The relationship between cell phone use and

4 downloads 0 Views 624KB Size Report
Mar 3, 2017 - alertness during transitions from automated to manual vehicle control. ..... STISIM Drive, Build 20802, and was run on a Dell Optiplex GX620 PC ...
Journal of Safety Research 61 (2017) 129–140

Contents lists available at ScienceDirect

Journal of Safety Research journal homepage: www.elsevier.com/locate/jsr

The relationship between cell phone use and management of driver fatigue: It's complicated Dyani Juanita Saxby, a,⁎ Gerald Matthews, b Catherine Neubauer c a b c

Medical College of Wisconsin, 8701 W Watertown Plank Rd, Milwaukee WI, 53226, United States Institute for Training and Simulation, University of Central Florida, 3100 Technology Pkwy, Orlando, FL 32826, United States USC Institute for Creative Technologies, 12015 East Waterfront Dr., Los Angeles, CA 90094, United States

a r t i c l e

i n f o

Article history: Received 26 February 2016 Received in revised form 23 January 2017 Accepted 23 February 2017 Available online 3 March 2017 Keywords: Passive fatigue Countermeasures Driver insight Automation Simulation

a b s t r a c t Introduction: Voice communication may enhance performance during monotonous, potentially fatiguing driving conditions (Atchley & Chan, 2011); however, it is unclear whether safety benefits of conversation are outweighed by costs. The present study tested whether personalized conversations intended to simulate hands-free cell phone conversation may counter objective and subjective fatigue effects elicited by vehicle automation. Method: A passive fatigue state (Desmond & Hancock, 2001), characterized by disengagement from the task, was induced using full vehicle automation prior to drivers resuming full control over the driving simulator. A conversation was initiated shortly after reversion to manual control. During the conversation an emergency event occurred. Results: The fatigue manipulation produced greater task disengagement and slower response to the emergency event, relative to a control condition. Conversation did not mitigate passive fatigue effects; rather, it added worry about matters unrelated to the driving task. Conversation moderately improved vehicle control, as measured by SDLP, but it failed to counter fatigue-induced slowing of braking in response to an emergency event. Finally, conversation appeared to have a hidden danger in that it reduced drivers' insights into performance impairments when in a state of passive fatigue. Conclusions: Automation induced passive fatigue, indicated by loss of task engagement; yet, simulated cell phone conversation did not counter the subjective automation-induced fatigue. Conversation also failed to counter objective loss of performance (slower braking speed) resulting from automation. Cell phone conversation in passive fatigue states may impair drivers' awareness of their performance deficits. Practical applications: Results suggest that conversation, even using a hands-free device, may not be a safe way to reduce fatigue and increase alertness during transitions from automated to manual vehicle control. © 2017 National Safety Council and Elsevier Ltd. All rights reserved.

1. Introduction The media has suggested that devices allowing for hands-free cell phone communication might serve to improve alertness and performance in monotonous, fatiguing driving situations (Pope, 2009). Anecdotal reports found on driving forums indicate some drivers seem to agree (Weatherson, 2010). Some drivers apparently use cell phone conversation routinely during long, monotonous drives because they perceive it helps reduce fatigue and improve alertness. Yet, this suggestion runs counter to the known dangers of distraction from phoning (Strayer & Drews, 2007), and few studies have investigated costs and benefits of phone conversation in fatigued drivers. Naturalistic driving studies have demonstrated that the risk of cell phone conversation while driving is not completely understood. Victor et al. (2014) analyzed data from The

⁎ Corresponding author at: University of Cincinnati McMicken College of Arts & Sciences, University of Cincinnati, 7148 Edwards One, Cincinnati, OH 45221-0037, United States. E-mail addresses: [email protected] (D.J. Saxby), [email protected] (G. Matthews), [email protected] (C. Neubauer).

http://dx.doi.org/10.1016/j.jsr.2017.02.016 0022-4375/© 2017 National Safety Council and Elsevier Ltd. All rights reserved.

Second Strategic Highway Research Program (SHRP2) and found that talking or listening on a cell phone was likely to decrease crash/near crash risk. Other naturalistic studies (e.g., Fitch et al., 2013; Olson, Hanowski, Hickman, & Bocanegra, 2009) have found that phone conversations do not raise crash risk. Nevertheless, the effects of cell phone use on risk while driving under passive fatigue conditions are often not distinguished from other driving conditions. In this article, we address the potential efficacy of simulated cell phone conversations as a counter to the fatigue that may be induced by vehicle automation, in relation to subjective and objective outcomes. 1.1. Automation and passive fatigue Technologies for vehicle automation have advanced to the point where full automation of driving tasks is feasible (Banks, Stanton, & Harvey, 2014), and ‘driverless cars’ are now on certain roads (Payre, Cestac, & Delhomme, 2014). There may be some benefits to high levels of automation (Jamson, Merat, Carsten, & Lai, 2013). Dopart (2015) has pointed out potential benefits including crash avoidance, reduced travel times, improved transportation system efficiency and improved

130

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

accessibility, particularly for persons with disabilities and aging adults; however, risks include loss of situational awareness (Young & Stanton, 2007). Attention to tasks may be impaired when the operator's role switches from active management of demands to passive display monitoring, as shown in studies of vigilance decrement (Warm, Parasuraman, & Matthews, 2008). A special problem in the highly-automated vehicle may be taskrelated driver fatigue (Neubauer, Matthews, & Saxby, 2012a). In general, driver fatigue may be either active or passive (Desmond & Hancock, 2001; May & Baldwin, 2009; Saxby, Matthews, Warm, Hitchcock, & Neubauer, 2013). Active fatigue is associated with overload and frequent control operations, whereas passive fatigue is elicited by underload and monotony (Desmond & Hancock, 2001). The distinction is important, in part, because the two different types of task fatigue call for different countermeasures (May & Baldwin, 2009). In recent research, Saxby et al. (2013) induced active fatigue during a simulated drive by exposing drivers to wind gusts to increase the required number of steering and acceleration changes. Passive fatigue was elicited by placing the driver in a supervisory role over an automated system in which the only task was to detect a signal of an occasional automation failure. In two studies, subjective fatigue was significantly higher in the passive condition, whereas active fatigue was characterized by increases in distress. Performance was assessed in one study. Passively fatigued drivers had significantly higher brake response times to an unexpected event (a van pulling into the road), and higher crash rates compared to actively fatigued drivers, although the former group actually showed better control of lateral position. A further study showed that allowing drivers voluntary control over use of automation was not effective in alleviating either subjective or objective impacts of full automation (Neubauer, Langheim, Matthews, & Saxby, 2012b). 1.2. Cell phone conversation as a countermeasure to automation-induced fatigue? It is likely that drivers will retain at least some level of control over the vehicle for the foreseeable future, for a variety of reasons including driver preference (Banks et al., 2014; Khan, Bacchus, & Erwin, 2012). However, vehicles may be operated in a mixed mode in which control shifts between automation and the human driver (Khan et al., 2012). For example, one application for full automation is the close-formation platooning of multiple vehicles to ease highway congestion (Shladover, 2007). In such cases, control would be restored to the driver on exiting the highway. Our previous research suggests that the driver would be vulnerable to passive fatigue and loss of alertness under such circumstances (Neubauer et al., 2012b; Saxby et al., 2013), raising the issue of what countermeasures would be effective. Currently, many drivers are exposed to some levels of automation coupled with minimal task demands such as use of cruise control on long stretches of straight highway; thereby, placing the driver in a largely supervisory role, which may induce passive fatigue. An intriguing possibility is that cell phone conversations may counter fatigue. Anecdotal reports of drivers commonly suggest that talking on a cell phone helps them feel subjectively more alert and less fatigued in monotonous road environments (Lissy, Cohen, Park & Graham, 2000). Tasks requiring voice interaction may be effective in countering sleepiness (Takayama & Nass, 2008) and fatigue associated with prolonged driving (Gershon, Ronen, Oron-Gilad, & Shinar, 2009; Oron-Gilad, Ronen, & Shinar, 2008). However, the cognitive demands of tasks used in these studies, such as playing trivia games (OronGilad et al., 2008), may differ from those of naturalistic conversation, especially when the topic is personally important or involving. Performance costs of conversation appear to be higher than those of information-processing tasks such as word games (Horrey & Wickens, 2006). Benefits of conversation for the fatigued driver must outweigh any safety hazards of phone use. Typically, researchers have concluded

that phone use is distracting, especially when the driver texts or uses smartphone functions (Strayer, 2015). Caird, Johnston, Willness, and Asbridge (2014) point out that different methodologies suggest different conclusions (although they converge for the most severe threats such as texting). Epidemiological and simulation studies fairly consistently show that phone use is associated with elevated crash risk and performance impairment (Strayer, 2015). However, some naturalistic studies (e.g., Fitch et al., 2013; Olson et al., 2009) found that phone conversations do not raise crash risk. If conversation is actually safe, it is worth considering it as a fatigue countermeasure. However, each methodology used has its own limitations (Caird et al., 2014). It is difficult to determine crash risk (Strayer, 2015), and, in naturalistic studies, to attribute safety critical events to fatigue (Knipling, 2015). Epidemiological and naturalistic studies also tend to neglect moderator factors that would be important in designing an intervention, such as variation in driver workload, the content of the conversation, the driver's level of fatigue and individual difference factors (Matthews, Saxby, Funke, Emo, & Desmond, 2011). The approach adopted in the current study is to focus on the basic science of the impact of a phone conversation on fatigued driver behavior, using simulation to provide experimental control. Two meta-analyses, covering both simulator and field studies, have investigated conversation effects (Caird, Willness, Steel, & Scialfa, 2008; Horrey & Wickens, 2006). Both analyses concluded that cell phone use has significant performance costs as indicated by delayed reaction times, with minimal impact on lane-keeping performance. Further, Caird et al. (2008) found that effect sizes did not differ significantly across simulator and on-road studies, implying that simulator studies can be informative about real driving. If phone conversation is associated with significantly delayed response times to potentially hazardous events (Strayer, Drews, & Crouch, 2006), then caution is needed in advocating for conversation as a countermeasure for fatigue. Phone use effects may vary with the mental state of the driver, but any claim for benefits of phone use requires rigorous substantiation. Evidence on interaction between phone use and fatigue is limited. An early study of simulated long distance truck driving (Drory, 1985) did show that voice communication improved braking response time to the appearance of tailgate lights during the simulator run, but also elevated subjective fatigue. By contrast, in a field study conducted on a test track, Jellentrup, Metz, and Rothe (2011) found that 5-min phone calls countered driver subjective fatigue, and had a “vitalizing” effect on their mental state. EEG alpha recordings were in line with their subjective report. Eyelid opening measures indicated a positive effect on alertness, but this effect declined with repeated phone calls. Analyses of performance were not reported. Two recent simulator studies have investigated performance impacts more systematically. Atchley and Chan (2011) specifically sought to explore whether a verbal task might improve performance (vehicle control and response times) during a monotonous drive. Participants were randomly assigned to one of three “interactive verbal tasks,” requiring free association, which included: no verbal task, continuous, verbal task, or late verbal task. Atchley and Chan (2011) found that drivers in the late verbal task condition had better vehicle control than those in the other condition. Control was assessed as standard deviation of lateral position (SDLP). They also tended to make fewer abrupt steering maneuvers. However, there was no significant effect of the task manipulation on brake response times to critical events. A followup study using a longer simulated drive (Atchley, Chan, & Gregersen, 2014) replicated the beneficial effects of a late verbal task on SDLP. However, as Atchley et al. (2014) note, it is arguable whether changes in SDLP reflect changes in alertness. The two studies did not confirm the earlier finding (Drory, 1985) that voice communication improved braking response time. It is also unclear whether findings can be generalized to the specific passive fatigue states induced by automated driving. Atchley and Chan (2011) also used wind gusts to maintain engagement, but this manipulation has been shown to induce active

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

fatigue, rather than monotony and passive fatigue (Saxby et al., 2013). Atchley et al.’s (2014) scenario required drivers to pass slow moving vehicles as well as to conduct four emergency stops, features that may also mitigate passive fatigue. It remains unknown whether passive fatigue specifically, which could be induced either by automation or very low workload driving (e.g., with cruise control) can be alleviated via cell phone conversation. 1.3. Evaluation of subjective fatigue effects The studies reviewed suggest that voice tasks may increase (Drory, 1985), decrease (Jellentrup et al., 2011) or have no effect (Atchley et al., 2014) on subjective fatigue. These inconsistencies may in part result from variation in the form of fatigue experienced, and the multidimensional nature of subjective state changes observed in simulator (Matthews & Desmond, 2002; Saxby et al., 2013) and on-road studies (Desmond & Matthews, 2009). There are also substantial individual differences in fatigue response to driving (Thiffault & Bergeron, 2003a; Verwey & Zaidel, 2000). Research has identified five driving-related personality traits (aggression, dislike of driving, fatigue proneness, hazard monitoring, and thrill seeking) that may relate to fatigue and stress response, measured by the Driver Stress Inventory (DSI: Matthews, Desmond, Joyner, Carcary, & Gilliland, 1997). A related issue is the effect of drivers' subjective fatigue on their insight into their own safety and performance. There is at least an approximate correspondence between subjective and objective indices of fatigue (Craig & Tran, 2012; Jones et al., 2006), but studies have also shown dissociations between subjective alertness and performance impairment (Christodoulou, 2012; Zhou et al., 2011). Drivers may also lack insight into the dangers of phone use. Although drivers are generally aware of potential risks of distraction (Cazzulino, Burke, Muller, Arbogast, & Upperman, 2014), cell phone use impairs drivers' ability to assess their performance accurately (Sanbonmatsu, Strayer, Biondi, Behrends, & Moore, 2015). Individuals also tend to underestimate their personal risk, perhaps because of an illusion of control, and younger drivers are especially prone to underestimate risk. Drivers are overconfident in their ability to compensate for a range of impairments, including phone use (Wohleber & Matthews, in press). Schlehofer et al. (2010) found that perceiving oneself as good at compensating for driving distractions and overestimating one's performance on a driving simulator predicted increased cell phone use in real-world driving scenarios. Moreover, individuals who overestimated their performance during cell phone use also tended to have poorer real-world driving records. Schlehofer et al. (2010) suggested that high illusory control may partially explain drivers' decisions to pick up a phone while driving. 1.4. Aims and hypotheses The potential of cell phone conversations to ameliorate the driver's vulnerability to distraction while passively fatigued has not been thoroughly researched. This is a critical area given that manufacturers are beginning to develop hands-free voice communication technology that has been touted as a having a possibility of mitigating fatigue (Pope, 2009). The principal aim of this study was to test whether passive fatigue and associated performance impairments elicited by monotonous conditions are alleviated by personalized conversations, aimed to mimic hands-free cell phone conversation. Multiple measures were taken to evaluate the impact of automation and cell phone use, including subjective state and objective performance measures, since the two may not always correspond (Craig & Cooper, 1992). The study also investigated whether drivers' beliefs about their performance under fatigue correspond to objective performance in the context of cell phone use, and tested for personality predictors of vulnerability to fatigue. We used a methodology validated in a previous study (Saxby et al., 2013) to produce substantial changes in fatigue following

131

automated driving, allowing a novel test of whether cell phone use counters this form of fatigue. Hypotheses were as follows. 1.4.1. Impact of cell phone use on performance of the fatigued driver Given reports of performance enhancement from secondary verbal tasks (Atchley et al., 2014), we hypothesized that a simulated cell phone conversation would enhance performance following a period of full automation, but not after normal driving. Similar to Drews, Pasupathi, and Strayer (2008), we elicited a conversation about a close call event (Bavelas, Coates, & Johnson, 2000). These authors argue that close call scenarios are more engaging and may overcome some of the limitations associated with more artificial types of voice communication. Although beneficial effects of verbal tasks on fatigued drivers appear more reliable for SDLP than for braking response time (e.g., Atchley & Chan, 2011), we also assessed speed of response to an emergency event, similar to Saxby et al. (2013). 1.4.2. Impact of cell phone use on subjective state of the fatigued driver Anecdotal evidence (Lissy et al., 2000) and a field study (Jellentrup et al., 2011) suggest phone use may enhance subjective alertness and energy, although evidence from the simulator is lacking (Atchley et al., 2014). Automation-induced fatigue is characterized by low mental workload and lack of challenge. Thus, we hypothesized that a simulated phone call following automated driving would increase task engagement, in terms of the 3-factor subjective state model (Matthews et al., 2002). However, given the multidimensional nature of state response, it is important to monitor the additional state factors of distress and worry also. In particular, distraction effects may be associated with cognitive interference – intrusive thoughts about task and personal concerns – and so we performed an additional analysis of the components of worry, which include cognitive interference. 1.4.3. Impact of cell phone use on performance worries We suggested above that fatigued drivers' positive perceptions of conversation might reflect illusory control (cf., Schlehofer et al., 2010; Wohleber & Matthews, in press). We aimed to test the effects of phone use on a measure of worries specifically about performance while fatigued (Matthews, Hitchcock, & Saxby, 2008). If drivers' positive perceptions are veridical then phone use should reduce performance worries in fatigued drivers, and reduced performance worry should be aligned with improvements in objective performance. 1.4.4. Individual differences in subjective fatigue response In previous field and simulator studies (Matthews et al., 2011), the DSI fatigue proneness scale was the most reliable predictor of fatigue response, although other dimensions may also be implicated. We aimed to test whether fatigue-prone drivers benefit more from simulated cell phone use, compared with those less susceptible to fatigue. 2. Method 2.1. Participants One-hundred and sixty (80 women, 80 men) undergraduate students from a large, urban university participated in the study for course credit. Ages ranged from 18 to 30 years (M = 19.42, SD = 1.76). All participants had a valid driver's license for an average of four years (SD = 1.72). The minimum amount of experience for participants was reportedly 1 year and the maximum 12 years. It is well documented that young novice drivers are at a higher risk for crashes (McCartt, Shabanova, & Leaf, 2003; Mayhew, Simpson, & Pak, 2003; Subramanian, 2005). On the basis that crash risk is at a lifetime high during the first 2 years of driving (NSC, 2005), participants with less than 2 years of driving experience may be considered novices. In the present study, 12 participants had a total of 2 years or less experience. There were no significant differences in years of driving experience amongst groups. Participants self-reported their

132

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

annual driving as follows: less than 5,000 miles: 20%; 5,000–10,000 miles: 36%; 10,000 to 15,000 miles: 20%; 15,000–20,000 miles: 14%, greater than 20,000 miles: 9%. Individuals with visual impairments were required to wear glasses or contacts during the study. Participants were told that they would perform a three-minute practice drive. They were asked to drive straight ahead and to adhere to traffic signals and minimum and maximum speed limit signs. Participants were advised that driving the simulator can be somewhat different from normal driving, and that they should use the practice drive as an opportunity to become accustomed to the controls. Participants signed an informed consent form approved by the IRB and were given course credit for their participation. 2.2. Design A 2 × 2 between groups design was used in this study, yielding 4 conditions containing 40 participants in each. Between subjects factors were cell phone condition (cell phone or no cell phone) and drive condition (passive or control). 2.3. Questionnaires Three relevant constructs were measured using questionnaires. The Dundee Stress State Questionnaire (DSSQ: Matthews, 2016; Matthews et al., 2002, Matthews, Szalma, Panganiban, Neubauer, & Warm, 2013) was administered before and after the drive to assess the overall, multidimensional subjective state response to the task manipulations, including loss of task engagement. The Driver Fatigue Questionnaire (DFQ: Saxby, Matthews, & Hitchcock, 2007) provides a finer-grained assessment of various aspects of fatigue state, including performance worries which were of special interest here. The DSI (Matthews et al., 1997) measures personality traits associated with stress vulnerability. More details of the scales are as follows. The Dundee Stress State Questionnaire (DSSQ) measures subjective state factors of task engagement (energetic arousal, motivation, and concentration), distress (tense arousal, hedonic tone, and confidence and control) and worry (self-focused attention, self-esteem, and two cognitive-interference scales) (Matthews et al., 2002, Matthews et al., 2013). This questionnaire has been shown to be sensitive to various driver stress and fatigue factors in laboratory and field studies (Matthews, 2016; Matthews et al., 2011). Changes in state are expressed as changes in standard deviation (SD) units. The Driver Fatigue Questionnaire (DFQ: Saxby et al., 2007) assesses multiple fatigue subscales. It allows the researcher to distinguish between different forms of driver fatigue including tiredness, loss of motivation, perceived cognitive impairment and changes in coping strategy. The present study focused on the performance worries subscale only, as a measure of insight into performance difficulties, which may be sensitive to fatigue. The DSI (Matthews et al., 1997) assesses driving history and five dimensions of driver stress vulnerability: aggression, dislike of driving, hazard monitoring, thrill seeking, and fatigue proneness. Scores are scaled to range from 0 to 100. Several studies have confirmed its validity as a predictor of subjective stress and fatigue during driving, as well as performance and safety criteria (Matthews, 2002; Matthews et al., 2011; Rowden, Matthews, Watson, & Biggs, 2011). It was included in the study to test whether aspects of personality relevant to driving moderated impacts of automation and cell phone use on driver fatigue and performance. Selected questionnaire data were analyzed to check that the participant characteristics of the experimental groups were similar. The DSI assesses frequency of driving on a 4-point scale ranging from 1 (less than weekly) to 4 (everyday). The mean rating was 2.64 (SD = 1.28). A 2 × 2 (cell phone × drive) between-subjects ANOVA showed no significant main or interactive effects of the experimental manipulations on driving frequency. The DSI assesses self-rated annual mileage on a

5-point scale ranging from 1 (b5,000 miles) to 5 (N 20,000 miles). The mean was 2.56 (SD = 1.22). A 2 × 2 ANOVA showed no significant effects of experimental manipulations. Thus, groups were matched for driving experience. We also tested for variation in the personality factors measured by the DSI using a 2 × 2 × 5 (cell phone × drive × DSI scale) mixed model ANOVA, with repeated-measures on scale. Means for all five scales were similar across conditions, and the ANOVA did not show any significant effects of experimental manipulations. 2.4. Simulated drive and fatigue manipulations The simulated drive was created using System Technologies, Inc., STISIM Drive, Build 20802, and was run on a Dell Optiplex GX620 PC, equipped with a 3.79 GHz processor and 2 GB of RAM. A single Westinghouse LVM-42w2 42-inch LCD monitor presented the simulated drive. Participants interacted by means of a Logitech MOMO Racing Force Feedback Wheel (model 963282 0403), which includes a steering wheel capable of providing realistic feedback by means of a computercontrolled torque motor, a gas and a brake pedal. Participants sat in an adjustable car seat. The driving simulator is illustrated in Fig. 1. All participants completed a 3-minute practice drive, followed by the 30-minute main drive and then a 5-minute supplementary drive. Table 1 summarizes the design of the drives for control and passive fatigue conditions. In the control condition, the main drive required normal driving with full manual control of the vehicle. In the passive condition, the main drive was fully automated. The purpose of the main drive was to induce fatigue in the passive condition, and so no performance measures were collected during this drive, in either condition. Following immediately after the main drive, the supplementary drive allowed participants in both conditions to complete the same drive so that performance could be compared across conditions. In the passive condition, automation was shut off after 30 min, so that all drivers had full manual control over the simulated vehicle during the supplementary drive. Performance was assessed during this supplementary drive. Participants began the 30-minute main drives approximately 1 min after the practice drive. The control and passive fatigue conditions included the same background scenery, traffic scenarios, and road geometry, which consisted of a two-lane highway including hills and curves. Following the lead of Thiffault and Bergeron (2003b), the roadway scenery in the present study was varied to reduce monotony. Posted minimum and maximum speed limit signs ranged from 40 m.p.h. min/50 max(64.37 k.p.h min/80.47 max) to 50 min/60 max (80.47 k.p.h min/96.56 max) and participants were informed that they would be assessed in part on adhering to the speed limit parameters.

Fig. 1. The driving simulator.

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

133

Table 1 Sequence of drives in control and passive fatigue conditions. Drive

Practice

Control or fatigue induction drive

Performance assessment

Time interval Vehicle control

0–3 min Manual control (both conditions)

30–35 min Manual control (both conditions)

Performance measures

None

0–30 min Manual control (control condition) Full automation (passive condition) None

Lateral vehicle control (SDLP) Braking to event (RT)

Note. The double line indicates a break in driving following practice.

Oncoming traffic and cross traffic were introduced occasionally, but there were no vehicles in the driver's lane. The passive fatigue drive was fully automated up until the final the performance assessment phase. Drivers in the control condition had full control of steering, braking, and acceleration. Drivers in the passive fatigue condition were required only to visually monitor the screen and to press the turn signal when they detected one of three “automation failures.” Two red diamonds were placed in the upper corners of the screen and participants were told that when either diamond switched to a downward pointing arrow, they must correct for an “automation failure” by pressing the corresponding turn signal. The turn signal was chosen as it was easily accessible and did not require participants to remove their hands from the steering wheel. Automation was actually never affected; however, participants were told that automation could still fail at any time despite their efforts to correct for “automation failures.” The experimenter turned off all automation prior to the beginning of the performance assessment phase (5-minute supplementary drive) so that drivers in all conditions were in full control of the vehicle at that point. Again, prior to the simulated drive, participants were informed that automation might fail completely at some point during the drive and that they should regain control of the vehicle if this occurred. They were not given further verbal warning regarding the automation failure during the actual drive. About 4000 ft or 1219.2 m towards the end of the main drive, a pedestrian and dog stepped out into a marked crosswalk to cue participants that they should be prepared to brake at any time, as in normal driving. It appeared that the dog was walking with the pedestrian, but there was not an actual leash attached to the dog. Consequently, slowed response to the subsequent critical event could not be attributed to drivers believing that such events would not occur during the simulation. The driving simulator recorded lateral control (SDLP), response times to the unexpected event, and crashes. The supplementary drive lasted 5 min and began immediately following the main drive. Performance assessment began at this time. The supplementary drive was completely flat and straight. There was no oncoming traffic during this period. A critical event was introduced about 2 min and 30 s into the supplementary drive. The event was a van initially parked on the side of the road, which pulled suddenly into the road. The background scenery was a sandy shoreline against the ocean with no other background stimuli, so that the van was conspicuous for each participant (see Fig. 2). In the cell phone conditions, the experimenter always initiated the event while the participant was in the process of verbally responding to a question. The experimenter sat several feet behind the participant in the same room, in order to control ending of vehicle automation.

in that participants had to engage in recall of past events and their personal impacts. The experimenter first asked participants to, “Tell me about an incident in which you had a close call experience.” If participants didn't have a close call experience to share, the experimenter adapted the conversation to ask about someone close to the participant having a close call. The definition of a “close call” story was left open to interpretation. The complete set of questions used to guide the conversation can be found in Appendix A. No device was used for the conversation as the experimenter was in the room with the participant; yet, the experimenter remained out of participant view. In addition, the experimenter was not watching the participant during the supplementary drive to avoid achieving a shared experience with the participant that an actual passenger might have (see Drews et al., 2008). Prior to the experiment, participants were told that they may have to answer questions at some point. They were not told that the task was to mimic a hands-free cell phone conversation so that the conversation would be more spontaneous (as though they were answering a call they did not expect to receive). For those assigned to the conversation condition, the experimenter prompted the conversation approximately 30 s into the supplementary drive. Previous studies (e.g., Saxby et al., 2013) had shown that 30 s was sufficient to allow drivers in the passive fatigue condition to fully regain control of the vehicle. 3. Results 3.1. Subjective state To determine the effect of drive condition and simulated cell phone conversation on subjective state, pre-drive scores were first analyzed to test whether there were pre-existing differences in subjective state scales between groups. This was accomplished using a 2 × 2 × 3 ANOVA (drive condition × cell condition × scales). There were no

2.5. Cell phone conversation Following Bavelas et al. (2000) and Drews et al. (2008), close-call stories were used to increase personal engagement. In the present study, the conversation was guided by a set of questions asked by the experimenter to provide structure. Primarily open-ended questions were asked regarding a close call experience that either the participant or someone close to them had experienced. The questions were designed to provide an emotionally and cognitively engaging conversation

Fig. 2. The van pulling into the road from the participants' view.

134

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

significant main or interaction effects of the drive and cell condition factors, showing that there were no pre-existing imbalances in task engagement, distress, and worry across groups. The effects of drive condition on the DSSQ change scores (post-task– pre-task) for task engagement, distress, and worry were analyzed using a series of 2 (drive condition) × 2 (cell phone condition) betweensubjects ANOVAs. Individuals in the passive fatigue conditions (M = −.76; SD = .79) experienced a significantly greater decrease in task engagement following the drive compared to individuals in the control conditions (M = −.43; SD = .72), F(1, 156) = 7.69, p b 0.01, partial η2 = 0.05. No significant differences in task engagement between cell phone conditions, p N 0.05, were found and the interaction between drive condition and cell phone condition was not significant, p N 0.05. No significant differences between drive conditions or cell phone conditions for distress were found, p N 0.05 in both cases. Furthermore, the interaction between drive condition and cell phone condition was not significant, p N 0.05. For worry, the effect of drive condition was not significant, p N 0.05; yet, the effect of cell phone condition was significant, F(1, 156) = 6.45, p = 0.01, partial η2 = 0.04. Individuals in the cell phone conditions (M = 0.12; SD = 0.69) experienced a significantly greater increase in worry after the drive compared to those in the no cell phone conditions (M = −0.14; SD = 0.61). The interaction between cell phone condition and driving condition was not significant, p N 0.05. To further explore worry for drivers in the cell phone conditions, 2 (drive condition) × 2 (cell phone condition) between-subjects ANOVAs were performed on cognitive interference related to the task and cognitive interference unrelated to the task. There were no significant effects of the experimental factors on task-related interference. However, there was a significant main effect of cell phone on cognitive interference unrelated to the task, F(1, 156) = 8.81, p b 0.01, partial η2 = 0.05. Individuals in the cell phone conditions (M = 0.04; SD = .99) had more worries unrelated to the task compared to those in the no cell phone conditions (M = −.44; SD = 1.07). Furthermore, the interaction between cell phone and drive conditions was significant, F(1, 156) = 4.14, p b 0.05, partial η2 = 0.03 such that phone use had a stronger effect on interference in the passive condition. Fig. 3 shows in the passive fatigue condition, task-unrelated cognitive interference was higher in the cell phone than in the no cell condition. Note that an elevation of personal worries for drivers in this cell phone condition could be due to the nature of the close call story and not simply to the conversation itself.

3.2. Insight We tested whether drivers in the passive fatigue condition were aware of the performance decrements (slowed brake response times) they experienced. It should be noted that naturalistic driving studies

Fig. 3. Task unrelated cognitive interference for all conditions. (All error bars are standard errors).

have not shown performance impairments associated with cell phone conversations in rear-end driving scenarios (see Victor et al., 2014). Yet, naturalistic studies have not thoroughly considered whether there might be an impact in passively fatigued drivers. The DFQ performance worries scale assesses a driver's concern about losing awareness of what is going on around them, perceptions of performance (e.g., speed, reaction time, braking ability), and perceived ability to pay attention and drive safely overall. Performance worries were low initially (M = 2.10; SD = 2.87), and did not differ across conditions prior to the simulated drive. Thus, a 2 (drive condition) × 2 (cell phone condition) between-subjects ANOVA was conducted to examine group differences in change in performance worries (i.e., post-drive worry–pre-drive worry). The effect of cell phone condition was not significant, p N 0.05; however, the effect of drive condition was significant, F(1, 143) = 4.5, p b 0.05, partial η2 = 0.03. Individuals in the passive fatigue conditions (M = 3.82; SD = 4.79) experienced a smaller increase in performance worries than drivers in the control conditions (M = 5.77; SD = 6.27). The drive × cell condition interaction was also significant, F(1, 141) = 7.3, p b 0.01, partial η2 = 0.05. Fig. 4 shows change in performance worries across all conditions. Individuals in all conditions had relatively low performance concerns prior to the drive. Following the drive, participants in the passive fatigue/cell phone condition had significantly lower performance worries compared to those in all other conditions, whereby individuals in the control, cell phone condition were the most worried about their performance. Passive fatigue tended to reduce concern about impairment only in drivers who were in the cell phone condition. 3.3. Performance The performance assessment phase was divided into three main segments: pre-van, van (response time), and post-van. Pre-van data were derived from an approximately 150 s interval, prior to the van pulling into the roadway. Pre-van data were divided into five, 30-second sections to provide more detailed analyses of performance change, but the first section was left out in order to allow participants time to regain vehicle control. Post-van data collection commenced approximately 30 s after the participant began to re-accelerate after either crashing or stopping for the van, which consisted of the last two minutes (approximately) of the drive. Table 2 shows the time sequence for data recording in the performance assessment phase. The driver's lateral control was indexed as SDLP, computed as the SD of the sequence of lateral positions of the vehicle logged by the simulator for each 30 s section. A 2 (drive condition) × 2 (cell phone condition)

Fig. 4. Performance worries across conditions.

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140 Table 2 Time sequence for data recording in the performance assessment phase. Time

Data recording

30 32 33 35

Performance data (SDLP) collection begins after the main drive Van pulls into the road and response time is recorded SDLP data collection is re-initiated The simulated drive ends

a

min and 30s min and 30sa mina min

Indicates approximate time.

between-subjects ANOVA on the average SDLP of all four pre-van sections revealed an effect of cell phone condition, F(1, 155) = 4.47, p b 0.05, partial η2 = 0.03. Both cell phone conditions had significantly lower overall SDLP (M = .66; SD = 0.20) compared to those in the no cell phone conditions (M = 0.76; SD = 0.35), indicating better vehicle control. The effects of drive condition as well as the interaction of cell phone × drive condition were not significant, p N 0.05 in both cases. Brake response time (RT) was a primary measure used to understand how drivers in each condition responded to the van suddenly pulling into the road. The RT was calculated as the time interval between the first movement of the initially stationary van towards the roadway and the driver's first foot contact with the brake pedal. Individuals who did not hit the brakes were given a RT of 5 s for that measure since the van was programmed to pull in front of the driver when he or she was 5 s away and so this is the latest possible time that the driver could have responded. Only the effect of drive condition was significant, F(1, 156) = 13.69, p b 0.01, partial η2 = 0.08. Drivers in the passive fatigue conditions had significantly delayed braking RTs (M = 2.25; SD = 1.07) compared to those in the control conditions (M = 1.68; SD = 0.87) regardless of simulated cell phone conversation. Mean braking RTs for participants in all conditions are shown in Fig. 5. Simulated cell phone use did not have a significant effect on braking RTs, and the interaction between drive and phone was also non-significant (p N 0.05). A chi-square test showed that the relative proportions of colliders in the two conditions (i.e., control and passive) deviated significantly from chance expectations, χ2(1, N = 160) = 5.58, p b 0.05. A second chi-square test was performed to examine the relation between the two cell phone conditions and the likelihood of collision avoidance, but this test was not significant, p N 0.05. Drivers in the passive fatigue conditions were more likely to collide with the van, regardless of simulated cell phone conversation. 36.3% of drivers crashed in the control condition, but 63.3% crashed in the passive fatigue condition. A 2 (drive condition) × 2 (cell phone condition) between-subjects ANOVA was conducted to test group differences in vehicle control

Fig. 5. Mean brake RTs by condition.

135

(SDLP) after the van pulled into the road. Results showed that the effect of drive condition was not significant, but the effect of cell phone condition was significant, F(1, 155) = 4.04, p b 0.05, partial η2 = 0.03. The cell phone group showed lower SDLPs (M = .72; SD = 0.24) than the no cell phone group (M = 0.80; SD = 0.23).

3.4. Predictors of subjective state and braking RT Table 3 shows the associations between the driver stress traits, measured with the DSI, and the three dimensions of subjective state, using data pooled across conditions. Driver stress factors were quite predictive of state prior to driving, indicating that the driving context elicits anticipatory individual differences in affective response. For example, those high in fatigue proneness tend to be disengaged, distressed, and worried even before driving. Correlates of state tended to be similar pre- and post-drive, although significant associations between aggression and lower engagement emerged only in post-drive data. The table also includes partial correlations between traits and post-drive states that control for pre-drive state. For example, the first partial in the table, with a value of −0.09, was calculated as the correlation between dislike of driving and post-drive task engagement, controlling for pre-drive engagement. Partials indicate whether traits predicted task-induced change in state. Fatigue proneness and aggression were associated with declining engagement, and dislike of driving, fatigue proneness and aggression with increasing distress. Next, we tested whether trait–state associations were moderated by the experimental manipulations, focusing on task engagement as the state most closely associated with fatigue. That is, can we identify DSI traits associated with elevated fatigue response to simulated cell phone use or to automation? Four-step hierarchical multiple regression analyses were conducted to determine predictors of task-induced changes in task engagement. Drive and cell phone condition effect-coded vectors were created (Pedhazur, 1997). The drive condition and cell phone condition effect-coded vectors were multiplied to calculate an additional vector (interaction term). The five DSI traits were centered. Two further sets of product terms were calculated to test for interactions between DSI traits and experimental manipulations: DSI × drive condition and DSI × cell phone condition. We did not test for DSI × drive × cell phone conditions in order to keep the analysis manageable. The first step of the regression controlled for pre-drive task engagement. At Step 2, the effect-coded vectors for drive and cell phone conditions were entered, followed by the interaction term vector. Step 3 entered the five DSI traits. Finally, the five interaction terms were entered (i.e., DSI × drive terms [in one regression: Step 4a] and five DSI × cell phone terms [in a second regression: Step 4b]). Table 4 gives summary statistics. The first three steps all added significantly to the variance explained. At Step 3, there were significant beta coefficients for pre-task engagement (.65, p b 0.01), automation condition (0.17, p b 0.01), fatigue proneness (0.17, p b 0.05), and aggression (0.14, p b 0.05). However, neither set of interaction terms added significantly to the variance explained at Step 4, implying that associations between DSI traits and task engagement response were not moderated by either experimental variable. Finally, we investigated DSI and DSSQ correlates of braking RT. Table 3 shows DSI correlates: slower initial braking (i.e., longer RT) was associated with higher DSI thrill seeking and lower DSI hazard monitoring. Braking RT was also negatively correlated with DSSQ post-drive task engagement (r = −0.26, p b 0.01), but not with postdrive distress or worry. We tested whether DSI–RT associations were moderated by task condition using a regression approach similar to that described. Three step regressions were run, entering condition vectors, linear DSI terms, and DSI × condition interaction terms in succession. No significant effects of the interactions terms were found.

136

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

Table 3 DSI correlates of subjective state and braking RT. State correlations include pre-task and post-task states, and partial correlation with post-drive state, controlling for pre-drive state. Task engagement

Dislike of driving Aggression Fatigue proneness Thrill seeking Hazard monitoring

Distress

Worry

Pre

Post

Partial

Pre

Post

Partial

Pre

Post

Partial

Braking RT

−0.03 −0.12 −0.32⁎⁎ −0.19⁎ 0.46⁎⁎

−0.09 −0.27⁎⁎ −0.39⁎⁎ −0.22⁎⁎ 0.36⁎⁎

−0.09 −0.25⁎⁎ −0.25⁎⁎ −0.13 0.10

0.32⁎⁎ 0.13 0.23⁎⁎ 0.04 −0.24⁎⁎

0.32⁎⁎ 0.22⁎⁎ 0.28⁎⁎ −0.05 −0.16⁎

0.17⁎ 0.18⁎ 0.18⁎ −0.10 −0.02

0.23⁎⁎ 0.27⁎⁎ 0.19⁎ 0.17⁎ 0.19⁎

0.26⁎⁎ 0.29⁎⁎ 0.24⁎⁎ 0.12 0.19⁎

0.13 0.14 0.14 −0.03 0.06

0.06 0.13 0.12 0.25⁎⁎ −0.19⁎

⁎ p b 0.05. ⁎⁎ p b 0.01.

4. Discussion 4.1. Fatigue and cell phone use: safety implications We confirmed that vehicle automation may have negative impacts leading to loss of situation awareness and safety (Young & Stanton, 2007). Specifically, we replicated Saxby et al.’s (2013; Study 2) findings that shifting from automated to manual driving reduced task engagement and slowed braking RT to a critical event, relative to normal driving. The combination of subjective disengagement and loss of behavioral alertness shown by Saxby et al. (2013) to be the hallmark of passive fatigue confirms that the fatigue induction was successful. The passive fatigue induced by automation, along with other monotonous driving conditions, may be particularly hazardous (May & Baldwin, 2009). We also found a significant correlation between higher task engagement and faster braking. Given the likelihood of higher levels of automation becoming introduced into future vehicles (Banks et al., 2014), developing countermeasures for automation fatigue is critical. Phone conversation may be effective in countering subjective fatigue (Jellentrup et al., 2011), and objective performance impairments (Atchley & Chan, 2011; Atchley et al., 2014), although studies of subjective fatigue have produced inconsistent outcomes. In the present study, simulated cell phone conversation did not appear to counter loss of subjective alertness induced by automation. Task engagement declined substantially following automated driving irrespective of the phone conversation. Longer conversations, or conversations on different topics, might conceivably elevate alertness, but the study provides no evidence that conversation has an immediate alerting impact in the fatigued driver. At the same time, the study showed some more subtle effects on subjective state. Participation in the phone conversation elevated worry, irrespective of automation, perhaps reflecting the personally relevant nature of the conversation. Conversations on more trivial topics may not elevate worry. By contrast, automation tended to reduce one form of worry, task-irrelevant cognitive interference, but interference of this kind was maintained following the conversation. In a real-life study of commercial drivers, Desmond and Matthews (2009) found that task-irrelevant interference was elevated after drives of around 12 h, suggesting that fatigue was accompanied by mind-wandering. Phone use that encourages task-irrelevant interference may threaten the safety of the fatigued driver.

Table 4 Summary statistics for regressions testing interactive effects of DSI traits and experimental factors on post-task engagement. Step

R2

ΔR2

F for Δ

Df

1. Pre-task engagement 2. Experimental factors 3. DSI traits 4. Interaction terms 4a. DSI × Automation 4b. DSI × Cell phone

0.43 0.46 0.52

– 0.03 0.06

116.76⁎⁎ 3.28⁎ 4.07⁎

1,158 3,155 5,150

0.55 0.53

0.03 0.01

1,81 0.55

4,155 4,155

⁎ p b 0.05. ⁎⁎ p b 0.01.

Results also suggest a hidden danger to engaging in cell phone conversation while driving. Drivers in the passive fatigue/cell phone condition reported lower levels of worry about their driving abilities, relative to the other conditions - even though their braking performance was significantly impaired. Yet, drivers in the passive fatigue/no cell phone condition seemed more realistically concerned about their performance deficits. Fatigue may impair drivers' ability to regulate their off-task thoughts, so that using the phone leads to loss of awareness of impairment. By contrast, non-fatigued drivers can reflect on the phone conversation and their own competence simultaneously. That is, fatigue may impair the ability to divide attention across conversation and selfmonitoring, consistent with evidence that fatigue interferes with multitasking (Stark, Scerbo, Freeman, & Mikulka, 2000). The lack of insight in fatigued drivers who use the phone is concerning since it may result in a false sense of security, and failure to take appropriate actions such as resting. This finding also suggests that anecdotal reports of improved alertness resulting from phone conversation (Lissy et al., 2000) may not be reliable, consistent with evidence for illusions of control (Schlehofer et al., 2010) and lack of awareness of errors (Sanbonmatsu et al., 2015) in phone users. Turning to objective performance, results dissociated the effects of the two manipulations. The effect of automation is troubling from a safety standpoint, given that drivers must maintain situation awareness and readiness to respond to unexpected situations such as pedestrians or animals suddenly stepping into the roadway, or other traffic merging into the driver's lane. However, simulated cell phone conversation did nothing to counter slowed response caused by fatigue. We did not find any impact of conversation on braking RT overall, which runs counter to the general trend identified in simulation studies (Caird et al., 2008; Horrey & Wickens, 2006). The Caird et al. (2008) meta-analysis lists reaction time costs associated with distraction as a function of study characteristics. The cost indicates the increase in braking RT that is associated with using a cell phone, relative to a no-phone control condition. We can use these data to judge whether the present findings conflict with the general trend found in similar previous studies. Relevant to the current study, estimated costs in the meta-analysis were 0.25 s overall, 0.19 s for young drivers, 0.18 s for hands-free phone conversation and .14 s for all conversation tasks (Caird et al., 2008). In our data, in the control condition, the 95% confidence interval for the difference in braking RT between no cell phone and cell phone conditions was −0.52 − 0.25 s (mean: −0.13). This confidence interval includes the means reported by Caird et al.; thus, data are not inconsistent with the meta-analysis. We expected a stronger impact of the conversation, but the study may have lacked the statistical power necessary to detect the rather modest RT differences identified by Caird et al. (2008). By contrast, the data match previous findings that a speech-based task can be effective in improving lateral control of the vehicle following an extended period of driving (Atchley & Chan, 2011; Atchley et al., 2014; Drory, 1985). The effect of phone use on SDLP is consistent with evidence that increasing cognitive workload may lead to a systematic decrease in lateral position variability (Cooper, Medeiros-Ward, & Strayer, 2013). These authors suggest that, within a hierarchical control loop model, attending to routine lower-level activities such as lane-

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

keeping, may be counterproductive. Adding workload diverts attention away from routine processing of lateral position, producing a paradoxical improvement in performance. Another explanation for such effects is that drivers tend to underestimate the effort required for routine driving, and adding workload produces more accurate matching of effort for task demands (Matthews, Sparkes, & Bygrave, 1996). Fatigue is known to disrupt effort-regulation and may produce declining effort (Hockey, 1997). Results from the present study and numerous others (e.g., Becic et al., 2010; Cooper et al., 2013; Kubose et al., 2006; Saxby et al., 2013) suggest that vehicle control as measured by SDLP should be interpreted with caution when assessing benefits of cell phone use during driving. Indeed, decreased variability in a dynamic environment may be a danger sign. According to Becic et al. (2010), “A variably moving system is potentially a more responsive system; thus, the greater variability in the continuous driving measures when there is no conversation may help explain nonconversing drivers' faster reactions to sudden external events” (p. 6). On the other hand, Matthews and Desmond (2002) found that task-induced fatigue impaired lateral control of the vehicle, accompanied by a decrease in smaller-magnitude steering movements. SDLP may be subject to several distinct influences, making it ambiguous as an index of driver performance. Thus, the results pose a dilemma for those who argue that cell phone conversation improves driving performance on the basis of superior vehicle control as measured by SDLP. For example, Atchley and Chan (2011) found that drivers given a verbal task late in a fatiguing drive had better vehicle control (as measured by SDLP) compared to drivers in continuous verbal and no verbal task conditions, suggesting that a strategically placed verbal task may help drivers maintain road position.

4.2. Individual differences in fatigue response There are pronounced individual differences in driver passive fatigue, suggesting a need to target countermeasures towards those who are most vulnerable (Matthews & Desmond, 2001). Generally, the predictors of subjective fatigue here were similar to those found in previous studies (Matthews, 2002). Fatigue-proneness was associated with decreased task-engagement post-drive, although aggression also emerged as a predictor of state change. This finding may signal the ‘reactive’ nature of driving aggression, as an expression of stress vulnerability and tendencies to cope with stress confrontationally when provoked (Matthews, 2002; Wickens, Wiesenthal, Flora, & Flett, 2011). Dislike of driving was substantially correlated with other aspects of stress–distress and worry–pre- and post-drive, as in previous studies (Matthews, 2002), but not with task-induced change in engagement. We conducted regression analyses to test whether any traits for driver stress were especially predictive of fatigue response in automation and in cell phone conditions, but the DSI trait × manipulation interactions failed to reach significance. Thus, the DSI traits pick up general sensitivity to fatigue and stress, but not any specific vulnerability to automation or phone use. For this purpose, more specialized scales such as those for trust in automation (e.g., Reagan & Bliss, 2013), beliefs about competency in using phones (Schlehofer et al., 2010) or in handling impairments (Wohleber & Matthews, in press) may be more useful. There was some dissociation between predictors of subjective state and braking RT, underscoring the need for multidimensional assessment of subjective and objective indices. It was DSI traits associated with affective response to risk – thrill-seeking and hazard monitoring – that emerged as predictors of braking RT, not fatigue proneness. Both traits were significantly associated, in opposite directions, with task engagement pre- and post-driving, although neither predicted task-induced change. We did find an association between task engagement and faster braking, consistent with other evidence linking engagement to superior vigilance (Matthews, Warm, Reinerman, Langheim, & Saxby, 2010; Matthews, Warm, Shaw, & Finomore, 2014). Thus,

137

fatigue proneness may have an indirect influence on objective alertness mediated by task engagement. 4.3. Practical implications: evaluation and countermeasures The findings have implications for the evaluation of driver passive fatigue effects and for the implementation of counter measures. From an evaluation standpoint, results emphasize the importance of multivariate assessment of fatiguing driving environments and of countermeasures to fatigue. Measurement of multiple dimensions of subjective state allows differentiation of fatigue and stress states that may differ in safety impacts, such as passive and active fatigue (Desmond & Hancock, 2001). Similarly, multiple performance indices, including but not limited to SDLP and braking RT, are necessary to gauge passive fatigue effects on objective driving behavior. Measurement of SDLP is not adequate for assessment of alertness, although lack of alertness may sometimes impact SDLP (Matthews & Desmond, 2001). The current findings concur with previous studies (May & Baldwin, 2009; Saxby et al., 2013) in suggesting that passive fatigue may be especially hazardous. The transactional approach divides countermeasures into those associated with external events and stimuli, with operator response, and with active coping with external demands (Matthews & Desmond, 2001; Neubauer et al., 2012a). From the external perspective, designing vehicles to maintain driver interest and challenge has potential; however, adaptive automation systems that relieve the driver of excessive workload may be more effective for active than for passive fatigue (May & Baldwin, 2009). Conversely, adding extra workload is not necessarily helpful to the fatigued driver, given that the additional cognitive demands of the close call conversation here did not enhance alertness. Task demands that add challenge and interest, such as interactive games (Gershon et al., 2009; Oron-Gilad et al., 2008; Williamson, 2012; Verwey & Zaidel, 1999) may have more potential. May and Baldwin (2009) also suggest that systems that monitor responses such as eyeblinks may be more effective for sleep-induced fatigue rather than task fatigue. However, the large magnitude of the subjective disengagement response seen here and in previous studies (Saxby et al., 2013) suggests that it may be premature to abandon the search for objective passive fatigue indicators. For example, Reinerman, Warm, Matthews, and Langheim (2008) showed a rapid decline in cerebral bloodflow velocity during a monotonous simulated drive. Phone use might present itself as an active coping strategy that the fatigued driver could initiate to counter fatigue. However, the current results do not support its efficacy, consistent with the rather mixed impacts of other self-alerting strategies that drivers may use such as opening the car window (Hanks, Driggs, Lindsay, & Merrill, 1999). Systematic training for fatigue management may be more successful, given the promise of programs for stress management that are based on enhancing coping with realistic driving scenarios (Machin, 2003). The transactional model also emphasizes individual differences in response to external challenges (Lazarus, 1999), suggesting the need for different intervention strategies according to personality factors such as fatigueproneness, hazard monitoring and thrill-seeking. 4.4. Limitations Of course, findings from simulation studies may not generalize to real-world driving. Still, simulation offers controlled environments that inform theoretical understanding about fatigue and cognitive distraction (Matthews et al., 2011). The use of simulator studies to complement field studies is especially important in the present context, because real driving may elicit uncontrolled admixtures of passive and active fatigue. Another challenge is lack of convergence from different methodologies for investigating phone conversation effects (Caird et al., 2014). A specific issue is whether the potentially emotional content of the close-call conversation used here may have been responsible for increased worry; other forms of conversation may not have had this effect. We adopted this methodology following Drews et al.’s (2008)

138

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

recommendation to use naturalistic, ecologically valid conversations in driving simulator studies. However, future studies might vary the content of the conversation to determine its impact on worry and performance. A further limitation is that the drive duration was relatively short. Real monotonous road conditions may extend over many hours. However, the present findings confirm that automation may produce passive fatigue over quite short time intervals (Saxby et al., 2013). Also, previous research has shown similar subjective state response patterns in real-world driving are seen at longer durations (Desmond & Matthews, 2009). It is unknown whether reversion to normal driving would over time eventually counter the fatigue effects seen here and in Saxby et al. (2013). Of relevance to future driverless vehicles, the present results highlight the driver's vulnerability when switching from a higher to lower level of automation. Another limitation of the present study is that it solely focused on task-induced fatigue. Sleep-deprivation and circadian rhythm effects may present further and perhaps subtly different safety impairments (Williamson et al., 2011). Future studies should explore whether interactive effects with passive fatigue and sleepiness and/or circadian rhythm effects might be shown (May & Baldwin, 2009). A further limitation is that the drivers in this study were relatively young and inexperienced. This population is vulnerable to risk underestimation (Machin & Sankey, 2008), which may lead to risky behavior (Cohn, Macfarlane, & Yanez, 1995), contributing to increased crash likelihood (Williams, 2003). Effects of automation and phone use on self-insight might be different in older drivers, who tend to have more realistic evaluations of their driving competence (Glendon, Dorn, Davies, Matthews, & Taylor, 1996). Finally, participants were asked to answer the simulated phone call when requested by the experimenter, though previous studies have shown that drivers' choice to answer a call may vary depending on the driving context (Tivesten & Dozza, 2015). 5. Conclusions In sum, results of this study call into question whether cell phone use should be considered a countermeasure for task-induced fatigue in monotonous driving situations. Findings highlight the potential of automation to induce passive fatigue, a threat to safety that will assume increasing salience as levels of automation in vehicles increase. However, simulated cell phone conversation did not counter the subjective and objective expressions of automation-induced fatigue, including braking speed. Cell phone conversation while passively fatigued may have a hidden danger in that it may impair self-awareness of deteriorating performance. Results also reinforce recent findings suggesting that measures of lateral vehicle control are not valid as indices of alertness. 6. Practical applications Countermeasures to passive fatigue in automated vehicles may require innovative strategies tailored to the individual driver. Secondary tasks, including cell phone use, have been promoted as possible fatigue countermeasures. The present study, which used a close call conversation, suggests possible dangers to such countermeasures, including impairment in the drivers' ability to judge their driving competency. Further research is necessary to determine whether cell phone based tasks can be configured to mitigate the dangers of fatigue in the automated vehicle. Acknowledgement Thank you to Ted Hitchcock at NIOSH for providing the car seat, fullsize steering wheel (Logitech MOMO Racing Force Feedback Wheel) and pedals.

References Atchley, P., & Chan, M. (2011). Potential benefits and costs of concurrent task engagement to maintain vigilance: A driving simulator investigation. Human Factors, 53, 3–12. Atchley, P., Chan, M., & Gregersen, S. (2014). A strategically timed verbal task improves performance and neurophysiological alertness during fatiguing drives. Human Factors, 56, 453–462. Banks, V. A., Stanton, N. A., & Harvey, C. (2014). Sub-systems on the road to vehicle automation: Hands and feet free but not ‘mind’ free driving. Safety Science, 62, 505–514. Bavelas, J. B., Coates, L., & Johnson, T. (2000). Listeners as co-narrators. Journal of Personality and Social Psychology, 79, 941–952. Becic, E., Dell, G. S., Bock, K., Garnsey, S. M., Kubose, T., & Kramer, A. (2010). Driving impairs talking. Psychonomic Bulletin and Review, 17, 15–21. Caird, J. K., Willness, C. R., Steel, P., & Scialfa, C. (2008). A meta-analysis of the effects of cell phones on driver performance. Accident Analysis & Prevention, 40, 1282–1293. Caird, J. K., Johnston, K. A., Willness, C. R., & Asbridge, M. (2014). The use of meta-analysis or research synthesis to combine driving simulation or naturalistic study results on driver distraction. Journal of Safety Research, 49, 91–96. Cazzulino, F., Burke, R. V., Muller, V., Arbogast, H., & Upperman, J. S. (2014). Cell phones and young drivers: A systematic review regarding the association between psychological factors and prevention. Traffic Injury Prevention, 15, 234–242. Christodoulou, C. (2012). Approaches to the measurement of fatigue. In G. Matthews, P. A. Desmond, C. Neubauer, & P. A. Hancock (Eds.), Handbook of operator fatigue (pp. 125–138). Aldershot, UK: Ashgate Press. Cohn, L. D., Macfarlane, S., & Yanez, C. (1995). Risk perception: Differences between adolescents and adults. Health Psychology, 14, 217–222. Cooper, J. M., Medeiros-Ward, N., & Strayer, D. L. (2013). The impact of eye movements and cognitive workload on lateral position variability in driving. Human Factors, 55, 1001-1014. Craig, & Cooper (1992). In A. P. Smith, & D. M. Jones (Eds.), Symptoms of acute and chronic fatigue. Handbook of human performance, Vol. 3. (pp. 289–339). San Diego, CA: Academic Press. Craig, A., & Tran, Y. (2012). The influence of fatigue on brain activity. In G. Matthews, P. A. Desmond, C. Neubauer, & P. A. Hancock (Eds.), Handbook of operator fatigue (pp. 185–196). Aldershot, UK: Ashgate Press. Desmond, P. A., & Hancock, P. A. (2001). Active and passive fatigue states. In P. A. Hancock, & P. A. Desmond (Eds.), Stress, workload, and fatigue (pp. 455–465). Mahwah, NJ: Erlbaum. Desmond, P. A., & Matthews, G. (2009). Individual differences in stress and fatigue in two field studies of driving. Transportation Research Part F: Traffic Psychology and Behaviour, 12, 265–276. Dopart, K. (2015). Automated vehicle research at the U.S. Department of Transportation. Washington, DC: USDOT Intelligent Transportation Systems Joint Program Office. Drews, F. A., Pasupathi, M., & Strayer, D. L. (2008). Passenger and cell-phone conversations in simulated driving. Journal of Experimental Psychology: Applied, 14, 392–400. Drory, A. (1985). Effects of rest and secondary task on simulated truck driving task performance. Human Factors, 27, 201–207. Fitch, G. A., Soccolich, S. A., Guo, F., McClafferty, J., Fang, Y., 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. (report no. DOT HS 811 757). Washington, DC: National Highway Traffic Safety Administration. Gershon, P., Ronen, A., Oron-Gilad, T., & Shinar, D. (2009). The effects of an interactive cognitive task (ICT) in suppressing fatigue symptoms in driving. Transportation Research Part F: Traffic Psychology and Behaviour, 12, 21–28. Glendon, A. I., Dorn, L., Davies, D. R., Matthews, G., & Taylor, R. G. (1996). Age and gender differences in perceived accident likelihood and driver competences. Risk Analysis, 16(6), 755–762. Hanks, W. A., Driggs, X. A., Lindsay, G. B., & Merrill, R. M. (1999). An examination of common coping strategies used to combat driver fatigue. Journal of American College Health, 48, 135–137. Hockey, G. R. J. (1997). Compensatory control in the regulation of human performance under stress and high workload: A cognitive-energetical framework. Biological Psychology, 45, 73–93. Horrey, W. J., & Wickens, C. D. (2006). Examining the impact of cell phone conversations on driving using meta-analytic techniques. Human Factors, 41, 196–205. Jamson, A. H., Merat, N., Carsten, O. M., & Lai, F. C. (2013). Behavioural changes in drivers experiencing highly-automated vehicle control in varying traffic conditions. Transportation Research Part C: Emerging Technologies, 30, 116–125. Jellentrup, N., Metz, B., & Rothe, S. (2011). Can talking on the phone keep the driver awake? Results of a field-study using telephoning as a countermeasure against fatigue while driving. Proceedings of the 2nd International Conference on Driver Distraction and Inattention. Jones, C. B., Dorrian, J., Jay, S. M., Lamond, N., Ferguson, S., & Dawson, D. (2006). Selfawareness of impairment and the decision to drive after an extended period of wakefulness. Chronobiology International, 23, 1253–1263. Khan, A. M., Bacchus, A., & Erwin, S. (2012). Policy challenges of increasing automation in driving. IATSS Research, 35, 79–89. Knipling, R. R. (2015). Naturalistic driving events: No harm, no foul, no validity. In D. V. McGehee, J. D. Lee, & M. Rizzo (Eds.), Driving assessment 2015: International Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design (pp. 196–202). Iowa City: Public Policy Center, University of Iowa. Kubose, T. T., Bock, K., Dell, G. S., Garnsey, S. M., Kramer, A. F., & Mayhugh, J. (2006). The effects of speech production and speech comprehension on simulated driving performance. Applied Cognitive Psychology, 20, 43–63. Lazarus, R. S. (1999). Stress and emotion: A new synthesis. New York: Springer.

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140 Lissy, K., Cohen, J., Park, M., & Graham, J. D. (2000). Cellular phones and driving: Weighing the risks and benefits. Harvard Center for Risk Analysis, 86, 1–6. Machin, M. A. (2003). Evaluating a fatigue management training program for coach drivers. In L. Dorn (Ed.), Driver behaviour and training (pp. 75–83). Aldershot, UK: Ashgate. Machin, M. A., & Sankey, K. S. (2008). Relationships between young drivers' personality characteristics, risk perceptions, and driving behavior. Accident Analysis and Prevention, 40, 541–547. Matthews, G. (2002). Towards a transactional ergonomics for driver stress and fatigue. Theoretical Issues in Ergonomics Science, 3, 195–211. Matthews, G. (2016). Multidimensional profiling of task stress states for human factors: A brief review. Human Factors (Online publication) 10.1177/0018720816653688 Matthews, G., & Desmond, P. A. (2001). Stress and driving performance: Implications for design and training. In P. A. Hancock, & P. A. Desmond (Eds.), Stress, workload and fatigue (pp. 211–231). Mahwah, NJ: Lawrence Erlbaum. Matthews, G., & Desmond, A. (2002). Task-induced fatigue states and simulated driving performance. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 55, 659–686. Matthews, G., Sparkes, T. J., & Bygrave, H. M. (1996). Stress, attentional overload and simulated driving performance. Human Performance, 9, 77–101. Matthews, G., Desmond, P. A., Joyner, L. A., Carcary, B., & Gilliland, K. (1997). A comprehensive questionnaire measure of driver stress and affect. In T. Rothengatter, & E. C. Vaya (Eds.), Traffic and transport psychology: Theory and application (pp. 119–127). Amsterdam: Pergamon. Matthews, G., Campbell, S., Falconer, S., Joyner, L., Huggins, J., Gilliland, K., ... Warm, J. (2002). Fundamental dimensions of subjective state in performance settings: task engagement, distress, and worry. Emotion, 2, 315–340. Matthews, G., Hitchcock, E. M., & Saxby, D. J. (2008). Driver fatigue scale: Initial summary report. Unpublished technical report for National Institute for Occupational Safety and Health. Cincinnati: University of Cincinnati. Matthews, G., Warm, J. S., Reinerman, L. E., Langheim, L. K., & Saxby, D. J. (2010). Task engagement, attention and executive control. In A. Gruszka, G. Matthews, & B. Szymura (Eds.), Handbook of individual differences in cognition: Attention, memory and executive control (pp. 205–230). New York: Springer. Matthews, G., Saxby, D. J., Funke, G. J., Emo, A. K., & Desmond, P. A. (2011). Driving in states of fatigue or stress. In D. Fisher, M. Rizzo, J. Caird, & J. Lee (Eds.), Handbook of driving simulation for engineering, medicine and psychology (pp. 29-1–29-11). Boca Raton, FL: Taylor and Francis. Matthews, G., Szalma, J., Panganiban, A. R., Neubauer, C., & Warm, J. S. (2013). Profiling task stress with the Dundee Stress State Questionnaire. In L. Cavalcanti, & S. Azevedo (Eds.), Psychology of stress: New research (pp. 49–90). Hauppage, NY: Nova Science. Matthews, G., Warm, J. S., Shaw, T. H., & Finomore, V. S. (2014). Predicting battlefield vigilance: A multivariate approach to assessment of attentional resources. Ergonomics, 57, 856–875. May, J. F., & Baldwin, C. L. (2009). Driver fatigue: The importance of identifying causal factors of fatigue when considering detection and countermeasure technologies. Transportation Research Part F: Traffic Psychology and Behaviour, 12, 218–224. Mayhew, D. R., Simpson, H. M., & Pak, A. (2003). Changes in collision rates among novice drivers during the first months of driving. Accident Analysis and Prevention, 35, 683–691. McCartt, A. T., Shabanova, V. I., & Leaf, W. A. (2003). Driving experience, crashes and traffic citations of teenage beginning drivers. Accident Analysis and Prevention, 35, 311–320. National Safety Council (2005). Teen driver: A family guide to teen driver safety. Washington, DC: National Safety Council. Neubauer, C. E., Matthews, G., & Saxby, D. J. (2012a). Driver fatigue and safety: A transactional perspective. In G. Matthews, P. A. Desmond, C. Neubauer, & P. A. Hancock (Eds.), Handbook of operator fatigue (pp. 365–377). Aldershot, UK: Ashgate Press. Neubauer, C., Langheim, L., Matthews, G., & Saxby, D. (2012b). Fatigue and voluntary utilization of automation in simulated driving. Human Factors, 54, 734–746. Olson, R. L., Hanowski, R. J., Hickman, J. S., & Bocanegra, J. (2009). Driver distraction in commercial vehicle operations (report no. FMCSA-RRR-09-042). Washington, D.C.: U.S. Department of Transportation. Oron-Gilad, T., Ronen, A., & Shinar, D. (2008). Alertness maintaining tasks (AMTs) while driving. Accident Analysis & Prevention, 40, 851–860. Payre, W., Cestac, J., & Delhomme, P. (2014). Intention to use a fully automated car: attitudes and a priori acceptability. Transportation Research Part F: Traffic Psychology and Behaviour (Advance online publication) 10.1016/j.trf.2014.04.009 Pedhazur, E. J. (1997). Multiple regression in behavioral research (3rd ed.). Orlando, FL: Harcourt Brace. Pope, B. (2009). Ford says cell-phone use danger exaggerated; supports texting ban. Retrieved March 1, 2011 from http://wardsauto.com/ar/ford_texting_ban_ 090929/ Reagan, I. J., & Bliss, J. P. (2013). Perceived mental workload, trust, and acceptance resulting from exposure to advisory and incentive based intelligent speed adaptation systems. Transportation Research Part F: Traffic Psychology and Behavior, 21, 14–29. Reinerman, L. E., Warm, J. S., Matthews, G., & Langheim, L. K. (2008). Cerebral blood flow velocity and subjective state as indices of resource utilization during sustained driving. Proceedings of the Human Factors and Ergonomics Society, 52, 1252–1256. Rowden, P., Matthews, G., Watson, B., & Biggs, H. (2011). The relative impact of workrelated stress, life stress and driving environment stress on driving outcomes. Accident Analysis & Prevention, 43, 1332–1340. Sanbonmatsu, D. M., Strayer, D. L., Biondi, F., Behrends, A. A., & Moore, S. M. (2015). Cellphone use diminishes self-awareness of impaired driving. Psychonomic Bulletin & Review, 1–7 (Online publication) 10.3758/s13423-015-0922-4

139

Saxby, D. J., Matthews, G., & Hitchcock, T. (2007). Fatigue states are multidimensional: Evidence from studies of simulated driving. Proceedings of the Driving Simulation Conference — North America 2007. Iowa City, IA: University of Iowa. Saxby, D. J., Matthews, G., Warm, J. S., Hitchcock, T. E., & Neubauer, C. (2013). Active and passive fatigue in simulated driving: Discriminating styles of workload regulation and their safety impacts. Journal of Experimental Psychology: Applied, 19, 287–300. Schlehofer, M. M., Thompson, S. C., Ting, S., Ostermann, S., Nierman, A., & Skenderian, J. (2010). Psychological predictors of college students' cell phone use while driving. Accident Analysis and Prevention, 42, 1107–1112. Shladover, S. (2007). PATH at 20—History and major milestones. IEEE Transactions on Intelligent Transportation Systems, 8, 584–592. Stark, J. M., Scerbo, M. W., Freeman, F. G., & Mikulka, P. J. (2000, July). Mental fatigue and workload: Effort allocation during multiple task performance. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, Vol. 44, No. 22. (pp. 863–866). SAGE Publications, 863–866. Strayer, D. L. (2015). Is the technology in your car driving you to distraction? Policy Insights from the Behavioral and Brain Sciences, 2, 157–165. Strayer, D. L., & Drews, F. A. (2007). Cell-phone-induced driver distraction. Current Directions in Psychological Science, 16, 128–131. Strayer, D. L., Drews, F. A., & Crouch, D. J. (2006). A comparison of the cell phone driver and the drunk driver. Human Factors, 48, 381–391. Subramanian, R. (2005). Motor vehicle traffic crashes as a leading cause of death in the United States, 2002, traffic safety facts: Research note. Washington, DC: NHTSA's National Center for Statistics and Analysis, National Highway Traffic Safety Administration. Takayama, L., & Nass, C. (2008). Assessing the effectiveness of interactive media in improving drowsy driver safety. Human Factors, 50, 772–781. Thiffault, P., & Bergeron, J. (2003a). Fatigue and individual differences in monotonous simulated driving. Personality and Individual Differences, 34, 159–176. Thiffault, P., & Bergeron, J. (2003b). Monotony of road environment and driver fatigue: A simulator study. Accident Analysis and Prevention, 35, 381–391. Tivesten, E., & Dozza, M. (2015). Driving context influences drivers' decision to engage in visual–manual phone tasks: Evidence from a naturalistic driving study. Journal of Safety Research, 53, 87–96. Verwey, W. B., & Zaidel, D. M. (1999). Preventing drowsiness accidents by an alertness maintenance device. Accident Analysis and Prevention, 31, 199–211. Verwey, W. B., & Zaidel, D. M. (2000). Predicting drowsiness accidents from personal attributes, eye blinks and ongoing driving behavior. Personality and Individual Differences, 28, 123–142. Victor, T., Bärgman, J., Boda, C. N., Dozza, M., Engström, J., Flannagan, C., ... Markkula, G. (2014). Analysis of naturalistic driving study data: Safer glances, driver inattention, and crash risk (no. SHRP 2 safety project S08A). Warm, J. S., Parasuraman, R., & Matthews, G. (2008). Vigilance requires hard mental work and is stressful. Human Factors, 50, 433–441. Weatherson, B. (2010). Cell phones and driving—The reality-based community. (Forum comment). Retrieved April 1, 2011 from http://www.samefacts.com/2010/01/ technology-and-society/cell-phones-and-driving/ Wickens, C. M., Wiesenthal, D. L., Flora, D. B., & Flett, G. L. (2011). Understanding driver anger and aggression: Attributional theory in the driving environment. Journal of Experimental Psychology: Applied, 17, 354–370. Williams, A. F. (2003). Teenage drivers: Patterns of risk. Journal of Safety Research, 34, 5–15. Williamson, A. (2012). Countermeasures to driver fatigue. In G. Matthews, P. A. Desmond, C. Neubauer, & P. A. Hancock (Eds.), Handbook of operator fatigue. Aldershot, UK: Ashgate. Williamson, A., Lombardi, D. A., Folkard, S., Stutts, J., Courtney, T. K., & Connor, J. L. (2011). The link between fatigue and safety. Accident Analysis and Prevention, 43, 498–515. Wohleber, R. W., & Matthews, G. (2016). Multiple facets of overconfidence: Implications for driving safety. Traffic Psychology and Behaviour: Transportation Research Part F (in press). Young, M. S., & Stanton, N. A. (2007). Back to the future: Brake reaction times for manual and automated vehicles. Ergonomics, 50, 46–58. Zhou, X., Ferguson, S. A., Matthews, R. W., Sargent, C., Darwent, D., Kennaway, D. J., & Roach, G. D. (2011). Sleep, wake and phase dependent changes in neurobehavioral function under forced desynchrony. Sleep, 34, 931–941.

Appendix A 1. Tell me about an incident in which you had a “close call” experience. [If they didn't have a close call experience, we will adapt the questionnaire to ask about “someone close to you” having a close call.] 2. It sounds like a scary ordeal. Describe how you felt emotionally during the experience. 3. What about your feelings after the experience? 4. That is certainly understandable. How did you cope with what had just happened? 5. What factors do you think contributed to you surviving the experience? 6. If you could go back, what things would you have done differently? 7. What do you think could have been done to avoid the situation entirely?

140

D.J. Saxby et al. / Journal of Safety Research 61 (2017) 129–140

8. Did your views on life and death change after the incident? 9. Many young people feel invincible. Did you ever feel this way and if so, did these feelings change after the close call? 10. Many people who have such experiences describe having an “enhanced sense of living in the present.” Did your experience affect you in this way? 11. Some people find that their priorities or interests change after having a close call. Did your experience affect you in this way? 12. How were your relationships with others affected by your close call experience for better or worse? 13. I was once involved in a minor close call hiking accident when I sprained my ankle. Luckily I had two friends with me. One stayed and the other got help. I’m still a little afraid to hike to this day. Did your close call cause you to avoid _______________________________ (whatever the person experienced a close call with)? 14. Thank you for sharing your story with me. Is there anything else you'd like to say about your close call? These questions were shown to produce a conversation lasting about 5 min. The experimenter was sitting out of view of the participant driving the simulator. In addition, the experimenter was not watching the participant during the supplementary drive to avoid achieving a common ground due to shared experiences as a “passenger” (see Drews et al., 2008).

Dyani Saxby is a clinical psychologist at the Clemont J. Zablocki VA Medical Center. She is also an assistant professor at the Medical College of Wisconsin. She received her Ph.D. in Psychology from the University of Cincinnati. She also holds an M.S. in Engineering Technology and Industrial Studies from Middle Tennessee State University. Research interests include driver stress and fatigue as well as clinical factors that may impact driving. Gerald Matthews is a research professor at the Institute for Simulation and Training, University of Central Florida. He received B.A. (1980) and Ph.D. (1984) degrees in Experimental Psychology from the University of Cambridge. He previously held faculty positions at the University of Cincinnati and the University of Dundee. He is Past-President of Division 13 (Traffic and Transportation Psychology) of the International Association of Applied Psychology. His current research interests include driver fatigue, stress factors in the operation of unmanned aerial vehicles, and optimization of human-robot interaction. Catherine Neubauer is a Postdoctoral Scholar at the Institute for Creative Technology at the University of Southern California (USC). She received her B.S. (2008) and Ph.D. (2014) in Experimental Psychology from the University of Central Florida and University of Cincinnati respectively. She also currently teaches in USC's Master's of Applied Psychology Program and works as an independent research consultant investigating effective user experience techniques. Her current research interests include unmanned aerial vehicle (UAV) operation, teamwork assessment, driver fatigue and stress, human computer interaction and automated environments.