Biological Psychology, 34, 237–257.
Klingner, J. M. (2010). Measuring cognitive load during visual tasks by combining pupillometry and eye tracking. Doctoral dissertation: Stanford University
Klingner, J., Kumar, R., & Hanrahan, P. (2008). Measuring the task-evoked pupillary response with a remote eye tracker. In Proceedings of the 2008 Symposium
on Eye Tracking Research & Applications (pp. 69–72).
Kramer, A. F. (1991). Physiological metrics of mental workload: A review of recent progress. In D. L. Damos (Ed.), Multiple-task performance (pp. 279–328).
London: Taylor & Francis.
Lajunen, T., Karola, J., & Summala, H. (1997). Speed and acceleration as measures of driving style in young male drivers. Perceptual and Motor Skills, 85, 3–16.
Lee, H. C., Cameron, D., & Lee, A. H. (2003). Assessing the driving performance of older adult drivers: On-road versus simulated driving. Accident Analysis and
Prevention, 35, 797–803.
Marquart, M., & De Winter, J. C. F. (2015). Workload assessment for mental arithmetic tasks using the task-evoked pupillary response. PeerJ Computer
Science, 1, e16.
Martinussen, L. M., Lajunen, T., Møller, M., & Özkan, T. (2013). Short and user-friendly: The development and validation of the Mini-DBQ. Accident Analysis
and Prevention, 50, 1259–1265.
Matthews, G. (2002). Towards a transactional ergonomics for driver stress. Theoretical Issues in Ergonomics Science, 3, 195–211.
Matthews, G., Reinerman-Jones, L. E., Barber, D. J., & Abich, J. (2014). The psychometrics of mental workload multiple measures are sensitive but divergent.
Human Factors: The Journal of the Human Factors and Ergonomics Society, 57, 125–143.
Maule, A. J., & Hockey, G. R. J. (1993). State, stress, and time pressure. In O. Svenson & A. J. Maule (Eds.), Time pressure and stress in human judgment and
decision making (pp. 83–101). New York, NY: Plenum Press.
McKenna, F. P. (2005). Why do drivers break the speed limit? Behavioral research in road safety 2005: Fifteenth seminar. London: Department for Transport.
Retrieved from http://www.bv.transports.gouv.qc.ca/mono/0996518.pdf.
Mehler, B., Reimer, B., Coughlin, J. F., & Dusek, J. A. (2009). Impact of incremental increases in cognitive workload on physiological arousal and performance
in young adult drivers. Transportation Research Record: Journal of the Transportation Research Board, 2138, 6–12.
Miller, J. G. (1960). Information input overload and psychopathology. American Journal of Psychiatry, 116, 695–704.
Mulder, L. J. M., Dijksterhuis, C., Stuiver, A., & De Waard, D. (2009). Cardiovascular state changes during performance of a simulated ambulance dispatchers’
task: Potential use for adaptive support. Applied Ergonomics, 40, 965–977.
Näätänen, R. (1992). Attention and brain function. Hillsdale, NJ: Lawrence Erlbaum Associates.
Nickel, P., & Nachreiner, F. (2003). Sensitivity and diagnosticity of the 0.1-Hz component of heart rate variability as an indicator of mental workload. Human
Factors: The Journal of the Human Factors and Ergonomics Society, 45, 575–590.
Organisation for Economic Co-operation & Development (2006). Young drivers: The road to safety. Paris, France: OECD.
Owsley, C., Stalvey, B., Wells, J., & Sloane, M. E. (1999). Older drivers and cataract: Driving habits and crash risk. The Journals of Gerontology Series A: Biological
Sciences and Medical Sciences, 54, M203–M211.
Palinko, O., Kun, A. L., Shyrokov, A., & Heeman, P. (2010). Estimating cognitive load using remote eye tracking in a driving simulator. In Proceedings of the
2010 symposium on eye tracking research & applications (pp. 141–144).
Parker, D., Reason, J. T., Manstead, A. S. R., & Stradling, S. G. (1995). Driving errors, driving violations and accident involvement. Ergonomics, 38, 1036–1048.
Portouli, E., Nathanael, D., & Marmaras, N. (2014). Drivers’ communicative interactions: On-road observations and modelling for integration in future
automation systems. Ergonomics, 57, 1795–1805.
Prado Vega, R., Van Leeuwen, P. M., Vélez, E. R., Lemij, H. G., & De Winter, J. C. F. (2013). Obstacle avoidance, visual detection performance, and eye-scanning
behavior of glaucoma patients in a driving simulator: A preliminary study. PLOS ONE, 8, e77294.
Recarte, M. A., & Nunes, L. M. (2000). Effects of verbal and spatial-imagery tasks on eye fixations while driving. Journal of Experimental Psychology: Applied, 6,
31–43.
Recarte, M. A., & Nunes, L. M. (2003). Mental workload while driving: Effects on visual search, discrimination, and decision making. Journal of Experimental
Psychology: Applied, 9, 119–137.
Recarte, M. Á., Pérez, E., Conchillo, Á., & Nunes, L. M. (2008). Mental workload and visual impairment: Differences between pupil, blink, and subjective
rating. The Spanish Journal of Psychology, 11, 374–385.
Reimer, B. (2009). Impact of cognitive task complexity on drivers’ visual tunneling. Transportation Research Record: Journal of the Transportation Research
Board, 2138, 13–19.
Rendon-Velez, E., Horvath, I., & Van der Vegte, W. (2012). A pilot study to investigate time pressure as a surrogate of being in haste. In Proceedings of the
Ninth International Symposium on Tools and Methods of Competitive Engineering (pp. 393–406).
Riener, A., Ferscha, A., & Matscheko, M. (2008). Intelligent vehicle handling: Steering and body postures while cornering. In U. Brinkschule, T. Ungerer, C.
Hochberger, & R. G. Spallek (Eds.), Architecture of computing systems–ARCS 2008 (pp. 68–81). Berlin: Springer.
Robert, G., & Hockey, J. (1997). Compensatory control in the regulation of human performance under stress and high workload: A cognitive-energetical
framework. Biological Psychology, 45, 73–93.
Rogers, S. D., Kadar, E. E., & Costall, A. (2005). Gaze patterns in the visual control of straight-road driving and braking as a function of speed and expertise.
Ecological Psychology, 17, 19–38.
168 E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169
Rothengatter, T. (1988). Risk and the absence of pleasure: A motivational approach to modelling road user behavior. Ergonomics, 31, 599–607.
Rowe, D. W., Sibert, J., & Irwin, D. (1998). Heart rate variability: Indicator of user state as an aid to human-computer interaction. In Proceedings of the SIGCHI
conference on human factors in computing systems, Los Angeles, CA (pp. 480–487).
Sivak, M. (1996). The information that drivers use: Is it indeed 90% visual? Perception, 25, 1081–1089.
Szalma, J. A., Hancock, P. A., & Quinn, S. (2008). A meta-analysis of the effect of time pressure on human performance. In Proceedings of the human factors and
ergonomics society 52nd annual meeting (pp. 1513–1516).
Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996). Heart rate variability: Standards of
measurement, physiological interpretation, and clinical use. Circulation, 93, 1043–1065.
Taubman-Ben-Ari, O., Mikulincer, M., & Gillath, O. (2004). The multidimensional driving style inventory—Scale construct and validation. Accident Analysis
and Prevention, 36, 323–332.
Tran, C., & Trivedi, M. M. (2010). Towards a vision-based system exploring 3d driver posture dynamics for driver assistance: Issues and possibilities. In IEEE
intelligent vehicles symposium (IV) (pp. 179–184).
Treat, J. R., Tumbas, N. S., McDonald, S. T., Shinar, D., Hume, R. D., Mayer, R. E., et al. (1979). Tri-level study of the causes of traffic accidents (Report DOT HS 805
099). Washington, DC: National Highway Traffic Safety Administration.
Van der Hulst, M., Rothengatter, T., & Meijman, T. (1998). Strategic adaptations to lack of preview in driving. Transportation Research Part F: Traffic Psychology
and Behaviour, 1, 59–75.
Van Galen, G. P., & Van Huygevoort, M. (2000). Error, stress and the role of neuromotor noise in space oriented behavior. Biological Psychology, 51, 151–171.
Van Leeuwen, P. M., Gòmez i Subils, C., Ramon Jimenez, A., Happee, R., & De Winter, J. C. F. (2015). Effects of visual fidelity on curve negotiation, gaze
behavior and simulator discomfort. Ergonomics, 58, 1347–1364.
Van Leeuwen, P. M., Happee, R., & De Winter, J. C. F. (2013). Investigating the effect of a visual search task for simulator-based driver training. In Proceedings
of the 7th international driving symposium on human factors in driver assessment, training and vehicle design, Bolton Landing, NY (pp. 425–431). Retrieved
fromhttp://drivingassessment.uiowa.edu/sites/default/files/DA2013/Papers/065_vanLeeuwen_0.pdf.
Van Leeuwen, P. M., Happee, R., & De Winter, J. C. F. (2014). Vertical field of view restriction in driver training: A simulator-based evaluation. Transportation
Research Part F: Traffic Psychology and Behaviour, 24, 169–182.
Veltman, J. A., & Gaillard, A. W. K. (1996). Physiological indices of workload in a simulated flight task. Biological Psychology, 42, 323–342.
Victor, T. W., Harbluk, J. L., & Engström, J. A. (2005). Sensitivity of eye-movement measures to in-vehicle task difficulty. Transportation Research Part F: Traffic
Psychology and Behaviour, 8, 167–190.
Vogel, K. (2003). A comparison of headway and time to collision as safety indicators. Accident Analysis and Prevention, 35, 427–433.
Wai, J., Lubinski, D., & Benbow, C. P. (2009). Spatial ability for STEM domains: Aligning over 50 years of cumulative psychological knowledge solidifies its
importance. Journal of Educational Psychology, 101, 817–835.
Warm, J. S., Parasuraman, R., & Matthews, G. (2008). Vigilance requires hard mental work and is stressful. Human Factors: The Journal of the Human Factors
and Ergonomics Society, 50, 433–441.
Watson, A. B., & Yellott, J. I. (2012). A unified formula for light-adapted pupil size. Journal of Vision, 12, 1–16.
Wells, P., Tong, S., Genderton, B., Grayson, G., & Jones, E. (2008). Cohort II: A study of learner and new drivers. Main report (Report No. 81) (Vol. 1) London, UK:
Transport Research Laboratory, Department for Transport.
Wickens, C. D., Lee, J. D., Liu, Y., & Gordon-Becker, S. E. (2004). An introduction to human factors engineering (2nd ed.). Upper Saddle River, NJ: Pearson Prentice
Hall.
Wientjes, C. J., Grossman, P., & Gaillard, A. W. (1998). Influence of drive and timing mechanisms on breathing pattern and ventilation during mental task
performance. Biological Psychology, 49, 53–70.
Wikman, A. S., Nieminen, T., & Summala, H. (1998). Driving experience and time-sharing during in-car tasks on roads of different width. Ergonomics, 41,
358–372.
Williams, L. J. (1988). Tunnel vision or general interference? Cognitive load and attentional bias are both important. The American Journal of Psychology, 101,
171–191.
Wright, P. (1974). The harassed decision maker: Time pressures, distractions, and the use of evidence. Journal of Applied Psychology, 59, 555–561.
Young, M. S., & Stanton, N. A. (2007). What’s skill got to do with it? Vehicle automation and driver mental workload. Ergonomics, 50, 1324–1339.
Zhai, S., Accot, J., & Woltjer, R. (2004). Human action laws in electronic virtual worlds: An empirical study of path steering performance in VR. Presence:
Teleoperators and Virtual Environments, 13, 113–127.
E. Rendon-Velez et al. / Transportation Research Part F 41 (2016) 150–169 169
... 1. Mean speed (kph) and absolute vehicle acceleration (m/s 2 ) were calculated as a measure of task efficiency, driving style, and driving safety. 2. The mean absolute steering speed (deg/s) and steer steadiness (%, defined as the percentage of time the absolute steering speed was lower than 1 deg/s) were used to represent steering activity [23, 24]. The mean absolute throttle speed (%/s) was used to indicate throttle activity. ...
... The percentage of gaze directed at the road center tended to increase with increasing driving speeds, whereas the steering activity decreased (i.e., steering steadiness increased) with increasing driving speeds. These results are similar to results found in driving simulator studies [23, 24, 27] , and illustrates the relative validity of the measurements obtained with our experimental setup. Consistent with the literature, the secondary arithmetic task resulted in an elevated physiological response. ...
... Specifically, the secondary task resulted in increased heart rate, a finding consistent with Reimer [28] who found similar results when participants performed an n-back arithmetic task (see also [29]). Our results also illustrate that the heart rate response was relatively slow (Fig. 4) [24] . The respiration rate responded quickly to the elevated cognitive load as the participants initiated the countback task. ...
Conference Paper
Full-text available
Driver distraction is a leading cause of crashes. The introduction of in-vehicle technology in the last decades has added support to the driving task. However, in-vehicle technologies and handheld electronic devices may also be a threat to driver safety due to information overload and distraction. Adaptive in-vehicle information systems may be a solution to this problem. Adaptive systems could aid the driver in obtaining information from the device (by reducing information density) or prevent distraction by not presenting or delaying information when the driver's workload is high. In this paper, we describe an on-the-road evaluation of a real-time driver workload estimator that makes use of geo-specific information. The results demonstrate the relative validity of our experimental methods and show the potential for using location-based adaptive in-vehicle systems.
Article
Full-text available
Pupillometry is a promising method for assessing mental workload and could be helpful in the optimization of systems that involve human-computer interaction. The present study focuses on replicating the studies by Ahern (1978) and Klingner (2010), which found that for three levels of difficulty of mental multiplications, the more difficult multiplications yielded larger dilations of the pupil. Using a remote eye tracker, our research expands upon these two previous studies by statistically testing for each 1.5 s interval of the calculation period (1) the mean absolute pupil diameter (MPD), (2) the mean pupil diameter change (MPDC) with respect to the pupil diameter during the pre-stimulus accommodation period, and (3) the mean pupil diameter change rate (MPDCR). An additional novelty of our research is that we compared the pupil diameter measure with a self-report measure of workload, the NASA Task Load Index (NASA-TLX), and with the mean blink rate (MBR). The results showed that the findings of Ahern and Klingner were replicated, and that the MPD and MPDC discriminated just as well between the lowest and highest difficulty levels as did the NASA-TLX. The MBR, on the other hand, did not interpretably differentiate between the difficulty levels. Moderate to strong correlations were found between the MPDC and the proportion of incorrect responses, indicating that the MPDC was higher for participants with a poorer performance. For practical applications, validity could be improved by combining pupillometry with other physiological techniques.
Book
Full-text available
The world is a dangerous place and recent events have served to make it less safe. There are many arenas of conflict and even combat across the world. Such situations are the quintessential expression of stress; you stand in imminent danger and live with the knowledge that you may be attacked, injured or even killed at any moment. How do people perform under these conditions? How do they keep a heightened level of vigilance when nothing may happen in their immediate location for weeks or even months? What happens when the bullets actually start flying? How is it you distinguish friend from foe, and each from innocent bystanders when in immediate peril of your life? Can we design technology to help people make good decisions in these ultimately hazardous situations? To what degree does your membership in a team act to dissipate these particular effects? Can we generate sufficiently stressful field exercises to simulate these conditions and can we train and/or select those most able to withstand such adverse conditions? How will the next generation of servicemen deal with these inherent problems? These are the sorts of questions that Performance Under Stress addresses. This book is derived largely from a multiple-year, multiple university initiative (MURI) on stress and soldier performance on the modern, electronic battlefield. It involved leading researchers from many institutions who have brought their individual expertise to bear on these crucial, contemporary concerns. United by a common research framework, these groups attacked the issue from different methodological and conceptual approaches, ranging from traditional laboratory modeling and experimentation, to realistic simulations; from involved field exercises to personal experiences of actual combat conditions. The insights generated have been distilled and presented as a benchmark of current understanding and provide future directions for research in this arena. Although this work focuses on soldier stress and soldier performance, the principles that are derived extend well beyond this single application. Their findings can be applied to people facing the demands of the business world or research as much as to those who meet life or death situations, such as homeland security, first responders, and law enforcement personnel. © Peter A. Hancock and James L. Szalma 2008. All rights reserved.
Article
Full-text available
Since the publication of William James's (1890) Principles of Psychology, most of James's questions about the relation between physiological events and molar psychological or behavioral processes remain unanswered. The slow progress in using physiological signals (PSs) to address general psychological questions is due in part to problems in quantifying PSs in humans and to the way in which investigators have been thinking about the relation between PSs and psychological operations. A framework is provided to foster analysis of psychological phenomena based on PSs. Psychological operations and physiological responses are defined in terms of configural and temporal properties, and psychophysiological relations are conceptualized in terms of their specificity (e.g., one-to-one vs many-to-one) and their generality (e.g., situation or person specific vs cross-situational and pancultural). This model yields 4 classes of psychophysiological relations: (a) outcomes, (b) concomitants, (c) markers, and (d) invariants. The model specifies how to determine whether a psychophysiological relation is an outcome, concomitant, marker, or invariant, and it describes limitations in inferences of psychological significance based on PSs when dealing with each. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Chapter
The first Handbook of Psychophysiology was published more than four decades ago (Greenfield & Sternbach, 1972). Coverage in that Handbook emphasized the peripheral nervous system (PNS), an emphasis that many still identify with the term psychophysiology in accord with the history of psychophysiology. As is the case for physiological and other scientific fields, however, psychophysiology has changed dramatically since the appearance of its first Handbook. With the advent of new and powerful probes of the central nervous system (e.g., brain imaging techniques), there is an increased emphasis in the field on investigating the brain and central nervous system (CNS) as they relate to behavior. Investigations of elementary physiological events in normal thinking, feeling, and interacting individuals are commonplace, and new techniques are providing additional windows through which the neural events underlying psychological processes can be viewed unobtrusively. Instrumentation now makes it possible for investigators to explore the selective activation of discrete parts of the brain during particular psychological operations in normal individuals and patients. Transcranial magnetic stimulation has made it possible to stimulate or temporarily disable a region of the brain to study its role in cognitive operations, and studies of patients with lesions are becoming more precise both in their definition of the lesion and in their specification of behavior. Developments in tissue and blood assays, ambulatory recording devices, non-contact recording instruments, and powerful and mobile computing devices make it possible to measure physiological, endocrinological, and immunological responses in naturalistic as well as laboratory settings. New, powerful assays, including DNA genotyping, are now possible using minimally invasive or non-invasive procedures. With recent developments in molecular biology, behavioral genetics and epigenetics are becoming important new players in the field. However, the views from these windows are clear only because of the deliberate efforts of knowledgeable investigators. Knowledge and principles of physiological mechanisms, biometric and psychometric properties of the measures, statistical representation and analysis of multivariate data, and the structure of scientific inference are important if veridical information is to be extracted from biological and behavioral data. These are among the topics covered in depth in this Handbook. The field of psychophysiology has changed dramatically in other ways as well. The field used to be divided into distinct territories, typically defined by organ systems (e.g., cardiovascular, somatic), with relatively little integration across these systems.
Article
A physiological measure of processing load or "mental effort" required to perform a cognitive task should accurately reflect within-task, between-task, and betweenindividual variations in processing demands. This article reviews all available experimental data and concludes that the task-evoked pupillary response fulfills these criteria. Alternative explanations are considered and rejected. Some implications for neurophysiological and cognitive theories of processing resources are discussed.
Chapter
Research investigating the effects of time pressure on judgment and decision making has tended to conceptualize the effects of time constraints in one of two ways. One approach has considered the imposition of a deadline as a stressor, emphasizing the mediating role of changes in affective state in influencing cognition (Maule & Mackie, 1990; Svenson & dland, 1989). The second approach has assumed that time constraints are one among a number of task and environmental factors that are included in a cost/benefit determination of cognitive strategy (Payne, Bettman & Johnson, 1988; Payne, Johnson, Bettman, & Coupey, 1990). The aim of the present chapter is to critically evaluate these approaches and to consider them in the light of recent research on the effects of stress on cognition. To meet these aims, the chapter is structured in the following way. First, there is a brief discussion of a number of key methodological issues. This is followed by a brief and selective review of major approaches investigating the effects of time constraints on judgment and decision making Finally, there is a review of Variable State Activation Theory (VSAT) (Hockey, 1986; Hockey, Briner, Tattersall, & Wiethoff, 1989; Hockey & Hamilton, 1983) and an analysis of how it may provide a unifying approach to develop our understanding of the effects of time constraints on judgment and decision making.
Conference Paper
Novice drivers tend to direct their gaze to the road ahead and not scan the environment properly. This study investigated the training effectiveness of a visual search task in a driving simulator, aimed at increasing young drivers' spread of visual search. Two groups of inexperienced drivers were instructed to drive as accurately as possible in the center of the right lane in a self-paced driving task of four 6-min sessions in a rural environment. While driving, one group performed a visual search task, consisting of detecting and fixating on visual stimuli in the peripheral area. The stimuli were purple dots that faded in slowly and disappeared when fixated by the participant. After training, both groups drove a transfer session in an urban environment, in which various hazardous situations occurred. Results showed that both groups improved their lane keeping performance, whereas the training group became more proficient in the visual search task. However, in the transfer session no group differences were detected. In conclusion, despite improvements in visual search performance during a relatively short training period, the visual search training did not detectibly influence the spread of visual search of novice drivers during a post training transfer session.