Individual Differences and Automation Choice in Simulated Driving

Abstract

Automated driving systems have the potential to lower driver workload, but may also induce fatigue and loss of situation awareness. This study investigated individual differences in fatigue response to automation that is under the driver's control. Of particular interest was the driver's choice of whether or not to use an automation option. The study included the Driver Stress Inventory (DSI), which assesses five personality traits that may control vulnerability to stress and fatigue: Aggression, Dislike of Driving, Fatigue Proneness, Hazard Monitoring, and Thrill Seeking. Participants were assigned to one of two experimental conditions, Automation Optional (AO) or Non-Automation (NA), and then performed a 35-minute simulated drive. Availability of automation failed to protect the driver from fatigue and distress; drivers in the AO condition actually showed poorer response times to a sudden traffic event. Individuals who chose to use automation in the AO condition showed higher levels of distress. DSI Fatigue-Proneness predicted higher levels of fatigue and stress, especially in automation-users. Thus, some individuals are especially prone to automation-induced fatigue, so that system designs should accommodate individual vulnerabilities.

Full text

INDIVIDUAL DIFFERENCES AND AUTOMATION CHOICE IN SIMULATED DRIVING
Catherine Neubauer, Gerald Matthews, Dyani Saxby, and Lisa Langheim
University of Cincinnati
Cincinnati, OH
Automated driving systems have the potential to lower driver workload, but may also induce fatigue and
loss of situation awareness. This study investigated individual differences in fatigue response to automation
that is under the driver's control. Of particular interest was the driver's choice of whether or not to use an
automation option. The study included the Driver Stress Inventory (DSI), which assesses five personality
traits that may control vulnerability to stress and fatigue: Aggression, Dislike of Driving, Fatigue
Proneness, Hazard Monitoring, and Thrill Seeking. Participants were assigned to one of two experimental
conditions, Automation Optional (AO) or Non-Automation (NA), and then performed a 35-minute
simulated drive. Availability of automation failed to protect the driver from fatigue and distress; drivers in
the AO condition actually showed poorer response times to a sudden traffic event. Individuals who chose to
use automation in the AO condition showed higher levels of distress. DSI Fatigue-Proneness predicted
higher levels of fatigue and stress, especially in automation-users. Thus, some individuals are especially
prone to automation-induced fatigue, so that system designs should accommodate individual
vulnerabilities.
INTRODUCTION
In recent years, numerous advancements have been
made in automated vehicle systems. In addition, driver stress
and fatigue are well known to be potentially dangerous during
vehicle driving (Hitchcock & Matthews, 2005). The seemingly
obvious solution to alleviate driver fatigue is to remove the
stressors that cause fatigue. In fact, development of more
automated vehicle systems might remove some of the task load
that can lead to fatigue. The purpose of these automated
systems is to improve performance of the human operator by
creating a less stressful, more workable environment, resulting
in optimal workload conditions. Examples of technologies that
may be found in future road transport systems include
automated highway system (AHS) and adaptive intelligent
cruise control (AICC) (Desmond & Hancock, 2001).
Automated highway systems consist of ‘smart’ road, with
driving tasks such as steering, braking, and speed controlled,
while adaptive intelligent cruise control emits a laser or radar,
which allows the vehicle to automatically slow or speed up
when needed.
These types of innovations might initially appear to be
a promising safety solution for mitigating fatigue effects,
because they reduce the prolonged workload placed upon
drivers. However, there are general concerns that automation
use may reduce situation awareness, as shown by evidence
suggesting a slowing of reaction time in drivers of automated
vehicles (Young & Stanton, 2007). In addition, the role of
individual differences in use of automation has not been much
studied in the transportation context. Potentially, personality
factors might influence how effectively the driver uses
automation as a means for regulating workload, stress and
fatigue. In addition, if automation does tend to reduce situation
awareness, drivers may differ in their susceptibility to loss of
alertness.
Studying individual differences in fatigue response is
challenging because fatigue itself is multifaceted and difficult
to define precisely (Desmond & Hancock, 2001). Studies based
on a multidimensional model of subjective states (Matthews et
al., 2002) have shown that fatigue manipulations may provoke
both loss of task engagement (e.g., tiredness) and elevated
distress (e.g., negative mood), depending on task demands
(Matthews & Desmond, 2002). Recent driving simulator
studies in our laboratory have been directed towards (1)
developing experimental methods for inducing fatigue states,
(2) exploring their consequences for performance and safety,
and (3) investigating the role of personality factors in fatigue
response.
Saxby et al. (2007, 2008) explored two qualitatively
different forms of fatigue, differentiated by Desmond and
Hancock (2001). Passive fatigue – produced by vehicle
automation – induced greater task disengagement than active
fatigue, associated with elevated workload for vehicle control
(induced by wind gusts). Active fatigue elicited higher distress.
It appears that the monotony of operating an automated vehicle
may be especially detrimental to subjective alertness and task
motivation. Further analyses of these data suggest that
individual differences in fatigue response on the simulator may
reflect variation in cognitive stress processes. For example,
drivers who appraise the task as lacking in challenge, and who
use high levels of avoidance coping, show lower task
engagement (Matthews, Neubauer, Saxby & Langheim, in
press).
Saxby et al. (2008) also investigated the consequences
for performance of fatigue states. Generally, studies using
driving simulators have identified a variety of behavioral
indices of fatigue, including deterioration in steering of the
vehicle and loss of attention to the external traffic environment
(Matthews & Desmond, 2002; Philip et al., 2003). Saxby et al.
(2008) tested the fatigued driver’s response to an emergency
event, a van pulling out in front of the driver unexpectedly.
Passive fatigue, induced through automation use, was
associated with slowed response times for braking and steering,
and a higher likelihood of crashing. The loss of alertness
Copyright 2011 by Human Factors and Ergonomics Society, Inc. All rights reserved DOI 10.1177/1071181311551326
PROCEEDINGS of the HUMAN FACTORS and ERGONOMICS SOCIETY 55th ANNUAL MEETING - 2011 1563

produced by passive fatigue may result from disruption of
effort-regulation, and failure to appropriately match effort to
task demands (Hancock & Warm, 1989; Young & Stanton,
2007).
The Saxby et al. (2007, 2008) studies demonstrate
potential dangers associated with automation, resulting from
passive fatigue. In these studies automation was imposed on
drivers, as might be the case in a future intelligent highway
system. However, many contemporary systems, such as AICC,
are under the individual driver’s control, such that drivers can
switch automation on or off at their discretion. The sense of
control may itself be beneficial, given that perceived control
helps mitigate adverse effects of stress (Michinov, 2005), and
may enhance attention (Dember, Galinsky & Warm, 1992).
Research investigating the role of control and automation found
that performance was better when participants were permitted
optional automation use (Harris et al., 1995). Furthermore,
drivers may differ in their ability to engage automated systems
most effectively. For example, perhaps some drivers are able to
use automation episodically in order to obtain a ‘rest break’
from some of the workload of driving so as to relieve fatigue.
Our previous studies have also investigated the role of
personality factors in fatigue vulnerability, factors which may
also relate to susceptibility to automation-induced
fatigue. There appear to be five personality traits linked to the
driving context that govern the individual’s vulnerability to
stress and fatigue. These are Aggression, Dislike of Driving,
Hazard Monitoring, Thrill Seeking and Fatigue Proneness, as
measured by the Driver Stress Inventory (DSI: Matthews,
2002). The DSI has been validated as a predictor of stress and
performance in both simulator and field studies, using samples
of professional and nonprofessional drivers (e.g., Desmond &
Matthews, 2009; Matthews, 2002). The first two traits are
predictive of various subjective states that correspond to anger
and anxiety, respectively. Hazard Monitoring indicates a coping
mechanism in which the driver continuously searches for signs
of danger and threat, while Thrill-Seeking reflects the
propensity to seek out dangerous situations.
Drivers scoring high on Fatigue Proneness experience
increased symptoms of fatigue and may be at an increased risk
from fatigue during both short and long drive durations. Indeed,
Fatigue Proneness predicts increased fatigue, relative to pre-
drive baseline in both simulated (Matthews & Desmond, 1998)
and real (Desmond & Matthews, 2009) driving. Rowden,
Matthews, Watson and Biggs (in press) found that, within an
occupational sample, Fatigue Proneness was associated with
higher self-reported driving errors, and with greater life stress.
Desmond and Matthews (2009) also found that Hazard
Monitoring related to resistance to fatigue states. These
personality factors may also influence fatigue response to
automation use.
The present study had two principal aims. The first
aim was to test whether individual differences in voluntary use
of automation were associated with fatigue state response. We
modified the paradigm used by Saxby et al. (2007, 2008).
Drivers were allocated to either an ‘automation optional’ (AO)
condition, affording choice, or a non-automation (NA)
condition representing normal driving. This design allows us to
test (1) whether simply making automation available alleviates
passive fatigue, and (2) whether – within the AO condition – it
helps those individuals who choose to use automation
experience less subjective fatigue and stress than non-users of
automation.
The second aim of this study was to investigate the
role of the DSI Fatigue-Proneness personality trait as an
influence on fatigue response in automated driving. Broadly,
similar to previous studies (Matthews et al., in press), we
expected Fatigue-Proneness to correlate with decreased post-
drive task engagement and increased distress in both AO and
NA conditions. We also hypothesized that fatigued drivers
would be more likely to use automation, in the AO condition, to
provide some respite from the task. We tested whether (1) the
Fatigue-Proneness trait, and (2) the task engagement state, was
associated with greater use of automation. We also tested –
again within the AO condition – whether Fatigue Proneness
moderated effects of automation use on subjective state
response.
The primary aims of the study were to investigate the
relationships between driver use of automation and fatigue
response, and the role of individual differences in Fatigue
Proneness. As a subsidiary aim we also tested whether the
effects on performance of the passive fatigue manipulation
reported by Saxby et al. (2008) were replicated.
METHOD
Participants
A total of 190 fully licensed drivers, from the
University of Cincinnati Introductory Psychology research pool
(80 males, 110 females) were recruited for the study.
Participants ranged in age from 18-30 (M = 21.28 years, SD
=3.13). All participants were required to have a valid driver’s
license and wear corrective lenses for those with a license
restriction B.
Materials and Apparatus
All groups participated in a simulated drive, which
lasted 35 minutes, on a Systems Technology, Inc., STISIM
Model 400 simulator, equipped with a 42” HD LCD video
monitor, which displayed the roadway and other elements of
the task. Controls included standard pedals and a steering wheel
that provided force feedback. The simulator is programmable to
create a variety of driving situations, which can produce stress
and fatigue symptoms in the driver.
Questionnaire Measures
Subjective stress states were assessed, before and after
the simulated drive, using the Dundee Stress State
Questionnaire (DSSQ: Matthews et. al., 2002). The pre-task
DSSQ assesses 11 dimensions of mood, motivation, and
cognition in performance settings. Scales are grouped into three
higher-order factors associated with task engagement (e.g.,
energy, task motivation, concentration), distress (e.g., tension,
low confidence) and worry (e.g., self-esteem, task-related
thoughts) symptoms. These three factors were analyzed for this
study. The post-task DSSQ assesses the same dimensions and
higher-order factors. It also includes an embedded version of
the NASA Task Load Index (NASA-TLX; Hart & Staveland,
PROCEEDINGS of the HUMAN FACTORS and ERGONOMICS SOCIETY 55th ANNUAL MEETING - 2011 1564

1998). The NASA-TLX is a standard measure of workload
based on ratings of task demands and subjective reactions to the
task that is widely used in human performance research. The
Driver Stress Inventory (DSI: Matthews, 2002) was used to
assess participants’ usual or typical feelings about driving. It
was scored to give the five stress vulnerability factors
previously described.
Procedure
Participants were individually tested and randomly
assigned to one of two experimental drives; Automation
Optional (AO) and Non-Automation (NA). Participants in the
AO condition had the option of initiating automation at their
discretion for periods of five minute blocks throughout the
course of the drive, while the NA group did not. The vehicle
was fully automated during these five minute intervals. In
addition to the drive, participants completed a series of
questionnaires used to assess trait and subjective states. Prior to
the drive, participants completed the DSI, which was used to
assess their stable predispositions to driving stress, followed by
the DSSQ, to attain baseline measurements of fatigue and
stress. Participants performed a 3 minute practice drive,
followed by the drive to which they had been assigned.
Following the drive, participants completed the post-task DSSQ
to assess changes in subjective state that may have resulted
from the simulated drive. High workload was achieved for both
conditions by programming a series of sharp curves and hills.
Performance was assessed during the final five minutes of the
drive, in which the AO group were obliged to control the
vehicle normally, without automation. During this interval, we
measured response times to an emergency event: a van pulling
out in front of the driver (similar to Saxby et al., 2008).
Response times were obtained for steering and braking. The
total experimental protocol lasted 1 ½ - 2 hours.
RESULTS
Four sets of analyses are summarized here. First, we
tested for effects of the drive on subjective state. Second, we
tested whether the choice to use automation was associated
with subjective state, within the AO group. Third, we
performed correlational and regression analyses to investigate
the role of stress vulnerability factors in state response. Fourth,
we tested for effects of automation on driver alertness in an
emergency situation.
Effects of Driving on Subjective State
State was assessed using the three DSSQ factors of
task engagement, distress and worry. For each state factor, we
ran a 2 (pre-post) x 2 (condition) ANOVA. ‘Pre-post’ was a
within-subjects factor contrasting pre- and post-drive state.
‘Condition’ was a between subjects factor contrasting the AO
and NA groups. Main effects of pre-post state were found for
Task Engagement, F(1, 185) = 88.86, p < .01, partial η2 = .33
and Distress, F(1, 185) = 160.70, p < .01, partial η2 = .47, but
not for worry. There were no effects of condition. Thus, the
drives were generally fatiguing and stressful, irrespective of the
availability of automation.
Effects of Automation Use on State
Next, we focused on differences in stress and fatigue
between users and nonusers of automation within the AO
group. Some of the variance in subjective state may be linked
to personality (stress vulnerability). Hence, to obtain a more
sensitive test of group differences, we ran ANCOVAs, in which
the DSI scales were included as covariates. The DSI scales
were unrelated to use or non-use of automation. Separate
ANCOVAs were run for each DSSQ dimension, including pre-
post as a within-subjects factor. The two critical tests here are
for the main effect of automation use, suggesting an overall
difference in state, and for the automation use x prepost
interaction, suggesting a time-dependent effect of automation.
In the analysis of task engagement, the main effect of
automation use was significant, F(1, 86) = 4.19, p < .05, partial
η2 = .05, but the interaction was not. For distress, it was the
automation use x prepost interaction that attained significance
F(1, 86) = 4.03, p < .05, partial η2 = .05, but not the main
effect. There were no effects of automation use on worry.
Figure 1. Pre-versus post-task engagement for users and non-
users of automation. Error bars are standard errors.
Figure 2. Pre-versus post-task distress for users and non-users
of automation. Error bars are standard errors.
Mean levels of engagement and distress in users and
nonusers of automation are shown graphically in Figures 1 and
2. Engagement was higher in non-users prior to the drive and
remained relatively high post-drive. Follow up t-tests
confirmed significant differences between groups both pre-
drive, t (92) = 2.05, p < .05, and post-drive, t (92) = 2.72, p <
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Pre-t as k Enga gemen t Pos t-t as k Enga ge men t
Task Engagement
Automation
Non-Users
Automation
Users
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Pre-t ask Dis tres s Pos t-tas k Dis tre s s
Distress
Automation
Non-Users
Automation
Users
PROCEEDINGS of the HUMAN FACTORS and ERGONOMICS SOCIETY 55th ANNUAL MEETING - 2011 1565

.01. It appears that lower engagement precedes the choice to
use automation. The group difference in distress is apparent
only in the post-drive data. There was no significant difference
between groups in pre-drive distress, but there was a significant
effect of use vs. nonuse of automation on post-drive distress, t
(92) = 2.76, p < .01. It seems that the using automated driving
increased distress.
Stress Vulnerability as a Correlate of Subjective State
Bivariate correlations between (1) the DSI scores and
(2) the post-drive DSSQ factors are shown in Table 1. The table
shows that the DSI predicted all three state factors. Fatigue
Proneness was associated with lower task engagement during
the drive, and also with elevated distress. Dislike of Driving
was most strongly associated with distress and worry, whereas
Hazard Monitoring correlated with higher engagement. Perhaps
surprisingly, Aggression was associated with mood
impairment, including higher distress and lower engagement.
Table 1 also gives partial correlations between DSI traits and
post-drive states, controlling for pre-drive states. The partials
suggest that Fatigue Proneness, Hazard Monitoring and
Aggression are associated with task-induced change in task
engagement.
Table 1.
Correlations between DSI and post-drive DSSQ scores.
Worry Task
Engagement
Distress
Correlations
Aggressive .272** -.246** .231**
Dislike of Driving .325** -.155* .355**
Hazard Monitoring -.005 .248** -.163*
Thrill Seeking .234** -.024 .100
Fatigue Prone .158* -.484** .439**
Partial correlations, controlling for pre-drive state.
Aggressive .035 -.208** .173*
Dislike of Driving .112 -.016 .103
Hazard Monitoring -.054 .058 -.155*
Thrill Seeking -.026 -.081 .055
Fatigue Prone .030 -.290** .344**
Note. * p < .05 ** p < .01
To further explore predictors of subjective state,
several multiple regressions were run. Fatigue-proneness was
significantly correlated with lower task engagement both pre-
drive (r = -.44, p<.01) and post-drive (r = -.48, p<.01). The trait
was also associated with higher distress pre-drive (r = .29,
p<.01) and post-drive (r = .44, p<.01). Correlation magnitudes
were similar in AO and NA conditions. Regression analyses
tested whether (1) fatigue-proneness moderated the task-
induced increases in fatigue and stress in the two experimental
conditions, and (2) whether fatigue-proneness moderated
effects of automation use on state within the AO condition.
For the first regression, post-drive task engagement
was the criterion. Four predictors were entered into the
equation in successive steps: pre-drive engagement, AO or NA
condition (effect-coded), fatigue-proneness (centered), and
condition × fatigue-proneness. The product term tests for the
interaction between fatigue-proneness and experimental
condition. A similar regression was run using distress in place
of engagement. Fatigue-proneness added to the prediction of
both task engagement (ΔR2=.048, p<.01) and distress
(ΔR2=.077, p<.01), but the interaction term was nonsignificant
in both cases. Fatigue-proneness predicted decreased
engagement and increased distress, with pre-drive states
controlled, but the effect was similar in both conditions.
Further regressions were run within the AO condition,
again with post-drive task engagement and distress as the
criteria. Four predictors were entered sequentially: pre- drive
(effect-coded), fatigue-proneness (centered), and automation
use × fatigue-proneness. Again, the linear fatigue-proneness
term predicted both decreased task engagement (ΔR2=.037,
p<.01) and increased distress (ΔR2=.083, p<.01). In addition,
the interaction term was significant for distress (ΔR2=.027,
p<.05). Thus, fatigue-proneness appears to depress engagement
irrespective of automation use. However, the effect of fatigue-
proneness on distress was accentuated in automation users. In
fact, additional regression analyses showed that fatigue-
proneness predicted post-drive distress, with pre-drive distress
controlled, in automation users (ΔR2=.096, p<.05), but not in
nonusers of automation (ΔR2=.010). Overall, results reveal that
the driver’s experience of the drive was substantially influenced
by the stress vulnerability traits measured by the DSI.
Performance
Mean response times to the emergency event for
participants in both conditions were measured. Independent
samples t-tests revealed a significant difference between groups
for the response times for steering t (150) = 2.06, p < .05 but
not for braking t (150) = .102, p > .05. Additionally, the NA
group showed faster response times for steering (M = 1.01, SD
= .82) and braking (M = 2.08, SD = 2.47), compared to the AO
group’s mean response times for steering (M = 1.45, SD = 1.60)
and braking (M = 2.10, SD = 1.96). Furthermore, mean
response times within the AO condition, comparing users and
non-users of automation were tested. Non-users of automation
were faster to respond to both braking and steering, compared
to users of automation, but the differences were not significant
for steering t (75) = 1.62, p > .05 or braking t (75) = -.285, p >
.05. We also tested whether DSI fatigue proneness and DSSQ
task engagement were correlated with steering and braking
response latencies. However, no significant associations were
found.
DISCUSSION
This study explored the effects of voluntary
automation use on subjective fatigue and stress states when
drivers were given the option of using automation in simulated
driving. More specifically, the study investigated the
associations between individual differences in automation use
and subjective state response. It also explored the effects of
passive fatigue on performance reported by Saxby et al. (2008).
Generally, the data confirmed previous findings (e.g., Saxby et
al., 2008) showing that high levels of fatigue (increased task
disengagement) and distress may be observed following
PROCEEDINGS of the HUMAN FACTORS and ERGONOMICS SOCIETY 55th ANNUAL MEETING - 2011 1566

automated driving, consistent with Desmond and Hancock’s
(2001) theory of passive fatigue. In addition, providing the
option to use automation does not in itself protect against these
adverse state changes.
Furthermore, it appears that choosing to use
automation relates to both the fatigue and stress responses. The
comparison of state changes in users and non-users of
automation, within the AO group, suggested more subtle inter-
relationships between automation use, stress, and fatigue. The
ANCOVAs showed that drivers who chose to use automation
were, on average, initially relatively low in task engagement
prior to the drive. Thus, fatigue may encourage the driver to use
automation. However, automation did not relieve fatigue in this
group. If anything, engagement dropped more rapidly in
automation users than in non-users. A different dynamic was
evident for distress. This stress state factor did not predict use
of automation, but distress was elevated in those who chose this
option. Thus, voluntary use of automation does not eliminate
the negative effects of passive fatigue (i.e., lower task
engagement), but in fact, seems to worsen them (i.e., elevating
distress). Thus, there may be a risk of a vicious cycle, in which
fatigue spurs the driver to use automation, which in turn
exacerbates distress.
There were several findings, which suggest that
personality factors may render drivers particularly vulnerable to
fatigue effects. Similar to previous findings (e.g., Desmond &
Matthews, 2009), drivers high in Fatigue Proneness showed
especially low levels of engagement during the drives, as well
as high distress. Drivers high in Dislike of Driving were prone
to the distress and worry states linked to anxiety. High
Aggression drivers also showed elevated distress and worry,
and reduced engagement, whereas high Hazard Monitoring
protected against fatigue. Fatigue Proneness did not influence
preference for automation. However, the trait moderated the
effect of automation in the AO group; individuals high in the
trait showed a greater elevation of distress following
automation use. These findings demonstrate a close connection
between individual differences in stress and fatigue
vulnerability. Those high in Fatigue Proneness may find the
experience of driving in monotonous conditions especially
frustrating. In these drivers, automation may serve to direct
their attention away from the driving task and onto their own
subjective discomfort, elevating distress.
We also replicated Saxby et al.’s (2008) finding that
steering response times to the emergency event were slower in
the AO than in the NA group, which is consistent with the
Desmond and Hancock (2001) theory that the effects of passive
fatigue, associated with automation use, may impair the
driver’s ability to actively monitor the traffic environment for
potential hazards.
The findings of this study have several practical
implications. First, they provide further support to those authors
who have identified possible dangers as well as benefits to
vehicle automation (e.g., Young & Stanton, 2007). Workload
reduction may interfere with the driver’s active engagement
with the task. At any rate, system designers need to evaluate
whether automation provokes fatigue and explore design
solutions that maintain driver engagement. Second, the data
show that people differ in their preferences for automation;
further research is needed on design features that render
automation more or less attractive to different individuals. Our
findings also show that individuals may make counter-
productive choices in use of automation, which exacerbate
fatigue, perhaps suggesting a need for training in fatigue
management. Lastly, the data suggest that questionnaire
measures may be used to identify drivers who are especially
fatigue-prone. Such measures may be used in selection of
professional drivers vulnerable to passive fatigue, such as long-
haul truck drivers.
REFERENCES
Dember, W.N., Galinsky, T.L., & Warm, J.S. (1992). The role of choice in
vigilance performance. Bulletin of the Psychonomic Society, 30,
201-204.
Desmond, P.A., & Hancock, P.A. (2001). Active and passive fatigue states. In
Hancock, P.A. & Desmond, P.A. (Eds.), Stress, workload, and
fatigue (pp. 455-465). Lawrence Erlbaum, Mahwah: NJ.
Desmond, P.A., & Matthews, G. (2009). Individual differences in stress and
fatigue in two field studies of driving. Transportation Research Part
F: Traffic Psychology and Behaviour, 12, 265-276.
Hancock, P.A., & Warm, J.S. (1989). A dynamic model of stress and sustained
attention. Human Factors, 31, 519-537.
Harris, W.C., Hancock, P.A., Arthur, E.J., & Caird, J.K. (1995). Performance,
workload, and fatigue changes associated with automation. The
International Journal of Aviation Psychology, 5, 169-185.
Hart, S. G., & Staveland, L. E. (1988). Development of a multi-dimensional
workload scale: Results of empirical and theoretical research. In
Hancock, P.A. & Meshkati, N. (Eds.), Human mental workload, (pp.
139-183). Amsterdam: North-Holland.
Hitchcock, E.M., & Matthews, G. (2005). Multidimensional assessment of
fatigue: A review and recommendations. Proceedings of the
International Conference on Fatigue Management in Transportation
Operations, Seattle, WA, September 2005.
Matthews, G. (2002). Towards a transactional ergonomics for driver stress and
fatigue. Theoretical Issues in Ergonomics Science, 3, 195-211.
Matthews, G., & Desmond, P.A. (2002). Task-induced fatigue states and
simulated driving performance. Quarterly Journal of Experimental
Psychology: Human Experimental Psychology, 55, 659-686.
Matthews, G., Campbell, S.E., Falconer, S., Joyner, L., Huggins, J., Gilliland,
K., Grier, R., & Warm, J.S. (2002). Fundamental dimensions of
subjective state in performance settings: Task engagement, distress
and worry. Emotion, 2, 315-340.
Matthews, G., Neubauer, C., Saxby, D., & Langheim, L. (in press). Driver
fatigue: The perils of vehicle automation. In. M. Sullman & L. Dorn
(Eds.), Proceedings of the 27th International Congress Of Applied
Psychology. Aldershot, UK: Ashgate Publishing.
Michinov, N. (2005). Social comparison, perceived control, and occupational
burnout. Applied Psychology: An International Review, 54, 99-118.
Philip, P., Sagaspe, P., Taillard, J., Moore, N., Guilleminault, C., Sanchez-
Ortuno, M., Akerstedt, T., & Bioulac., B. (2003). Fatigue, sleep
restriction, and performance in automobile drivers: A controlled
study in a natural environment. Fatigue, Sleep Restriction, and
Performance, 26, 277–280.
Rowden, P., Matthews, G., Watson, B., & Briggs. H. (in press). The relative
impact of occupational stress, life stress, and driving environment
stress on driving outcomes. In. Accident, Analysis and Prevention.
Saxby, D.J., Matthews, G., Hitchcock, E.M., & Warm, J.S. (2007).
Development of active and passive fatigue manipulations using a
driving simulator. Proceedings of the Human Factors and
Ergonomics Society, 51, 1237-1241.
Saxby, D.J., Matthews, G., Hitchcock, E.M., Warm, J.S., Funke, G.J., &
Gantzer, T. (2008). Effects of active and passive fatigue on
performance using a driving simulator. Proceedings of the Human
Factors and Ergonomics Society, 52, 1252-1256.
Young, M. S., & Stanton, A. (2007). Back to the future: Brake reaction times
for manual and automated vehicles. Ergonomics, 50, 46-58.
PROCEEDINGS of the HUMAN FACTORS and ERGONOMICS SOCIETY 55th ANNUAL MEETING - 2011 1567