Voluntary- and Involuntary-Distraction Engagement: An Exploratory Study of Individual Differences

Abstract

Objective

The aim of this study was to explore individual differences in voluntary and involuntary driver-distraction engagement.

Background

Distractions may stem from intentional engagement in secondary tasks (voluntary) or failing to suppress non-driving-related stimuli or information (involuntary). A wealth of literature has examined voluntary distraction; involuntary distraction is not particularly well understood. Individual factors, such as age, are known to play a role in how drivers engage in distractions. However, it is unclear which individual factors are associated with voluntary- versus involuntary-distraction engagement and whether there is a relation between how drivers engage in these two distraction types.

Method

Thirty-six participants, ages 25 to 39, drove in a simulator under three conditions: voluntary distraction with a self-paced visual-manual task on a secondary display, involuntary distraction with abrupt onset of irrelevant visual-audio stimuli on the secondary display, and no distraction.

Results

The number of glances toward the secondary display under voluntary distraction was not correlated to that under involuntary distraction. The former was associated with gender, age, annual mileage, and self-reported distraction engagement; such associations were not observed for the latter. Accelerator release time in response to lead-vehicle braking was delayed similarly under both conditions.

Conclusion

Propensity to engage in voluntary distractions appears to be not related to the inability of suppressing involuntary distractions. Further, voluntary and involuntary distraction both affect braking response. These findings have implications for design of in-vehicle technologies, which may be sources of both distraction types.

Keywords

driver distraction driving simulator voluntary distraction involuntary distraction individual differences

Introduction

Distraction is a significant contributor to motor vehicle crashes. The U.S. government reported 3,477 fatalities and estimated 391,000 injuries from distraction-affected crashes for 2015 (National Highway Traffic Safety Administration, 2017). The SHRP 2 Naturalistic Driving Study (NDS) found that drivers engaged in observable distraction in more than 50% of the baseline epochs analyzed and that distraction is associated with a crash risk 2 times that of model driving (where the driver is alert, attentive, and sober; Dingus et al., 2016). Although a large portion of distractions identified in SHRP 2 NDS was carried out purposefully (e.g., handheld cell phone and in-vehicle device use), some distractions appeared to occur without intention, such as when the drivers’ attention was diverted to external objects (Dingus et al., 2016).

In their definition of driver distraction, Lee, Young, and Regan (2008) state that “distraction can occur when extremely salient perceptual cues compel drivers to attend to a source against their will or when drivers willingly divert their attention from the road or from the primary driving task” (p. 34). A wealth of literature has examined intentional, or voluntary, driver distraction; however, involuntary driver distraction in general, and its relation to voluntary distraction engagement in particular, are not well studied. Like other risky driving behaviors, some drivers are at an elevated risk for distracted driving compared with other drivers. For example, Pöysti, Rajalin, and Summala (2005) reported that younger age, male gender, and higher mileage in driving were associated with more frequent use of cell phones while driving. However, whether these drivers are at an elevated risk for both voluntary and involuntary distractions is an open research question, the answer to which can guide the design of personalized interventions.

In naturalistic studies, voluntary visual-manual secondary tasks, such as dialing and texting on the phone, have been found to be particularly detrimental to driving and to increase crash risk significantly (Dingus et al., 2016; Fitch et al., 2013). Simulator studies have also shown engagement in such secondary tasks to affect driving performance negatively and degrade responses to hazards (e.g., Drews, Yazdani, Godfrey, Cooper, & Strayer, 2009; Horberry, Anderson, Regan, Triggs, & Brown, 2006). In terms of the level of engagement in voluntary distractions, some drivers experience stronger social-psychological facilitators to do so. For example, self-reported engagement in voluntary distractions has been associated with attitudes (Chen, Donmez, Hoekstra-Atwood, & Marulanda, 2016; Horrey & Lesch, 2008; Li, Gkritza, & Albrecht, 2014), personality (Feng, Marulanda, & Donmez, 2014; Lansdown, 2012), and normative influences (Carter, Bingham, Zakrajsek, Shope, & Sayer, 2014). Demographics are also known to play a role in voluntary distraction engagement but also in the ability to respond to hazards while engaged in voluntary distractions. Pöysti et al. (2005) found that among those who reported phone use while driving, young age, occupational position, and low safety motivation—but not gender or driving skill level—increased self-reported hazards related to phone use.

Contrary to voluntary distraction, which involves deliberate, goal-directed attention selection processes, involuntary distraction is better described as a bottom-up selection process that may be part of the exploration of the driving environment or the result of automatic responses to the physical properties of an irrelevant stimulus (see Trick & Enns, 2009, for a framework differentiating attentional processes in driving). Research has shown that drivers do fixate on irrelevant stimuli during a large portion of their driving time (Green, 2002; Hughes & Cole, 1986). However, such fixations may become problematic when a stimulus captures attention at the wrong time; it is known that even short “inopportune glances” can lead to crashes (Victor et al., 2015). In fact, the SHRP 2 data revealed that drivers made glances of extended duration to external objects in about 0.93% of the baseline epochs analyzed, with a substantial crash risk that is 7.1 times that of model driving (Dingus et al., 2016).

An Australian study showed that 8.9% of drivers who reported distractions prior to crashing cited distraction by outside persons, objects, or events (Beanland, Fitzharris, Young, & Lenné, 2013). Increased crash rates have also been associated with road sections with a greater number of roadside billboards (see Wallace, 2003, for review), and simulator studies have shown that driving on a road with billboards may lead to risky behaviors, such as dangerous intersection crossings (Bendak & Al-Saleh, 2010) and hard braking around video advertisements (Chattington, Reed, Basacik, Flint, & Parkes, 2009). These studies revealed safety concerns for engaging in involuntary distraction while driving, but their focus has been mostly on distractions external to the vehicle despite the fact that involuntary stimuli can also exist inside the vehicle (e.g., a phone buzzing inside the car).

Although limited research has associated selective-attention abilities, measured using cognitive tasks, to traffic crashes (Arthur & Doverspike, 1992), little is known about the attentional mechanisms and facilitators behind driver engagement in involuntary distraction. Basic visual search experiments have demonstrated that salient and sudden onset of stimuli can capture attention automatically despite suppression attempts (Irwin, Colcombe, Kramer, & Hahn, 2000; Theeuwes, 1992; Theeuwes & Godijn, 2001), but the ability to selectively attend to task-relevant stimuli and to suppress irrelevant stimuli have been found to vary significantly among individuals (e.g., Murphy, 2002). Demographics have been explored as a potential factor that can explain this variability. Merritt et al. (2007) showed that males and females exhibit qualitatively different responses to different types of cues in Posner cuing tasks, and Stoet (2010) found that females are more strongly affected by the task-irrelevant distractions in a flanker task. Age-related increases have also been observed in Stroop interference (e.g., Bugg, DeLosh, Davalos, & Davis, 2007; Weir, Bruun, & Barber, 1997), and Maylor and Lavie (1998) suggest that aging reduces both inhibitory control and processing capacity.

Recent research on selective attention has also demonstrated that attentional capture can occur with inconspicuous, but value-driven, task-irrelevant stimuli (Anderson, Laurent, & Yantis, 2011; Theeuwes & Belopolsky, 2012). The consideration of value (i.e., previously associated with reward) in stimuli is particularly relevant in driving, given that involuntary driver distraction stimuli are often associated with some value to the driver. For example, a social-media notification may be valuable to the driver by previous experience, when the driver had found the content of such notifications useful and engaging. As such, a nonsalient incoming notification may become a compelling source of involuntary distraction due to the associated value, albeit its lack of saliency. In addition to previous experiences, vulnerability to value-driven stimuli also varies with working-memory capacity and trait impulsivity in individuals (Anderson et al., 2011).

Overall, we aim to explore how drivers are susceptible to involuntary distraction and voluntary distraction, respectively, at an individual level. Knowledge of the individual characteristics associated with increased susceptibility to voluntary or involuntary distraction, as well as a deeper understanding of the safety consequences of these distractions, can have significant implications for the design of in-vehicle systems and the traffic environment. Data were collected for 36 participants in a driving simulator under three distraction conditions: (a) voluntary distraction, that is, driving while performing a self-paced visual-manual task on a secondary display with no added incentives for engagement; (b) involuntary distraction, that is, driving while stimuli irrelevant to driving appeared abruptly on the secondary display, and (c) driving with no secondary task or stimulus. Our focus was on identifying similarities/differences in voluntary and involuntary distraction engagement at the individual driver level rather than how voluntariness of distraction engagement affects safety.

To study individual differences in distraction engagement, we recruited individuals selected from a wide range of self-reported distraction engagement frequency. As an initial step, we focused on a fairly narrow age range (i.e., 25 to 39) but had a sample that was balanced for gender. Data on annual mileage were also collected as a proxy for driving experience. We expected to see variability across participants in the amount of engagement with the secondary task and with the irrelevant stimuli. Based on studies described earlier, we expected to see males and those with higher mileage to engage more with the secondary task and females to be more affected by the presence of irrelevant stimuli. Although age plays a role in distraction engagement, we did not expect major effects for age in our study given the narrow range we focused on.

To examine how the voluntary and involuntary distractions implemented in our experiment affected driving performance, we utilized lead-vehicle braking events. Based on the literature cited earlier, we hypothesized that driving performance in the voluntary-distraction condition would deteriorate compared with baseline driving. We expected to see little, if any, negative impact on driving performance in the involuntary distraction condition, given the brief and irrelevant nature of the stimuli designed for this study; further details about the task and stimuli are provided in the Method section.

Method

The experiment followed a repeated-measures design with one independent variable, distraction type (three levels): voluntary, involuntary, and none (baseline). The presentation order was counterbalanced to control for learning and fatigue. Ethics approval was obtained from the University of Toronto Research Ethics Board.

Participants

Thirty-six participants (18 females), between the ages of 25 and 39 (M = 29.1, SD = 3.76), recruited from the Greater Toronto Area using online and poster advertisements, completed the study. All had a valid full Canadian driver’s license and normal or corrected-to-normal vision with the ability to wear contact lenses (to enhance eye-tracking accuracy). In addition to these criteria, participants were also recruited based on their self-reported frequency of engagement in distracted driving. Stratified random sampling was used, in which six male and six female participants were recruited in low (12 total), medium (12 total), and high (12 total) distraction engagement categories, using the Distraction Engagement scale from the Susceptibility to Driver Distraction Questionnaire (SDDQ; Feng et al., 2014). Potential participants responded to how frequently (on 5-point Likert scales ranging from never to very often) they engaged in a list of six distractions:

When driving, you: (1) hold phone conversations, (2) manually interact with a phone, (3) adjust the settings of in-vehicle technology, (4) read roadside advertisements, (5) continually check roadside crash scenes if there are any, and (6) chat with passengers if you have them.

Responses across these six items were averaged to provide a score (1–5) of self-reported distraction engagement (SRDE). Following the distribution of SRDE scores from over 500 respondents in Feng et al. (2014; male, M = 3.1, SD = 0.49; and female, M = 3.0, SD = 0.45), participants in the current study were selected based on their level of distraction engagement corresponding with the following SRDE scores: low (1–2.6), medium (2.8–3.2), and high (3.5–5). The gaps in scores between levels were implemented to provide a clear boundary between levels of SRDE in participants selected for the study. Participants were compensated at C$15 per hour and were incentivized with an additional C$5 for study completion.

Apparatus

A NADS quarter-cab MiniSim Driving Simulator was utilized (Figure 1). This fixed-based simulator has three 42-in. widescreen displays, creating a 130° horizontal and 24° vertical field of view at a 48-in. viewing distance. The simulator collects driving data at 60 Hz. A Microsoft Surface Pro 2 was used to present the self-paced secondary task and the involuntary distraction stimuli (described later) and recorded video for glance data with its front-facing camera at 720p. The Surface Pro 2 was positioned to the right of the dashboard where it would not be visually obstructed by the steering wheel. A dashboard-mounted Seeing Machines faceLAB 5.1 eye tracker also collected gaze data at 60 Hz; however, the data were found to be not reliable, and hence eye-glance data were coded manually from video data.

Figure 1.

Simulator setup with (1) eye tracker and (2) Surface Pro 2 for displaying the voluntary distraction secondary task and involuntary distraction stimuli.

Driving Task

Four driving scenarios were created, one for each of the three experimental conditions and one for the practice drive. These scenarios used the same road network, which took approximately 10 min to drive through. The road network began in a rural environment and entered an urban environment for the second half of the drive; oncoming traffic was present throughout the drive. Prior to each drive, participants were told, “Your main task in this study will be the safe operation of the vehicle. Please drive as you would in your own vehicle and prioritize safety as you would in your own vehicle.”

The participants began by following a lead vehicle in the rural environment, where four lead-vehicle braking events took place (two on straight and two on curved sections); they were instructed to follow the lead car at 50 mph (80.5 km/h) unless the lead car braked with its brake lights on. The rural environment, with a posted speed limit of 50 mph, was a two-lane highway with a shoulder on each side of the roadway and yellow lines separating opposing traffic flow. At a fixed distance (223 m) before lead-vehicle braking onset, a mechanism for controlling headway between the participant’s vehicle and the lead vehicle was triggered: The lead vehicle began to adjust its speed smoothly to obtain a 1.8-s gap time. The headway control ceased at predetermined locations along the roadway, where the lead-vehicle brake lights turned on and the lead vehicle braked at a rate of 0.2 g for 7 s. The lead vehicle accelerated and drove away after the final braking event, and the participants continued in the rural environment, passing bicycles on the side of the road, one of which crossed the road in front of the participant. The urban environment, with a posted speed limit of 35 mph (56.3 km/h), was a four-lane road divided by a double solid line and contained intersections, sidewalks, parked cars, buildings, and stationary pedestrians on each side of the road. Participants had to make left turns at two intersections, and at one upcoming intersection, a sudden right-turning vehicle emerged onto the road ahead.

The busy driving environment was designed to command attention from the drivers, creating safety consequences for distraction engagement. Although the event locations were altered across the different drives to minimize learning, participants were able to anticipate a large portion of these events, with the exception of lead-vehicle braking. Therefore, our driving performance analysis focused on the four lead-vehicle braking events that occurred in the rural portion of the road network.

Voluntary-Distraction Task

A self-paced word-matching task was presented on the Surface Pro 2 (Figure 2a). This visual-manual task was shown to delay accelerator release times (ARTs) in response to lead-vehicle braking in Donmez, Boyle, and Lee (2007). Participants were to select one out of 10 phrases to match either discover with its first word, project with its second word, or missions with its third word. All phrases consisted of three words and there was only one correct answer in the list of 10 candidate phrases. Two phrases were displayed at one time, and participants could tap the up and down arrows with their fingers to scroll through all the options. Participants pressed the Submit button to enter their selection and received feedback on whether their entry was correct or incorrect. A new set of 10 phrases became available, regardless of the answer selected being correct or incorrect. The task was available throughout the drive, and participants decided when to start a new task and did so by hitting the Start button. Before driving through the voluntary-distraction condition, participants were given the following instruction:

During the drive, this task will be available at all times. You can choose when to perform the task. Perform the task only when you feel comfortable doing so and at a pace that you are comfortable with. This is not an experiment in risk taking; your primary task, as in the real world, is to drive safely at all times, so please prioritize driving as you normally would.

Figure 2.

(a) The voluntary distraction task as it appeared on the 208-dots-per-inch secondary display. (b) The involuntary distraction stimuli as they appeared on the secondary display. Each animation was 500 × 500 pixels and was displayed once per drive.

Involuntary-Distraction Stimuli

The involuntary-distraction stimuli aimed to capture attention but not to encourage voluntary interaction. Contrary to the self-paced distraction task, which solicited conscious activities that involve visual, manual, and cognitive efforts, the involuntary distraction was a mere stimulus, not activity, designed to reflect the perceptual component emphasized by Lee et al. (2008). The distraction consisted of a chime sound followed by an abrupt onset of a 5-s geometric animation selected at random from a set of three animations adopted from Lucas (2012; Figure 2b). Whereas the self-paced, voluntary-distraction task mimics the type of visual-manual interactions one may find with in-vehicle infotainment systems, the visual animation and audio stimuli of the involuntary-distraction condition may also find parallels in digital roadside advertisements and sudden audiovisual notifications from a carried-in device.

The stimuli were presented 11 times at fixed locations along the roadway, four of which were synchronized by location with the four lead-vehicle braking events. Prior to the involuntary-distraction drive, participants were instructed, “For this drive, there will be a sound and an animation that appears on the display periodically. You do not need to interact with it.”

Procedure

After the informed-consent process, the participants went through eye-tracking calibration, were shown how to perform the voluntary task, and practiced the task (without driving) five times. After seeing a map of the driving route, the participants completed the practice drive, which included all driving events included in the experimental drives. The voluntary task was also available during the practice drive, and participants were asked to complete the task at least twice. The participants then completed the three experimental drives.

As part of a larger study on the link between cognition and susceptibility to driver distraction, but outside the scope of this paper, participants performed two cognitive tasks using a standard PC setup, one before and one after the simulator portion. Participants also completed the SDDQ for a second time to assess its test-retest reliability (see Marulanda, Chen, & Donmez, 2015, for results). Altogether, the study took approximately 3 hr to complete.

Measures

Table 1 provides a summary of the measures used in our analysis.

Table 1:

Summary of Measures

Distraction engagement

Glance behavior toward secondary display

Average glance duration
Total number of glances
Number of glances >2s

Manual interactions with secondary display (during voluntary distraction)

Number of tasks completed
Number of display taps

Driving performance: Lead-vehicle brake response

Accelerator release time (ART)
Brake transition time (BTT)
Minimum time to collision (TTC_min)
Maximum deceleration

Distraction engagement

Distraction engagement was assessed by examining glance behavior toward the secondary display in both voluntary- and involuntary-distraction conditions. For the voluntary-distraction condition, manual interactions with the secondary display were also assessed using number of tasks completed and number of taps on the display.

Glance data were manually coded using videos recorded by the Surface Pro 2’s camera, post–data collection, following ISO 16673:2007. Glance duration was defined as the time from the direction of gaze moving toward the secondary display to the moment the gaze moved away from it (i.e., fixation time plus transition to the target). Manual coding was performed independently by two coders and verified by one of the authors. Glances less than 100 ms were removed from the analyses as they may not represent meaningful fixations (Crundall & Underwood, 2011). Glance metrics used were average glance duration to the secondary display and the number of glances made toward the secondary display in a drive. The voluntary-distraction condition also included an additional metric to examine number of glances over 2 s, as off-road glances over 2 s have been associated with doubling crash risk (Dingus et al., 2006). Glances over 2 s were very rare in the involuntary condition and hence were not analyzed.

Driving performance

Lead-vehicle braking-event response was assessed using the following metrics: accelerator release time (ART; from lead-vehicle brake onset to accelerator pedal being completely released), brake transition time (BTT; from accelerator release to contact with the brake pedal), minimum time to collision (TTC_min; shortest time required for the lead and following vehicles to collide if they continue at their present speed and on the same path), and maximum deceleration. Note that ART and BTT assume that the accelerator pedal was pressed at the onset of lead vehicle braking and thus exclude cases where the accelerator pedal was already released prior to lead vehicle braking. We used the same stipulation for TTC_min and maximum deceleration.

As mentioned earlier, there were four braking events that occurred in each drive. Some participants failed at times to maintain the prescribed speed, and hence the lead-vehicle braking events did not materialize as intended. Thus, these events (n = 59) were excluded from analysis, and they represented roughly 13% of all braking events. In total, 317 instances of braking events (voluntary distraction, 96; involuntary distraction, 110; and baseline, 111) were included in our analyses.

Results

Our statistical analysis was carried out in R using the nlme package (Pinheiro, Bates, DebRoy, Sarkar, & R Core Team, 2015). In addition to distraction type, we explored SRDE, gender, age, and annual mileage as predictors in our models on distraction engagement. Two age groups, younger than 30 years old (n = 16) and 30 years or older (n = 20), were formed along with two mileage groups, 15,000 km or less in the past year (n = 19) and greater than 15,000 km (n = 17). The thresholds were selected to create approximately equal number of observations within each group.

Voluntary-Distraction Engagement

Voluntary engagement with the task varied widely. One participant never engaged with the task, whereas another completed 26 tasks during the drive. On average, the 36 participants completed 9.2 tasks (SD = 5.3 tasks) with an overall task accuracy of 94.9% (SD = 19.7%) and tapped on the display an average of 95.8 times (SD = 71.4). On average, participants glanced 119 times toward the display (SD = 48.8 glances), corresponding to 22.6% of total drive time (SD = 10.3%), with a mean duration of 1.24 s (SD = 0.4 s).

Number of glances during the voluntary-distraction drive was analyzed in a generalized linear model with log link function, including gender, age group, mileage, and their interactions as predictors. Total drive time in the voluntary condition was included as an offset variable, and quasi-Poisson distribution was used given the presence of overdispersion. Nonsignificant interactions were dropped. Gender had a significant effect, χ²(1) = 5.25, p = .02: Male participants had 1.42 times (95% confidence interval, or CI [1.05, 1.94], p = .03) the rate of glances to the secondary display compared with females. There was a significant interaction between age group and mileage, χ²(1) = 8.98, p = .003. Younger participants (under 30) who had higher mileage were found to have 2.24 times (95% CI [2.01, 2.48], p < .0001) the rate of glances compared with those who had lower mileage. Such a difference was not observed in participants 30 and over (p = .15).

We also analyzed the engagement metrics observed in the voluntary-distraction drive against participants’ SRDE. Generalized linear models with log link function and quasi-Poisson distribution were built to determine if rates of glances, long glances (over 2 s), tasks completed, and taps on the display were different across participants recruited from the low, medium, and high levels of SRDE (see Participants section). Drive time was included as an offset variable. There was a significant relationship between rate of glances and SRDE, χ²(2) = 9.34, p = .009 (Figure 3a). Participants in the high-SRDE group had 1.70 times the rate of glances compared with those in the low-SRDE group (95% CI [1.10, 2.62], p = .01) and a marginally significant 1.49 times the rate of glances compared with those in the medium-SRDE group (95% CI [0.98, 2.27], p = .07); there was no difference between the medium and low groups (p = .79). A marginally significant relationship was found between rate of tasks completed and SRDE, χ²(2) = 5.68, p = .06: The high-SRDE group completed 1.70 times the rate of tasks completed by the low-SRDE group (95% CI [0.99, 2.90], p = .054) but with no significant difference between medium and low (p = .66) or high and medium groups (p = .31). No significant effects for SRDE were observed for other voluntary-distraction engagement variables: average glance duration, F(2, 32) = 0.83, p = .44; rate of glances over 2 s, χ²(2) = 0.02, p = .99; and rate of taps on the secondary display, χ²(2) = 2.68, p = .26.

Figure 3.

Box plots (showing quartiles information; black diamonds represent raw means) of the number of glances made toward the secondary display during (a) voluntary- and (b) involuntary-distraction conditions.

Involuntary Distraction by Irrelevant Stimuli

In the involuntary condition, 32 out of the 36 participants glanced at least once at a distraction stimulus. On average, the 36 participants glanced 5.22 times at the involuntary stimuli with a sizable variability (SD = 5.46) and a mean duration of 548 ms (SD = 207). A generalized linear model with the log link function and quasi-Poisson distribution showed that the number of glances made to the irrelevant stimuli was not related to age group, χ²(1) = 0.34, p = .56; gender, χ²(1) = 0.007, p = .93; or annual mileage, χ²(1) = 0.49, p = .48.

Although we hypothesized the voluntary-distraction engagement metrics to be associated with SRDE, we did not have this hypothesis for involuntary-distraction glance metrics. To confirm the latter, we also carried out analyses of the involuntary-distraction glance metrics against SRDE. Not surprisingly, SRDE was not associated with the number of glances made to the irrelevant stimuli, χ²(2) = 1.19, p = .55 (Figure 3b) or with the average glance duration toward the stimuli, F(2, 29) = 0.72, p = .50. Furthermore, the numbers of glances made in the voluntary and the involuntary conditions were not significantly correlated, r(34) = 0.12, p = .47, suggesting that propensity to engage in voluntary distractions may not be related to the inability in suppressing involuntary distractions for the population investigated in this experiment.

Responses to Lead-Vehicle Braking Under Distraction

Linear mixed models were built to analyze braking response. The four braking event responses were averaged based on road curvature, leading to two data points (one for straight and one for curved road sections) per condition for each participant. Distraction type was included as a fixed factor and participant as a random factor. Road curvature and gap time at the lead-vehicle brake onset were included as covariates to control for potential differences resulting from speed and road condition. Logarithmic transformations were applied to meet linear model assumptions. Nonsignificant interaction terms were dropped. Table 2 provides the F statistics for the models built, along with Ω₀², a measure for the correlation between the fitted and the observed values (Xu, 2003).

Table 2:

Model Results for Lead-Vehicle Braking Response

Variable	Distraction Type		Gap Time		Road Curvature
Variable	F Value	p	F Value	p	F Value	p	Ω₀²
Accelerator release time (ART)	F(2, 157) = 4.06	.02	F(1, 157) = 7.53	.007	F(1, 157) = 0.55	.46	.40
Brake transition time (BTT)	F(2, 157) = 0.91	.41	F(1, 157) = 4.06	.046	F(1, 157) = 15.0	.0002	.63
Minimum time to collision (TTC_min)	F(2, 157) = 3.66	.03	F(1, 157) = 0.56	.45	F(1, 157) = 38.14	<.0001	.69
Maximum deceleration	F(2, 157) = 0.15	.86	F(1, 157) = 13.86	.0003	F(1, 157) = 0.03	.86	.60

Distraction type had a significant effect on ART (Figure 4a). Compared with baseline driving without distraction, participants on average were 24% slower to release the accelerator pedal (ART) under voluntary distraction (95% CI [5, 46], p = .01) and 20% slower under involuntary distraction (95% CI [3, 39], p = .01). ART was not different between voluntary and involuntary conditions (p = .67). Distraction type did not have a significant effect on BTT (Figure 4b). Distraction type also did not have an effect on maximum deceleration, but it was significant for TTC_min. Under involuntary distraction, participants had 11% shorter TTC_min (95% CI [0.01, 21], p = .04; Figure 5), but there were no differences in TTC_min between voluntary distraction and the baseline conditions (p = .99).

Figure 4.

Boxplots (showing quartile information; black diamonds represent raw means) of (a) accelerator release times and (b) brake transition times averaged across the four braking events participants experienced in each condition.

Figure 5.

Box plots (showing quartiles information; black diamonds represent raw means) of minimum time to collision averaged across the four braking events participants experienced in each condition.

Discussion

The present work explored individual differences (namely, gender, age group, annual driving mileage, and self-reported frequency in driver distraction engagement) in how drivers engage with and are affected by voluntary and involuntary driver distraction. A driving simulator study was carried out to examine voluntary engagement in a self-paced secondary task and involuntary response to visual/audio stimuli with abrupt onset. We observed large variability in the level of engagement participants had with both the self-paced secondary task and the irrelevant stimuli. Number of glances to the secondary display in a single drive (~10 min) ranged from zero to 251 in the voluntary distraction condition and from zero to 18 in the involuntary one. The correlation, r(34) = 0.12, p = .47, between the two types of glances was not significant, suggesting that individuals more prone to voluntary distraction are not necessarily more prone to involuntary distraction and vice versa. Even if this correlation were to be statistically significant with a larger sample size, the size of the effect observed in our experiment, a correlation coefficient of 0.12, would represent a small effect size that likely is not practically significant.

Our findings provide support to the conjecture that voluntary and involuntary distraction are likely driven by different attentional mechanisms. Previous research has shown that voluntary-distraction engagement may be facilitated by social-psychological factors (Chen et al., 2016; Chen & Donmez, 2016; Horrey & Lesch, 2008), which were in part reflected in the significance of demographic factors examined in our study. The male participants made more voluntary glances to the secondary task compared with the females, and those who were under 30 years old and reported higher annual mileage also glanced more at the secondary task compared with others in their age group who reported less annual mileage. The effect of mileage was not observed for the participants who were 30 years or older. Our results are in line with the fact that male drivers are generally known to be prone to risk- or sensation-seeking behaviors (Arnett, 1994), and it appears that additional experience gained through higher annual mileage by younger drivers may increase their comfort in engaging in voluntary distractions. On the other hand, attentional research has supported an age-related decline or impairment in the efficiency of inhibitory processes (Hartley, 1993; Maylor & Lavie, 1998). However, such decline in cognitive abilities was unlikely evident in our participants’ age range of 25 to 39 years. It is also possible that the limited sample size had insufficient power to detect differences in involuntary-distraction engagement associated with demographic factors. With the same sample, associations emerged for the voluntary-distraction condition, suggesting that even if such associations exist for involuntary distraction, the size of the effect is lower.

Participants appeared to be aware of their tendency to engage in driver distractions: Those who reported more frequent engagement in distraction in the real world did engage more with the voluntary task used in our simulator study (more glances and marginally more tasks completed), but there was no relationship between this SRDE level and their engagement with the involuntary distraction stimuli. This finding may not be surprising because the scale employed for SRDE consisted mostly of voluntary-distraction items (e.g., hold phone conversations and manually interact with a phone). Finally, we note the overall high accuracy in secondary tasks completed (95%), which suggested that the larger number of tasks completed associated with the higher level of self-reported engagement was likely a result of willingness to engage rather than the ability to perform the secondary task.

For driving performance, the negative impact of voluntary distraction was as expected. Similar delays in ARTs were observed in Donmez et al. (2007), who used the same self-paced secondary task but provided a monetary incentive for task engagement. However, the level of interference involuntary distraction generated on driving performance was rather surprising. Even though the distraction stimuli were irrelevant and brief, participants took longer to release the accelerator in response to lead-vehicle braking in the involuntary condition, compared with baseline driving, at levels similar to those observed with voluntary distraction. Recent work (e.g., Fukuda & Vogel, 2011; Theeuwes, Atchley, & Kramer, 2000) that suggests all people are susceptible to attentional capture but vary in their recovery time from an attentional capture may provide a plausible explanation on this matter. It may be the resources allocated to recovery that led to delayed ART, rather than the abrupt attentional capture that we designed to assess. The current analysis does not directly assess residual impact of either the voluntary- or involuntary-distraction engagement and thus cannot provide further insights on this topic.

However, in a follow-up experiment to investigate involuntary distraction using a similar experimental setup, we found higher self-reported everyday distractibility scores from the Cognitive Failures Questionnaire (CFQ) to correlate with longer glances, but not the number of glances made, toward the irrelevant stimuli (Hoekstra-Atwood, Chen, & Donmez, 2017). These relationships suggest that the CFQ scale may correlate better with the ability to disengage from a distraction than with the ability to suppress automatic attentional capture, thereby supporting the dual mechanisms behind involuntary attentional capture (Hoekstra-Atwood et al., 2017).

A limitation of the lead-vehicle braking response analysis was the need to exclude data because, due to participant vehicles falling considerably behind the lead vehicle, there was no need to respond to the braking of the lead vehicle. Although we observed that some participants exhibited a general tendency to drop below the posted speed limit, it is also possible that some of the slowdowns might have been a result of distraction engagement. Future work may employ longer stretches of empty roadway driving to examine how self-paced or involuntary-distraction engagement may affect longitudinal control in the absence of a lead vehicle. In addition, as an exploratory study, current findings are limited by sample size and, subsequently, the number of social-demographic factors that can be addressed. The design of the study also prevented direct comparison of the voluntary and involuntary distractions, as they involved different types of stimuli (text vs. abstract animation) and perceptual channels (visual vs. audiovisual).

Authors of future work may consider aligning the distractions, such as removing the audio component of the irrelevant stimuli, and adapting the current involuntary stimuli for a revised voluntary secondary task. For example, participants may be asked to respond to questions regarding features of the abstract animation (e.g., shape or color). We also note that both voluntary and involuntary distraction in the real world may be much more engaging or compelling, compared with the contrived secondary task and stimuli found in this study. Driving in a simulator was also unlikely to produce the same level of perceived risk or urgency participants may associate distraction with in the real world, given the minimal safety consequences of being involved in a collision in a simulator.

Finally, although we have examined and discussed voluntary and involuntary distractions as a dichotomy in this study, it is not always straightforward to categorize distraction precisely by intentionality. Although drivers may be involuntarily distracted by a roadside advertisement, prolonged engagement due to interest in its content may be voluntary. On the other hand, checking/reading text messages while driving is largely intentional, but the incoming alert itself can involuntarily distract drivers regardless of their intent to respond. When a cell phone is ringing, to not engage in phone calls while driving, the driver requires both an unwillingness to pick up the call and the ability to not be compelled by the ringing to divert his or her attention away from driving. As such, intentionality may be better explored as a continuum, and authors of future research can investigate the interaction or additive effect of voluntary versus involuntary components of a distraction. Authors of future research can also investigate how voluntary and involuntary distraction may manifest in physiological responses (particularly useful for smart driver monitoring) and how to strategize approaches to combat distractions accordingly, taking into consideration individual differences in driving populations (e.g., blocking out stimuli for those who have difficulty suppressing irrelevant stimuli or providing postdrive education for those who are unusually willing to engage in distractions).

With the multitude of technologies being introduced into the driving environment (e.g., smart devices, in-vehicle systems, automated driving, and intelligent transportation systems), distraction will only become more complex and increasingly overwhelming for the everyday automobile driver. One example is how current smartphones can push notifications through multiple perceptual channels—audio, visual, and vibrotactile—and can increase the saliency of such notifications by conveying meanings (e.g., snippets) or by conceivably designing stimuli to elicit affects. In other words, what was once a passive, primarily involuntary source of distraction may become infused with meaning and motivation to engage a driver voluntarily. Designers of any technology to be implemented in the context of driving must consider not only how drivers may consciously interact with it but also how drivers may be distracted involuntarily by the various features associated with the technology. Our results suggest that there is considerable variability among drivers in terms of their susceptibility to voluntary and involuntary distractions and that in-vehicle system designs should benefit from a consideration of the individual driver.

Key Points

There may be large individual differences in how drivers engage with voluntary versus involuntary distraction, evident in the variability observed in a driving simulator study.

Voluntary and involuntary distractions are likely driven by different attentional mechanisms. There was no correlation observed between voluntary and involuntary engagement in our study, and demographic factors (age group, gender, and mileage) were predictive of engagement in voluntary distraction but not involuntary distraction.

Voluntary and involuntary distractions both delayed accelerator release times in response to lead-vehicle braking events.

Footnotes

Acknowledgements

This work was supported by funding from the Toyota Collaborative Safety Research Center (CSRC) and Auto 21 Network of Centres of Excellence. We gratefully acknowledge James Foley, Kazu Ebe, and Chuck Gulash from Toyota CSRC for their feedback and support. We also would like to thank members of Human Factors and Applied Statistics Laboratory at the University of Toronto for their insightful feedback.

Huei-Yen Winnie Chen is an assistant professor at the University at Buffalo, State University of New York, Department of Industrial and Systems Engineering. She received her PhD in industrial engineering from the University of Toronto.

Liberty Hoekstra-Atwood is a human factors researcher at Battelle Memorial Institute. She received her MASc in industrial engineering at the University of Toronto, where she worked in the Human Factors and Applied Statistics Laboratory.

Birsen Donmez is an associate professor at the University of Toronto, Department of Mechanical and Industrial Engineering. She received her PhD in industrial engineering from the University of Iowa in 2007.

References

Anderson

B. A.

Laurent

P. A.

Yantis

(2011). Value-driven attentional capture. Proceedings of the National Academy of Sciences, 108, 10367–10371.

Arnett

(1994). Sensation seeking: A new conceptualization and a new scale. Personality and Individual Differences, 16, 289–296.

Arthur

Doverspike

(1992). Locus of control and auditory selective attention as predictors of driving accident involvement: A comparative longitudinal investigation. Journal of Safety Research, 23, 73–80.

Beanland

Fitzharris

Young

K. L.

Lenné

M. G.

(2013). Driver inattention and driver distraction in serious casualty crashes: Data from the Australian National Crash In-Depth Study. Accident Analysis & Prevention, 54, 99–107.

Bendak

Al-Saleh

(2010). The role of roadside advertising signs in distracting drivers. International Journal of Industrial Ergonomics, 40, 233–236.

Bugg

J. M.

DeLosh

E. L.

Davalos

D. B.

Davis

H. P.

(2007). Age differences in Stroop interference: Contributions of general slowing and task-specific deficits. Aging, Neuropsychology, and Cognition, 14, 155–167.

Carter

P. M.

Bingham

C. R.

Zakrajsek

J. S.

Shope

J. T.

Sayer

T. B.

(2014). Social norms and risk perception: Predictors of distracted driving behavior among novice adolescent drivers. Journal of Adolescent Health, 54(5, Suppl.), S32–S41.

Chattington

Reed

Basacik

Flint

Parkes

(2009). Investigating Driver Distraction: The Effects of Video and Static Advertising (No. PPR409). Wokingham, UK: Transport Research Laboratory.

Chen

H.-Y. W.

Donmez

(2016). What drives technology-based distractions? A Structural Equation Model of social and psychological factors related to technology-based driver distraction engagement. Accident Analysis & Prevention, 91, 166–174.

10.

Chen

H.-Y. W.

Donmez

Hoekstra-Atwood

Marulanda

(2016). Self-reported engagement in driver distraction: An application of the theory of planned behaviour. Transportation Research F: Traffic Psychology and Behaviour, 38, 151–163.

11.

Crundall

Underwood

(2011). Visual attention while driving: Measures of eye movements used in driving research. In Porter

B. E.

(Ed.), Handbook of traffic psychology (pp. 137–148). San Diego, CA: Academic Press.

12.

Dingus

T. A.

Guo

Lee

Antin

J. F.

Perez

Buchanan-King

Hankey

(2016). Driver crash risk factors and prevalence evaluation using naturalistic driving data. Proceedings of the National Academy of Sciences of the United States of America, 113, 2636–2641.

13.

Dingus

T. A.

Klauer

S. G.

Neale

V. L.

Petersen

Lee

S. E.

Sudweeks

J. D.

… Gupta

(2006). The 100-Car Naturalistic Driving Study, Phase II: Results of the 100-car field experiment (No. DOT HS 810 593). Washington, DC: National Highway Traffic Safety Administration.

14.

Donmez

Boyle

L. N.

Lee

J. D.

(2007). Safety implications of providing real-time feedback to distracted drivers. Accident Analysis & Prevention, 39, 581–590.

15.

Drews

F. A.

Yazdani

Godfrey

C. N.

Cooper

J. M.

Strayer

D. L.

(2009). Text messaging during simulated driving. Human Factors, 51, 762–770.

16.

Feng

Marulanda

Donmez

(2014). Susceptibility to driver distraction questionnaire: Development and relation to relevant self-reported measures. Transportation Research Record: Journal of the Transportation Research Board, 2434, 26–34.

17.

Fitch

G. M.

Soccolich

S. A.

Guo

McClafferty

Fang

Olson

R. L.

… Dingus

T. A.

(2013). The impact of hand-held and hands-free cell phone use on driving performance and safety-critical event risk (No. DOT HS 811 757). Washington, DC: National Highway Traffic Safety Administration.

18.

Fukuda

Vogel

E. K.

(2011). Individual differences in recovery time from attentional capture. Psychological Science, 22(3), 361–368.

19.

Green

(2002). Where do drivers look while driving (and for how long)? In Olson

P. L.

Dewar

R. E.

(Eds.), Human factors in traffic safety (2nd ed., pp. 77–110). Tucson, AZ: Lawyers & Judges.

20.

Hartley

A. A.

(1993). Evidence of the selective preservation of spatial selective attention in old age. Psychology and Aging, 8, 371–379.

21.

Hoekstra-Atwood

Chen

H.-Y. W.

Donmez

(2017). Simulator study of involuntary driver distraction under different perceptual loads. Transportation Research Record: Journal of the Transportation Research Board, 2663, 12–19.

22.

Horberry

Anderson

Regan

M. A.

Triggs

T. J.

Brown

(2006). Driver distraction: The effects of concurrent in-vehicle tasks, road environment complexity and age on driving performance. Accident Analysis & Prevention, 38, 185–191.

23.

Horrey

W. J.

Lesch

M. F.

(2008). Factors related to drivers’ self-reported willingness to engage in distracting in-vehicle activities. In Proceedings of the Human Factors and Ergonomics Society 52nd Annual Meeting (pp. 1546–1550). Santa Monica, CA: Human Factors and Ergonomics Society.

24.

Hughes

P. K.

Cole

B. L.

(1986). What attracts attention when driving? Ergonomics, 29, 377–391.

25.

International Organization for Standardization. (2007). Road vehicles—Ergonomic aspects of transport information and control systems—Occlusion method to assess visual demand due to the use of in-vehicle systems (ISO Report No. 16673:2007). Geneva, Switzerland: International Organization for Standardization.

26.

Irwin

D. E.

Colcombe

A. M.

Kramer

A. F.

Hahn

(2000). Attentional and oculomotor capture by onset, luminance and color singletons. Vision Research, 40, 1443–1458.

27.

Lansdown

T. C.

(2012). Individual differences and propensity to engage with in-vehicle distractions: A self-report survey. Transportation Research Part F: Traffic Psychology and Behaviour, 15, 1–8.

28.

Lee

J. D.

Young

K. L.

Regan

M. A.

(2008). Defining driver distraction. In Regan

M. A.

Lee

J. D.

Young

K. L.

(Eds.), Driver distraction: Theory, effects, and mitigation (pp. 31–40). Boca Raton, FL: CRC Press.

29.

Gkritza

Albrecht

(2014). The culture of distracted driving: Evidence from a public opinion survey in Iowa. Transportation Research Part F: Traffic Psychology and Behaviour, 26, Part B, 337–347.

30.

Lucas

(2012). 89-A. Retrieved from http://www.89a.co.uk/archive

31.

Marulanda

Chen

H.-Y. W.

Donmez

(2015). Test–retest reliability of the susceptibility to driver distraction questionnaire. Transportation Research Record: Journal of the Transportation Research Board, 2518, 54–59.

32.

Maylor

E. A.

Lavie

(1998). The influence of perceptual load on age differences in selective attention. Psychology and Aging, 13, 563–573.

33.

Merritt

Hirshman

Wharton

Stangl

Devlin

Lenz

(2007). Evidence for gender differences in visual selective attention. Personality and Individual Differences, 43, 597–609.

34.

Murphy

(2002). Inhibitory control in adults with attention-deficit/hyperactivity disorder. Journal of Attention Disorders, 6, 1–4.

35.

National Highway Traffic Safety Administration. (2017). Distracted driving 2015 (DOT HS 812 381). Washington, DC: Author.

36.

Pinheiro

Bates

DebRoy

Sarkar

, & R Core Team. (2015). nlme: Linear and nonlinear mixed effects models. R package version 3.1-122 [Computer software]. Retrieved from https://CRAN.R-project.org/package=nlme

37.

Pöysti

Rajalin

Summala

(2005). Factors influencing the use of cellular (mobile) phone during driving and hazards while using it. Accident Analysis & Prevention, 37, 47–51.

38.

Stoet

(2010). Sex differences in the processing of flankers. Quarterly Journal of Experimental Psychology, 63, 633–638.

39.

Theeuwes

(1992). Perceptual selectivity for color and form. Perception & Psychophysics, 51, 599–606.

40.

Theeuwes

Atchley

Kramer

A. F.

(2000). On the time course of top-down and bottom-up control of visual attention. In Driver

(Ed.), Attention and performance XVIII: Control of cognitive processes (pp. 105–124). Cambridge, MA: MIT Press.

41.

Theeuwes

Belopolsky

A. V.

(2012). Reward grabs the eye: Oculomotor capture by rewarding stimuli. Vision Research, 74, 80–85.

42.

Theeuwes

Godijn

(2001). Attentional and oculomotor capture. In Folk

C. I.

Gibson

B. S.

(Eds.), Attraction, distraction and action: Multiple perspectives on attentional capture (pp. 121–150). Amsterdam, Netherlands: Elsevier.

43.

Trick

L. M.

Enns

J. T.

(2009). A two-dimensional framework for understanding the role of attentional selection in driving. In Castro

(Ed.), Human factors of visual and cognitive performance in driving (pp. 63–73). Boca Raton, FL: CRC Press.

44.

Victor

Dozza

Bärgman

Boda

C.-N.

Engström

Markkula

(2015). Analysis of naturalistic driving study data: Safer glances, driver inattention, and crash risk (No. SHRP2 Report S2-S08A-RW-1). Washington, DC: National Academy of Sciences.

45.

Wallace

(2003). Driver distraction by advertising: Genuine risk or urban myth? In Proceedings of the Institution of Civil Engineers—Municipal Engineer 156 (pp. 185–190). London: Institution of Civil Engineers.

46.

Weir

Bruun

Barber

(1997). Are backward words special for older adults? Psychology and Aging, 12, 145–149.

47.

(2003). Measuring explained variation in linear mixed effects models. Statistics in Medicine, 22, 3527–3541.