Justin M. Owens, Richard Lehman
Franklin & Marshall College
Lancaster, Pennsylvania USA
E-mail:
JM_Owens@email.FandM.edu
Summary: The objective of this study was to
investigate the effects of driver distraction – both cognitive and visual – on
reaction time to unexpected road hazards. Participants operated a driving
simulator while intermittently answering prerecorded questions of various
difficulty (holding a “conversation” with the computer), or dialing specified
numbers into a cellular telephone. Two
road hazards were presented at unpredictable times and locations, including red
brake lights and a red pedestrian-shape of approximately the same area as the
brake lights. Targets were presented in
two different locations: directly in
front of the driver at the bottom of the screen, and off to the side of the
road. The results showed a significant overall increase in reaction time for
older subjects, as well as a strong interaction with the dialing task
condition. There were no significant differences from the control for either
easy or difficult verbal response conditions.
In addition, stimuli on the side of the road took significantly longer
to respond to, especially when combined with the dialing task. These data suggest a strong link between
age, visual task load, stimulus location, and increased reaction time to
unexpected stimuli.
INTRODUCTION
Today’s
society, with its almost universal fascination with computers and perpetual
information availability, seems eager to accept every new technology which
combines many functions into one. A
prime example of this is the current move to integrate an office environment
into automobiles; cutting edge technology is making it possible to allow an
ever-increasing array of functions to occupy a driver’s attention, including
but not limited to telephones, navigation systems, and even internet and email
capabilities. As in many cases of
technology, however, there may be a downside to counter the benefits. Here we refer explicitly to the effects of
divided driver attention; increasingly, drivers must divide their cognitive and
visual resources between talking on the telephone, reading a navigational map,
drinking coffee, adjusting the stereo, and, finally, operating their automobile
at highway speeds.
How does this affect drivers’ ability to
perform driving (their primary task) competently; specifically, how does all
this distraction (both cognitive and visual) influence drivers’ reaction times
to unexpected stimuli? Our purpose was
to quantitatively analyze the potentially dangerous effects of electronic
devices in cars; to this end, we have concentrated on the effects of mobile
telephones, which have become relatively commonplace in automobiles over the
past ten years.
Several studies
done in recent years have addressed this question, including two conducted by
Alm and Nilsson in 1994 and 1995. The
1994 study attempted to analyze differences in reaction time to stimuli between
groups of subjects who were performing a memory test while driving on a
simulator, and subjects who were doing simulated driving alone. The results
indicated that during the easy driving task, the division of attention between
driving and performing the memory task had a negative effect on the
experimental group, producing significantly longer reaction times than
participants in the control group.
There was no significant difference in reaction times between these groups
when driving the difficult course. The
authors suggest that while driving the difficult course drivers are in a more
aroused, frustrated state, and that they therefore pick up visual cues more
quickly (this is supported by research done by Moss & Triggs (1997)). They also suggest that as the road is more
difficult, drivers are paying more attention to the driving task to begin with
than to the dual task. The 1995 study
demonstrated increased reaction time for both age and task load. The current study focused on the effects of
gender, age, hazard type and location, distraction task load, and distraction
type (cognitive vs. visual). We
hypothesized that reaction time would not be affected by gender, but that there
would be a significant increase with age, and especially with increased task load
(none, easy and hard cognitive tasks, and visual tasks). In addition, it was hypothesized that
reaction times would increase for stimuli outside central visual focus; i.e.
stimuli appearing on the side of the road.
Participants
Participants
were of equal numbers male and
female college students (mean age 20.1 years) and middle aged members of the
college community (mean age 46.4 years).
There were 28 participants overall.
Apparati
The
driving simulator consisted of the Sony PlayStation program Gran Turismo II,
run off a PlayStation connected to a EIKI LC color video projector. The image was projected onto a 2 meter by
1.7 meter white screen approximately 8 meters from the projector. The screen formed a visual angle of 33°
23.9’ x 28° 4.3’. The test track was a
circuit approximately 3.2 kilometers in length, with easy to moderately
difficult left and right hand bends.
There were no other cars on the track.
Participants sat at a console approximately 3.3 meters from the
screen. The console contained a
steering wheel, with gas and brake pedals realistically positioned on the
floor. Visual stimuli and verbal
distracter instructions were presented by a program written for a Macintosh G3. The video feed from the computer was spliced
into the video from the PlayStation console by means of a TeleGen electronic
mixer, so that stimuli were overlaid on the test track.
There were two
types of visual targets presented in two locations. The first target was two round,
red circles, which simulated brake lights.
The second target was a stick-figure pedestrian; it too was red, and it
covered approximately the same area as the brake lights. The stimuli were
presented at either of two points on the screen: the first was in the center of the field of view at the bottom of
the screen, and the second was at the far right edge of the screen. The
computer was programmed to present these stimuli at seemingly random times (one
stimulus/location at a time), and to measure the time elapsed between initiation
of the stimulus and the time at which the brake pedal was depressed
halfway. This timer was accurate to
hundredths of a second. The stimuli
were programmed to appear an equal number of times in control and dual task
conditions, and during each type of task.
However, stimuli only occurred in 1/2 of the total tasks, to minimize
predictability.
Method
Participants
were instructed in how to operate the simulator, and were asked to drive as
though the test track were a highway with a speed limit of 55 mph. They were then given practice time driving
around the track. When participants mastered control of the simulator,
instructions were read and testing began. Approximately every 15 seconds, the
computer either asked the participant a question that required a verbal
response, or requested that a particular seven digit number be entered into the
cellular telephone to the participant’s right.
Stimuli were presented 2.5 s after the completion of the command during 50% of the trials.
An equal number of stimuli was presented between distractor trials.
RESULTS
This
study contained five independent variables and one dependent variable. The independent variables, as stated above,
were age, gender, stimulus location, stimulus type, and driving condition (control,
easy questions, difficult questions, and dialing). The dependent variable was reaction time, in hundredths of
seconds.
There
was a significant difference in reaction times between the age groups
in overall reaction time; F (1,761) =
6.75, p<0.05. There was no significant difference between
males and females, and no interaction between gender and age.
Results
from the dialing condition provided several significant findings. First, dialing the telephone produced
significantly longer reaction times for both age groups than the control
condition (for the young group p=0.01;
for the middle age group, p<0.001). Second, for the middle age group, reaction
times while dialing were significantly higher than during difficult question
answering, p<0.001. Third, there was an interaction between age
and task load, as the only significant difference between age groups came
during the dialing task (p=0.002). In the control condition, there was a
nonsignificant trend for the young age group to respond more quickly. The easy
question task showed similar results, with the middle age group taking slightly
longer to respond; again, this was a nonsignificant trend. Difficult questions
provided a larger young/middle age group reaction time gap, which approached
significance at p=0.07 Overall, however, there were no significant
differences in reaction time among age and the first three distracter tasks.
There was no
significant difference between age groups for the brake light stimulus, but the
young age group responded more quickly to the figure shaped stimulus (p<0.01)
than did the middle age group.
There
was no significant difference between the age groups for stimuli presented in
the middle of the road. However,
participants in the middle age group had significantly longer reaction times to
stimuli presented at the side of the road than did young participants; p=0.006. In addition, middle age participants had significantly higher
reaction times to the roadside stimuli than to stimuli in the center of the
road.
A
significant interaction occurred between Task and Stimulus Location [F(3,761)=67.20, p<0.0001]. In the
control condition, stimuli on the side of the road elicited significantly
longer reaction times than centrally presented stimuli (p=0.001). In addition, during the dialing task reaction times to
stimuli presented at the side of the road were significantly longer than
reaction times to central stimuli during dialing (p<0.0001). They were
also significantly longer than reaction times to side-presented stimuli during
all other distraction conditions (for all cases, p<0.001).
Age x Task x
Stimulus Location
There was a significant three-way interaction
between age, task load, and stimulus location [F(3,761)=15.10, p=0.003].
Post-hoc tests revealed several significant differences. First, in the control condition,
side-appearing stimuli elicited significantly longer reaction times for the
middle age group (p=0.002), but not
for the young group. Second, when answering difficult questions, participants
in the middle age group took significantly longer than participants in the
young group to react to stimuli at the side of the road (p=0.01). When reacting to
stimuli in the center of the road, however, there was no significant difference
between the age groups.
In the dialing condition, reaction times to
stimuli at the side of the road were the largest for both age groups of any distracter/location
combination. The young group had
significantly higher reaction times in the dialing/side condition than in the
control/side condition (p<0.0001). In addition, they had a significantly higher
reaction time during the dialing/side condition than to stimuli appearing in
the dialing/center condition. The
effect was even greater for participants in the middle age group; for them,
reaction times in the dialing/side condition were significantly greater than
the dialing/center (p<0.0001), as
well as for all other tasks and locations for either gender (for all, p<0.05).
These data show several interesting
trends, both consistent with and contrary to the hypotheses. In general, the hypothesis of increasing
difficulty in dual-task processing with increasing age and task complexity was
supported. However, not as much
evidence was found to support the hypothesis that cognitive distractions can
degrade reaction times on a level comparable to visual distractions.
Age proved to be a complex and rich
variable; it was significant not only as a main
effect, but in several interactions with other variables. This is very important, and cuts to the
central point of the experiment: the ability
of humans to divide their attention to performing two complex tasks seems to
suffer degradation with advancing age.
This is especially interesting when one considers that there were no
participants over the age of 54 in this study, and that the mean age for the “middle”
age group was 46.4. It would be fascinating to do a similar study on
retirement-age participants; however, it is possible that they would find a simulator quite difficult to operate.
The distraction conditions also provided a valuable source of
data. Of primary importance, and
counter to prediction, there was practically no difference in performance
between the control and question-and-answer conditions. Indeed, reaction times in general were not
significantly affected by even complex questions. There are two potential explanations for this fact. The first is simply that people, even into
their middle years, are quite good at thinking and speaking one thing while
looking for and reacting to another.
The second possibility is that there was a ceiling effect among the
questions conditions; although counterbalancing was used, there was still a
greater probability that stimuli would appear on any given trial than in the
real world.
Dialing, on the other hand, contributed
to significantly longer reaction times for both age groups. When combined with stimuli that were to the
side of the road, the effect became particularly prominent. Several times during the study, participants
(particularly those in the middle age group) completely failed to see roadside
obstacles while driving. When one
pauses to consider that this was an environment with high-contrast obstacles
that appeared in fully half of the trials, the real hazards of looking away
from the road become apparent.
Stimulus location was shown to significantly affect reaction time; in conjunction with the task and age variables, some interesting interactions occurred. In every distracter condition besides the Easy Question condition, reaction times were higher for stimuli appearing on the side of the road. These differences were significant in the Control, Difficult Question, and Dialing conditions for middle age participants, and in the Control and Dialing conditions for younger subjects. This discrepancy is interesting - it is important to remember that roadside targets appeared equally as often as central targets. Even with counterbalancing, there was far more likelihood of hazards appearing in this study than in an actual driving environment, and all stimuli had a high contrast with the surround road and scenery. At a speed of 55 m.p.h., the difference in mean reaction times (nearly a second) between middle and young conditions could cost drivers almost 65 feet. This effect is largely due to the influence of the dialing task. There was a very sizable interaction between dialing and stimuli location, especially for middle-age participants.
There seem to be two major trends at work here. First, in nearly all cases, the middle age
group had longer reaction times. This
difference grew as the difficulty of the task increased, especially when the
stimuli were presented at the side of the road. The second trend, related to the
first, was the tendency for reaction times in general to be longer for stimuli
at the side of the road, especially while dialing. For example, in the dialing/side condition for the middle age
group, reaction time was nearly double that
of the young age group, and was significantly different from every other
task/location combination for either age group. This effect was predicted and
is of course reasonable. Primarily, these data underscore the true
danger of unexpected hazards emerging from the roadside, in combination with
age and driver distraction.
REFERENCES
Alm, H., & Nilsson, L.
(1994). Changes in driver behaviour as
a function of handsfree mobile phones - a simulator study. Accident
Analysis and Prevention, 26, 441-451.
Alm, H., & Nilsson, L.
(1995). The effects of a mobile
telephone task on driver behaviour in a car following situation. Accident
Analysis and Prevention, 27, 707-715.
Lamble, D., Kauranen, T.,
Laakso, M., Summala, H. (1999). Cognitive load and detection thresholds in
car following situations: safety
implications for using mobile (cellular) telephones while driving. Accident
Analysis and Prevention, 31.
617-623.
Moss, S.A. & Triggs, T.J.
(1997). Attention switching time: A comparison between young and experienced
drivers. In Noy, Y.I. (Ed), Ergonomics
and safety of intelligent driver interfaces. 381-392. Mahwah, N.J. Lawrence Erlbaum Associates, Publishers.
Srinivasan, R., &
Jovanis. P. (1997). Effect of selected in-vehicle route guidance
systems of driver reaction times. Human Factors, 39. 200-215.