On average, drivers are distracted 50% of the time while driving (Dingus et al., 2016). It is notuncommon for a driver to arrive at a destination, and then wonder how he or she made it there.This phenomenon, first termed highway hypnosis, was thought to occur when “the monotony ofthe surroundings and the necessity to attend only to a very small part of the visual field mightinduce some kind of hypnotic trance” (Williams, 1963 as cited in Wertheim, 1978, p. 111). Thisphenomenon is now commonly understood as mind wandering.
Mind wandering is defined as “ashift of attention away from a primary task toward internal information” (Smallwood &Schooler, 2006, p. 946). Several recent studies have evaluated how mind wandering affects driving performance(Bencich, Gamboz, Coluccia, & Brandimonte, 2014; He, Becic, Lee, & McCarley, 2011; Yanko& Spalek, 2014). Of these studies, the primary mind wandering detection methods were eitherself-caught or probe-caught.
Using the self-caught method, participants are instructed to press aresponse button as soon as they notice their mind wandering (He et al., 2011). Alternatively,using the probe-caught method, participants are instructed to press a response button to indicatewhether their mind was wandering when they hear a probe tone (Bencich et al., 2014; Yanko &Spalek, 2014). If, at this time, the participant’s mind was not wandering, they are instructed topress an alternative button to indicate a state of alertness (or task-related thoughts). He et al.
(2011) used the self-caught method and found that variability in velocity increasedwhen drivers were alert compared to mind wandering. They also found that lateral control (i.e., 79 PROCEEDINGS of the Ninth International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design steering, lane deviation, lateral position variability) was unaffected by the driver’s state ofinternal attention.
Similarly, when using the probe-caught method, Bencich et al. (2014) foundthat speed and speed variability increased when drivers were alert compared to mind wandering.Conversely, Yanko and Spalek (2014), who also used the probe-caught method, found thatparticipants drove at greater speeds when mind wandering compared to alert. Given these inconsistent results, the purpose of the present investigation was to examineperformance differences when drivers were actively thinking about the driving task (herereferred to as an alert state) versus experiencing task-unrelated thoughts (here referred to as mindwandering) using two common detection methods, self-caught and probe-caught. Althoughresearchers (e.g., Yanko & Spalek, 2014) suggest that both detection methods are equallyreliable, to our knowledge this has not been previously explored, at least in regard to driving. One issue with the use of the self-caught method is that drivers’ must be aware of their mindwandering in order to report it.
Numerous studies, however, demonstrate that people are oftennot aware, at least for much of the time, that they are mind wandering (for a review see Schooleret al., 2011). Therefore, using a self-caught method may result in reduced reports of mindwandering relative to the probe-caught method. Further, those self-caught instances of mindwandering (when people have the meta-awareness of their mind wandering) may differsystematically from probe-caught instances. Specifically, providing probes are likely to bringdrivers’ awareness back to the task at hand, interferring with the overall nature of the task.
Based on the observation that focusing attention on automatized tasks frequently decreasesperformance (referred to as choking; see Ehrlenspiel, Wei, & Sternad, 2010; Koedijker & Mann,2015), it was hypothesized, as assessed by operational control metrics, that driving performancewould be less variable while mind wandering than alert. Specifically, it was expected that speedvariability, lane deviation, lateral position variability and steering reversal rate would be lowerwhen mind wandering. This finding would support the results of previous studies evaluatingmind wandering while driving (Bencich et al., 2014; He et al., 2011). It was also hypothesizedthat performance decrements elicited by mind wandering would not differ using the twodetection methods, but the frequency of reported mind wandering would be reduced in the Self-Caught Experiment.
SELF-CAUGHT EXPERIMENTMethod Participants. Twenty students (9 men, 11 women) from George Mason University participated inthe study for research credit. All participants were at least 18 years of age, held a valid driver’slicense, and had normal or corrected-to-normal vision and hearing.
Visual acuity of at least 20/20was verified using the Rosenbaum near vision test (Rosenbaum, 1984). On average, participantswere 19.95 (SD = 4.
43) years of age and had 4.00 (SD = 4.09) years of driving experience. Apparatus. Participants completed a series of questionnaires assessing mind wandering andinattention, a demographics and driving history questionnaire, and the Karolinska Sleepiness 80 PROCEEDINGS of the Ninth International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design Scale (KSS; A?kerstedt & Gillberg, 1990). The KSS was administered twice, once prior to theexperiment (along with the other questionnaires), and once after the experiment.
A medium fidelity driving simulator with a 240o rotation steering wheel, and pedals was used inthis study. The independent variables evaluated were attentional state (alert, mind wandering)and drive (drive one, drive two). As is common in previous studies (He et al., 2011), periods ofmind wandering and alertness were evaluated via pairs of 10-second windows pre and post self-caught button press. Here, mind wandering was defined as the period of -13 to -3 seconds priorto the onset of the button press, while alertness was defined as the period of 3 to 13 secondsfollowing the button press (see Figure 1).
Figure 1. Self-caught mind wandering detection times The dependent variables were average speed (km/h), speed variability (standard deviation ofspeed), lane deviation (root-mean-squared-error in meters from lane center), lateral positionvariability (standard deviation in meters of each participant’s mean lateral position), and steeringreversal rate (see He et al., 2011). Procedure. Participants signed an informed consent form and completed a series of surveys.
Participants then completed a practice drive (7 minutes), drive one (15 minutes), the sustainedattention to response task (SART; Robertson, Manly, Andrade, Baddeley, & Yiend, 1997) for 15minutes, drive two (15 minutes), and then the second KSS. During the practice drive, participantswere instructed to maintain a speed of 96.56 km/h (60 mph), stay in the right lane, and whenprompted, to exit the highway and proceed back onto the highway in the opposite direction.Following the practice drive, participants were instructed to read a mind wandering definitionadapted from Singer and Antrobus (1972). The instructions for the two experimental drives weresimilar to the practice drive with one exception.
Participants were instructed to press a button onthe steering wheel when they noticed they were mind wandering (He et al., 2011; Smallwood &Schooler, 2015). Between the experimental drives, participants completed the SART to inducecognitive fatigue (Robertson et al.
, 1997). Performance data on the SART is not presented here.