Individual Differences in Driving-Related Multitasking.
ABSTRACT We conducted an experiment with 22 participants to investigate the effect of secondary task presentation style on driving-related performance. Prior to the experiment, participants were presented with three cognitive ability tests and answered an online survey consisting of the Domain-Specific Risk-Taking Scale (DOSPERT), the Driver Behaviour Questionnaire (DBQ), and some demographic questions. The participants then performed a 1-D tracking (primary) task which simulated longitudinal control of a car. They also performed a vowel counting secondary task (counting the number of vowels in a list of multiple letters) under a variety of conditions. These conditions combined different modalities (audio/visual), presentation styles (simultaneous/sequential), task complexity (the number of distractors), and list lengths. We discuss the experimental results in terms of the impact of individual differences, in risk tolerance and cognitive ability, on how the tasks were performed.
- SourceAvailable from: Christopher D WickensTransportation Research Record. 01/2007; 1.
- [Show abstract] [Hide abstract]
ABSTRACT: This paper describes a method for remote, non-contact point-of-gaze (POG) estimation that tolerates free head movements and requires a simple calibration procedure in which the subject has to fixate only on a single point. This method uses the centers of the pupil and at least two corneal reflections (virtual images of light sources) that are estimated from eye images captured by at least two cameras. Experimental results obtained with a prototype system that tolerates head movements in a volume of about 1 dm3, exhibited RMS POG estimation errors of approximately 0.6-1 degrees of visual angle. This system can enable applications with infants that, otherwise, would not be possible with existing POG estimation methods, which typically require multiple-point calibration procedures.Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 02/2007; 2007:4556-60.
- [Show abstract] [Hide abstract]
ABSTRACT: In order to provide a reliable measure of the similarity of uppercase English letters, a confusion matrix based on 1,200 presentations of each letter was established. To facilitate an analysis of the perceived structural characteristics, the confusion matrix was decomposed according to Luce’s choice model into a symmetrical similarity matrix and a response bias vector. The underlying structure of the similarity matrix was assessed with both a hierarchical clustering and a multidimensional scaling procedure. This data is offered to investigators of visual information processing as a valuable tool for controlling not only the overall similarity of the letters in a study, but also their similarity on individual feature dimensions.Perception & Psychophysics 06/1979; 25(5):425-31. · 1.37 Impact Factor
Paper for the 3rd International Conference on Driver Distraction and Inattention (DDI2013),
September 4-6, 2013, Gothenburg, Sweden
Individual Differences in Driving-Related Multitasking
Sachi Mizobuchi1&2, Mark Chignell1&2, David Canella2, Moshe Eizenman2
1 Vocalage Inc., Canada
2University of Toronto, Canada
We conducted an experiment with 22 participants to investigate the effect of secondary
task presentation style on driving-related performance. Prior to the experiment,
participants were presented with three cognitive ability tests and answered an online
survey consisting of the Domain-Specific Risk-Taking Scale (DOSPERT), the Driver
Behaviour Questionnaire (DBQ), and some demographic questions. The participants
then performed a 1-D tracking (primary) task which simulated longitudinal control of a
car. They also performed a vowel counting secondary task (counting the number of
vowels in a list of multiple letters) under a variety of conditions. These conditions
combined different modalities (audio/visual),
(simultaneous/sequential), task complexity (the number of distractors), and list lengths.
We discuss the experimental results in terms of the impact of individual differences, in
risk tolerance and cognitive ability, on how the tasks were performed.
With the adoption of increasingly complex information systems in passenger vehicles,
interface designers are being asked difficult questions concerning how large quantities
of information can be presented to the driver without creating unsafe levels of
distraction. A naïve view of Multiple Attentional Resource Theory (Wickens, 1980)
would suggest that since driving tasks are primarily visual in nature, then auditory
interfaces should be used to avoid overloading visuo-spatial attentional resources.
However, this issue is complicated by the existence of key differences in the ways that
humans attend to and process different sensory modalities. For example, audio is an
inherently "streaming" (time-dependent) medium. Thus the sequential acquisition of
auditory information may sometimes be less efficient than the more simultaneous
acquisition of visual information. Furthermore, sequential processing of speech may
create a working memory load, as appears to happen with some text-to-speech
interfaces (Strayer, 2012).
The research reported here sought to examine how people with different levels
of cognitive ability and different levels of awareness of, and tolerance towards, risk,
deal with the demand of carrying out a driving related task in the presence of a
distracting secondary task.
The research questions addressed in this paper were:
• How do cognitive ability and style of presentation of the secondary task affect
attention towards the primary task?
• How, if at all, do attitudes towards risk affect how much visual attention people
pay to the primary task, and does this effect depend on cognitive abilities?
This question was motivated by the assumption that individual differences in attitude
and behavior towards risky situations may affect the allocation of visual attentional
resources while driving. We also assumed that individual differences in cognitive ability
would affect multitasking performance, as we have found in our previous research
(Mizobuchi et al. 2011, 2012). Thus, we designed an experiment in which participants
performed dual-tasks with different secondary task conditions. Prior to the experiment,
we measured participants' risk tolerance (attitude and behavior towards risky situations)
and cognitive abilities, using methods explained in the next section.
Three Steps of the Experiment
1. Online questionnaire
Forty-four people (Male=30, Female=14, aged from 18 to 34, Mean=24.68, SD=4.46
years-old) were recruited through invitations that were posted on the University of
Toronto campus, sent to university e-mail lists, advertised on the Internet, and presented
to a class. These participants answered an online survey consisting of the Domain-
Specific Risk-Taking Scale (DOSPERT) and the Driver Behaviour Questionnaire
The DOSPERT, originally developed by Weber et al. (2002), is a psychometric
scale that assesses risk taking in five decision domains: financial (separately for
investing versus gambling), health/safety, recreational, ethical, and social. Respondents
to the instrument rate the likelihood that they would engage in domain-specific risky
activities. In this research, we used the revised DOSPERT with 30-items as presented
by Blais et al. (2006).
driving style and investigates the relationship between driving behaviour and accident
involvement. In this study, we used a version of the DBQ adapted for use in the United
States (Reimer et al., 2005), which was deemed to be more appropriate for our Canadian
The DBQ, originally developed by Reason et al. (1990), measures self-reported
following cognitive ability test and dual task experiment. The composition of this 22-
person sample was determined by the following constraints. First, respondents had to
express a willingness to participate further steps. Second we wanted the sample to be
relatively balanced for gender. The sample for the follow-on studies was chosen so that
it included at least 10 females. Subject to the first two constraints noted above, the final
set of 22 participants was chosen so that people had more extreme (high/low) scores on
the DOSPERT and DBQ. This created a group with contrasting risk tolerances intended
to make tests of risk-related effects more sensitive. The resulting sample had twenty-
two (Female=10, Male=12) participants between 18 and 33 years old (Mean= 25, SD=
4.6). All participants held a driver’s license or learner’s permit and were judged to be
proficient in English.
Twenty-two of the 44 respondents for the online questionnaire went on to do the
2. Cognitive ability test
Prior to the experiment, the 22 participants performed three cognitive tests to measure
their ability on three Central Executive (CE) functions (shifting, updating, and
inhibition). These three functions were chosen based on Miyake et al.’s (2000) findings
concerning mappings between tasks and CE functions. Tyey carried out various types of
cognitive test on 137 participants. Confirmatory factor analysis (CFA) and structural
equation modelling (SEM) showed that much of the variance in cognitive task
performance could be explained by three CE functions (inhibition, shifting, and
updating) that are moderately correlated with one another, but are also clearly separable.
We selected three tasks that were closely related to the tasks that Miyake et al found
were strongly related to inhibition, shifting, and updating respectively.
(1) Stroop test (Inhibition task): Six color-related words (‘black’, ‘white’, ‘yellow’,
‘orange’, ‘purple’, and ‘green’) were presented in one of the six corresponding font
colors individually and at random. There were 36 possible word-font color
combinations. On each trial, three color names (response alternatives) were presented in
black at the bottom of the display. The participant's task was to respond with the color
in which the stimulus word was written, by pressing a corresponding key. The three
response alternatives were mapped to the left arrow key, down arrow key, and right
arrow key, respectively.
(2) Color monitoring test（Updating task): Participants were shown blue, yellow and
red circles (8cm in diameter) one at a time for 500ms in randomized order with an inter-
stimulus interval of 2500ms. The task was to respond by pressing a down arrow key
when the third instance of each circle color was presented (e.g., after seeing the third
blue/yellow/red circle), which required participants to monitor and keep track of the
number of times each color had been presented. For example, if the sequence was ‘‘blue,
red, yellow, yellow, red, blue, yellow, blue, red’’ then the participant should have
responded to the third blue, yellow and red circle (italicized). In order for momentary
mental lapses to have less impact on task performance, the circle count for each color
was automatically reset to 0 if the participant made a key press for that color, and
participants were informed of this feature before starting the task. Prior to completing
the trial blocks, participants received a practice session, which continued until they
made 3 correct responses.
(3) Wisconsin Card Sort Test (WCST; Shifting task): In this task, four stimulus cards
were displayed as images on a computer screen. The objects on the cards could differ in
color, quantity, and shape. The participants were then shown an additional card and
were asked to choose which one of the four original cards conformed to the same
category as the additional card. The selection was made by clicking on one of the four
original cards with the mouse. As the classification rule was not provided to the
participants, they had to guess the rule. They did this based on the pattern of feedback
provided to them (“correct” or “incorrect”), after they chose one of the four cards to
match with the additional card. In this experiment, the classification rule changed after
10 correct responses under the rule. The task was finished when a participant completed
8 different rules or 128 trials, whichever came earlier. We used the number of
perseveration errors as the performance measure based on our previous research
(Mizobuchi et al., 2012)
3. Dual-task experiment
Participants performed a 1-D tracking (primary) task which simulated longitudinal
control of a car. They also performed a vowel counting secondary task (counting the
number of vowels in a list of multiple letters) under a variety of conditions. The various
conditions combined different modalities (audio/visual), presentation styles (sequential
/simultaneous), task complexity (the number of distractors), and list length.
Estimation of Distraction
Two methods were used for estimating the amount of visual distraction. The first
(direct) method was to observe the proportion (and duration) of time that people were
looking away from the primary task monitor. The second (indirect measure) was the
variability (standard deviation) of the throttle input for the primary task. Our reasoning
was that if people were visually focused on the primary task, then they would be able to
track and control the size of the rectangle more smoothly, leading to relatively
consistent throttle inputs. On the other hand, if they were distracted from the primary
task and returned to it to find the rectangle (almost) out of bounds then they were be
likely to make sudden changes in acceleration to remediate the situation. Thus
variability in throttle input might also be a proxy for visual distraction.
Information was presented visually on up to two LCD monitors and auditorily through
computer speakers. Data was collected using pedals from a Logitech Driving Force GT
game controller and with a standard US Keyboard. In addition, a two-camera Remote
Eye-Gaze Estimation (REGT) system (Guestrin et al. 2007) collected eye gaze data
throughout the experiment. The arrangement of the equipment is shown in Figure 1.
Figure 1 Experiment settings.
The experiment consisted of a pedal tracking (primary) task, and a simultaneously
performed vowel counting (secondary) task.
Primary task (Pedal tracking task)
We used the same pedal tracking task as in our previous research (Mizobuchi et al.,
2012). This task was designed to simulate the maintenance of inter-vehicle distance on
roadways, as originally proposed by Uno and Nakamura (2010). They reported that, in
comparison among a lateral tracking task with a steering-wheel, a longitudinal tracking
task with a foot pedal and a detect response task, performance of the pedal tracking task
was most sensitive to different levels of difficulties of both auditory and visual
secondary tasks. In the pedal tracking task, a target rectangle in blue (analogous to a car
ahead) and a frame-shaped area in yellow (the ideal following distance) were presented
on the primary monitor (M1). The target rectangle fluctuated based on a mixed sign
wave signal. As a participant tapped/released the foot pedal, the target rectangle
expanded/shrunk accordingly. The participants' goal was to keep the outer edge of the
target rectangle inside the yellow area by manipulating a single foot pedal. The target
rectangle turned to red when it went outside the yellow area as shown in Figure 2.
Figure 2 Pedal tracking task visualizations.
a lead vehicle driving at variable speed, the size (side length) of the target rectangle (D)
was defined by equation (1).
To simulate the control dynamics of adjusting accelerator input while following
equalled 0 km/h and dt was 0.1sec. Sf represented the fluctuating signal while Lt was a
percentage of the first order lag of the throttle opening. D was the second-order integral
of the difference between the fluctuating signal (corresponding to the acceleration of the
car in front) and the control signal (corresponding to the acceleration of one’s own car;
the first order lag of the throttle opening %). The fluctuating signal was generated from
a mixture of four sine waves.
Initially D0 was equal to half the width of the acceptable area (yellow area). V0
Secondary task (Vowel counting task)
In this task, participants were presented with a sequence of letters in list form (e.g.,
“AABA”) and instructed to count the number of vowels in the list before indicating
whether the total number of target items was odd or even by pressing a corresponding
key (“1” key for Odd and “2” key for Even). Participants were cued to respond either by
a change in stimulus appearance (visual conditions) or voice (audio condition).
The experiment consisted of twelve conditions, which varied according to
presentation style (the stimuli were presented sequentially in audio, sequentially in
visual, or simultaneously in visual), list length (4 vs 12 letters), and number of
distractors (“AB” vs “AIUCFM”). The capital letters “AIUCFM” were chosen based on
an analysis of audio confusability (Conrad, 1964; Hull, 1973) and visual (Townsend,
1971; Gilmore, Hersh, Caramazza, & Griffin, 1979). Each condition consisted of eight
trials. Within the twelve conditions, the combinations of presentation style, list length,
and number of distractors were randomly ordered.
The inter-item interval (for sequential conditions) was set to 1s and the inter-trial
interval was 2s. Participants had up to a maximum of 3s to select a modality and up to
7.5s to answer whether the total number of vowels was even or odd. Across all
conditions reaction times were recorded from the moment that the final stimulus was
Procedure and Incentive Structure
Participants were run in the experiment individually. First, they were introduced to the
primary (pedal tracking) task and given time to practice. Next, they were introduced to
the secondary (vowel counting) task and given time to practice each presentation style
with 4 item lists and with different numbers of distractors. Then, the tasks were
combined to allow practice with 12 item lists, in both modalities, with multiple
distractors (AIUCFM). After calibrating the Eye-Tracking System, the participants
performed both the primary and secondary task simultaneously under 12 counter-
balanced conditions that consists of eight trials each. During the experiment,
participants were encouraged to take short breaks. Each experiment session took
approximately 90 minutes for each participant. To promote engagement with both tasks
equally at all times, we told participants that they would be awarded based on combined
primary and secondary task performance with an emphasis on accuracy. Prior to the
experiment the research protocol was approved by the University of Toronto Research
The analysis looked at the relationship between risk perception, cognitive ability, and
performance of the primary and secondary tasks.
Cognitive Abilities and Eye Gaze
To examine whether there was a relationship between cognitive abilities and
presentation style (of the secondary task) on eye gaze patterns, we carried out two
stepwise entry regression analyses in which the dependent variables were (a) the mean
percentage of time that a participant viewed the primary display (Monitor 1: M1) in a
session (M1 gaze rate) and (b) the mean (per condition) maximum dwell time on the
secondary display (Monitor 2: M2) by each participant (M2_MaxDwell_mean). The
predictor variables in these analyses were CE ability test scores: accuracy and response
time of correct response (CRT) of a color monitoring test (updating); a Stroop test
(inhibition); the number of completed trials and the number of perseveration errors of
Figure 3 The relationship between updating ability and mean percentage of
time spent dwelling on Monitor 1 (M1Gazerate) for the each of the
three presentation styles.
looking at the primary monitor (M1) to increase with updating ability, this tendency was
most pronounced with the visual simultaneous presentation style, where the slope of the
linear fit was significant, F[1,18] = 9.54, p < .01 (Table 1). As can be seen in Figure 3,
Figure 3 shows that while there was a tendency for the proportion of time spent
simultaneous (versus sequential) visual presentation of information on a secondary task
generally facilitated visual attention to the primary task, and this benefit was greater for
people with higher updating ability. Table 1 shows a summary of the regression analysis
Table 1 Results from individual linear regression modelling of the effects of
updating ability measures on M1Gazerate.
Figure 4 The relationship between inhibition ability and mean percentage of
time spent dwelling on Monitor 1 (M1Gazerate).
monitor for participants with greater ability on a CE function, in this case inhibition
ability (note that lower reaction times in the inhibition task were indicative of increased
inhibition ability). As with updating ability, higher inhibition ability increased attention
to the primary task in the visual sequential condition F[1,18] = 7.14, p < .05, but in
contrast to the results for inhibition ability it also increased attention to the primary task
in the visual simultaneous condition, F[1,18] = 5.94, p < .05. Table 2 shows a summary
of results for the corresponding regression analyses.
Figure 4 shows a similar tendency for greater attention to the primary task
Table 2 Results from individual linear regression modelling of the effects of
Inhibition ability measures on M1 Gaze rate.
Pedal Tracking Performance
We examined the effect of the experimental conditions on the accuracy of the pedal
tracking task. First, we calculated the percentage of time when the target rectangle was
inside the acceptable area (target-in rate) as a dependent variable. The target-in rate was
very high in general, and likely due to this ceiling effect we did not find any significant
effect of conditions. Then we carried out analysis of the effect of presentation style of
the secondary task on the number of conditions (out of each of the 4 combinations of
list length and number of distractors in the secondary task) in which participants went
out of bounds. The result showed a significant effect of presentation style (F[2,
42]=3.60, p<.05). Planned comparisons showed a significant difference in the out-of-
the-bounds error rate between the visual-sequential and visual-simultaneous conditions
(Figure 5, p<.05). Thus the adverse effect of sequential presentation of the secondary
task (visually) was reflected not only in less visual attention to the primary task, but also
in lower task performance.
of-bounds errors (Error bars: 95% CI)
The number of conditions (out of 4) in which participants made out-
primary task, we also carried out stepwise entry regression analyses using the risk-
related measures as predictors:
• DOSPERT scores: total score
financial_investment, financial_gambling, financial_combined, health_safety,
and recreational items.
• DBQ scores: total score of all items, summed scores of errors, lapses, and
We carried out stepwise entry regression analyses in which the dependent
variable was the standard deviation of the throttle input, and the predictor variables were
the total DOSPERT and DBQ scores, as well as the measures of cognitive ability used
in the analyses reported above.
The best fitting regression model (p<.001) contained an inhibition ability
measure, Acc_inhib (s-beta=.66, p<.001), an updating ability measure, CRT_updating
(s-beta=.27, p<.05) and the DOSPERT_total risk measure (s-beta=-.40, p<.005). The
model explained 76% (r=.87) of variance in mean SD of pedal input. People who were
sensitive to risks and who had lower inhibition and updating ability operated the pedal
To further investigate the effect of individual differences on attention to the
of all items, summed scores of
in a smaller range, while people with higher ability in inhibition operated the pedal in a
larger range regardless of their DOSPERT score.
Secondary Task Performance
Since reaction times in sequential vs. simultaneous presentation conditions were not
comparable, only accuracy results are reported here.
Figure 6 shows the mean accuracy for the secondary tasks, with a generally high
level of accuracy being observed. Repeated measures within-participants 3 (presentation
style) x 2 (distractors) x 2 (list length) factorial ANOVA was conducted to test the
differences amongst conditions. The results showed a significant main effect of list
length (F[1,21] = 14.5, p<.01) and distractors (F[1, 21] = 21.6, p<.001) indicating lower
accuracy in conditions with longer lists and more distractors. There was also a
significant interaction between presentation types and list length (F[2, 42] = 3.5, p<.05)
indicating a stronger impact of list length on sequential conditions than on simultaneous
The Effect of the interaction between presentation style and list
length on Accuracy of the secondary task.
secondary task performance. We carried out stepwise entry regression analyses in which
the dependent variables were the mean accuracy of the secondary task, and predictor
variables were the same variables as used in the analysis in the preceding section.
The best fitting model (p<.005) contained WCST_Perrors (s-beta=-.66 p<.005)
DOSPERT_recreational (s-beta=.53, p<.005) and DBQ_lapses (s-beta=.38, p<.05) and
it explained 59% (r=.77) of the variance in mean accuracy of the secondary task. Thus,
people with higher shifting ability (i.e., a smaller number of perseveration errors) were
more accurate in the secondary task. Stronger sensitivity to risks and reported risky
behavior also predicted higher accuracy in the secondary task.
We also investigated the relationship between individual differences and the
The methodology used in this study was designed to examine the impact of distraction
due to a secondary task. As expected, visual presentation was significantly more
visually distracting (in terms of gazing away from the primary task monitor) than audio
presentation. Primary task performance (in terms of pedal-tracking accuracy) with
visual-simultaneous was neither better nor worse than that for audio presentation in both
the pedal tracking (primary) and secondary tasks. However, not only was visual
sequential presentation of the secondary task more distracting to the primary task, but
performance on the secondary task was also worse in the visual sequential condition.
Thus sequential presentation of visual information should probably be avoided when
adding information technologies in vehicles as it seems likely to have a damaging affect
not only on the secondary task but also on the primary (driving) task.
How do cognitive ability and style of presentation of the secondary task affect attention
towards the primary task?
We measured attention towards the primary task both in terms of the proportion of time
spent gazing at the primary monitor, and in the variability in throttle use. We would
expect that a person who was attending closely to the primary task would have a
relatively constant level of throttle input. On the other hand, someone who was more
distracted would tend to let the primary tracking task get closer to the extreme values
(bordering on being out of range) leading to greater variability in throttle input as larger
corrections were needed.
As expected, there was more visual distraction when the secondary task was
presented visually, and this affect was greater when the visual stimuli were presented
sequentially rather than simultaneously. This effect was moderated by updating ability,
with the secondary task tending to be less distracting (as measured by M1 gaze rate) for
people with higher updating ability, particularly when visual sequential presentation
was used. A similar moderating effect was found with inhibition ability, except that the
benefit of high inhibition ability in allowing a higher proportion of gazing on the
primary monitor was found for sequential, as well as simultaneous, visual presentation.
How, if at all, do attitudes towards risk affect how much visual attention people pay to
the primary task and does this effect depend on cognitive abilities?
We found that people who were measured as being sensitive to risks manipulated the
pedal in a smaller range and spent more time viewing the primary task monitor.
There was an interesting dissociation between the three central executive
functions in this study. While attention to the primary task was influenced by inhibition
and updating ability, the influence of shifting ability seemed to be largely limited to
secondary task performance in this study.
NHTSA Guidelines (2012) recommend that devices and tasks be designed so that they
do not require a glance duration of more than two seconds away from the road. In our
experiment, median glance duration was quite a bit shorter than that recommended
amount (ranging between 600 and 900ms.) However, distribution of glance duration
was positively skewed with the maximum glance duration on the secondary task
monitor in visual conditions being greater than 8 seconds in the worst case (v-
seq12AB). This should be taken into account when designing visual user interfaces for
in-vehicle systems. As discussed by Horrey and Wickens (2007), the unsafe conditions
that are likely to produce a motor vehicle crash reside not in frequently occurring
conditions (e.g., near the centre of the distribution), but rather in the tails of the
The take home message from this study is that people with higher cognitive
ability were able to attend to the primary task more, without negatively impacting their
performance on the secondary task. The visual sequential style was the most distracting
form of visual presentation, both in terms of reducing the amount of time spent looking
at the primary task monitor, and in terms of increase the amount of variability in
operating the primary task throttle. There was also a tendency for people with low risk
awareness to be more distracted by the secondary task.
While higher inhibition and updating ability seemed to help people with
selective attention i.e., focusing on the primary task), this increased attention to the
primary tasks did not result in lower secondary task performance. Instead, secondary
task performance was associated with shifting ability, with higher shifting ability
leading to better secondary task performance.
We are currently in the early stages of explicating the role of cognitive abilities
in determining how people react to distracting secondary tasks within vehicles. We
believe that the research reported in this paper has identifed some useful
presuppositions or hypotheses for future research. Secondary tasks are likely to be more
distracting for people with lower cognitive abilities and amongst those people, those
with a greater awareness of risk are more likely to maintain their focus on the primary
task. In the vowel counting task that we used sequential presentation of the letters in the
task was found to be harmful. Further research is needed with more complex (and
possibly more ecologically valid) tasks to see whether sequential presentation of
information in a secondary task is inherently more distracting or if the distraction is
specific to the kind of secondary task that we used in this study. The present results also
provide further evidence that it is necessary to consider the impact of different central
executive functions rather than ascribe differences in performance to more general
constructs such as working memory or mental workload. In the study reported here,
inhibition and updating ability affected the amount of attention paid to the primary task,
while shifting ability affected performance on the secondary task.
We would like to thank Sayaka Yoshizu, Chihiro Sannomiya, Kazunari Nawa, and
Junko Suzuki at Toyota InfoTechnology Center for their support on this research, We
would also like to thank Pierre Duez for developing software used in this study, Tiffany
Tong and Devan Hurst for assisting us with data collection.
Blais, A.R., and Weber, E. U. 2006. A Domain-Specific Risk-Taking (DOSPERT) scale for
adult populations. Judgment and Decision Making, 1, 33-47.
Conrad, R. 1964. Acoustic Confusions in Immediate Memory. British Journal of Psychology,
Gilmore, G. C., Hersh, H., Caramazza, A., and Griffin, J. 1979. Multidimensional Letter
Similarity Derived from Recognition Errors. Perception & Psychophysics, 25(5), 425-
Guestrin, E. D. and Eizenman, M. 2007. Remote point-of-gaze estimation with free head
movements requiring a single-point calibration, Proceedings of Engineering in Medicine
and Biology Society 2007, 4556 - 4560.
Horrey, W.J. and Wickens, C.D. 2007. In-vehicle glance duration: Distributions, tails and a
model of crash risk. Proceedings of the 86th Annual Meeting of the Transportation
Research Board. Washington, DC: National Academy of Sciences, 22-28.
Hull, A. J. 1973. A Letter-Digit Matrix of Auditory Confusions. British Journal of Psychology,
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H. and Howerter A. 2000. The unity
and diversity of executive functions and their contributions to complex ‘‘Frontal Lobe’’
tasks: A latent variable analysis. Cognitive Psychology, 41, 49–100.
Mizobuchi, S., Chignell, M., Suzuki, J., Koga, K., and Nawa, K. 2011. Central executive
functions likely mediate the impact of device operation when driving. In Proceedings of
the 3rd International Conference on Automotive User Interfaces and Interactive
Vehicular Applications, 129-136.
Mizobuchi, S., Chignell, M., Suzuki, J., Koga, K., and Nawa, K. 2012 The Impact of Central
Executive Function Loadings on Driving-Related Performance. Adjunct Proceedings of
AutomotiveUI '12, 68-75. Available at: http://www.auto-ui-cogload.unh.edu/wp-
content/uploads/2012/09/CLW2012-1.pdf [Accessed: 20 August 2013].
Mizobuchi, S., Chignell, M.H., Canella, D., Eizenman, M., Yoshizu, S., Sannomiya, C., Nawa,
K., and Suzuki, J. 2013. Looking or Listening? Impacts of Secondary Task Timing and
Difficulty on Tracking Performance and Modality Selection. To Appear in Proceedings
of the Human Factors and Ergonomics Society's 56th Annual Meeting. San Diego, CA.
NHTSA. 2012. Visual-Manual NHTSA Driver Distraction Guidelines for In-Vehicle Electronic
Devices. Available at: http://www.gpo.gov/fdsys/pkg/FR-2013-04-26/pdf/2013-
09883.pdf [Accessed: 20 August 2013].
Reason, J.T., Manstead, A.S.R., Stradling, S., Baxter, J., and Campbell, K. 1990. Errors and
violations on the roads. Ergonomics, 33, 1315–1332.
Reimer, B., D'Ambrosio, L., Gilbert, J., Coughlin, J., Biederman, J., Surman, C., Fried, R. and
Aleardi, M. 2005. Behavior Differences in Drivers with Attention Deficit Hyperactivity
Disorder: The Driving Behavior Questionnaire. Accident Analysis and Prevention, 37(6),
Strayer, D. 2012. Effects of Cell Phones on Driving Performance. Presentation for International
Conference on Distracted Driving. Available at:
http://www.distracteddriving.ca/english/program.cfm [Accessed: 20 August 2013].
Townsend, J. T. 1971. Theoretical Analysis of an Alphabetical Confusion Matrix. Perception &
Psychophysics, 19(1A), 40-50.
Uno, H., and Nakamura, Y. 2010 Study on surrogate laboratory techniques to assess driver
workload induced from voice-input interface. Proceedings of the JASE annual congress
(Fall), No.99-10, 1-6. (in Japanese)
Weber, E. U., Blais, A.R., and Betz, N. 2002. A domain-specific risk-attitude scale: Measuring
risk perceptions and risk behaviors. Journal of Behavioral Decision Making, 15, 263-290.
Wickens, C. D. 1980. The structure of attentional resources, in R. Nickerson (ed.), Attention and
Performance VIII (Hillsdale, NJ: Lawrence Erlbaum), 239-257.