«Contents lists available at ScienceDirect Transportation Research Part F journal homepage: The short-term effect of ...»
Transportation Research Part F 14 (2011) 251–260
Contents lists available at ScienceDirect
Transportation Research Part F
journal homepage: www.elsevier.com/locate/trf
The short-term effect of verbally assessing drivers’ state on vigilance
indices during monotonous daytime driving
Eike A. Schmidt a,b,⇑, Michael Schrauf c,d, Michael Simon c,e, Axel Buchner b, Wilhelm E. Kincses c
Federal Highway Research Institute (BASt), Section U3: Trafﬁc Psychology, Trafﬁc Medicine, Brüderstraße 53, 51427 Bergisch Gladbach, Germany b Heinrich-Heine-University Düsseldorf, Institut für Experimentelle Psychologie, Universitätsstr. 1, D-40225 Düsseldorf, Germany c Daimler AG, Research and Development, GR/PAP, HPC 50-G024, D-71059 Sindelﬁngen, Germany d University of Regensburg, Institute of Experimental Psychology, Universitätsstr. 31, 93053 Regensburg, Germany e University of Tübingen, Wilhelm-Schickard-Institut für Informatik, Sand 13, D-72076 Tübingen, Germany article info abstract Article history: To investigate the effects of verbal assessment of subjective driver state on objective indi- Received 1 February 2010 cators of vigilance state during a monotonous daytime drive, a real road driving study was Received in revised form 9 December 2010 conducted. During a 4-h drive participants’ subjective state (sleepiness, inattention, Accepted 20 January 2011 monotony) was assessed every 20 min by an investigator accompanying the drive. The assessment procedure consisted of roughly 1 min of verbal interaction. Physiological indicators (EEG alpha spindle rate, blink duration, heart rate) revealed a signiﬁcant
improvement of vigilance state during the communication episode as compared to a Vigilance pre-assessment baseline. The activation persisted for up to 2 min following the end of Fatigue the verbal interaction. Reaction times supported these ﬁndings by indicating a signiﬁcant Monotony decrease after the communication. The P3 amplitude of the auditory event-related poten- Driving tial did not show any consistent results. It can be concluded that a short verbal assessment Self assessment has positive effects on drivers’ vigilance stat
1. Introduction The negative effects of prolonged monotonous night- and daytime driving on driver vigilance have been repeatedly investigated (e.g., Brookhuis & De Waard, 1993; Horne & Reyner, 1999; O’Hanlon & Kelly, 1977; Philip et al., 2005; Schmidt et al., 2009; Thiffault & Bergeron, 2003). In an attempt to reduce the number of fatigue-related accidents, numerous researchers have been working on the task of reliably detecting vigilance-related deﬁcits in drivers (for reviews see Kecklund et al., 2006; Wright, Stone, Horberry, & Reed, 2007). The general absence of a single valid measure for every individual (Kecklund et al., 2006) makes it necessary to acquire a wide variety of vigilance indicators including subjective self-assessment measures of driver state. Subjective measures, often assessed verbally, bear the potential of inﬂuencing the state being under investigation (Kaida, Åkerstedt, Kecklund, Nilsson, & Axelsson, 2007) and by their nature cannot be recorded continuously.
The objective of this study was to evaluate the magnitude and especially the duration of the potential inﬂuence of verbal assessment on driver state in real trafﬁc. This should indicate how often a subjective assessment is feasible without significantly inﬂuencing the state under investigation. As the verbal assessment resembles a short communication between driver ⇑ Corresponding author at: Federal Highway Research Institute (BASt), Section U3: Trafﬁc Psychology, Trafﬁc Medicine, Brüderstraße 53, 51427 Bergisch Gladbach, Germany Tel.: +49 176 2060 2075; fax: +49 211 811 5037.
E-mail addresses: firstname.lastname@example.org (E.A. Schmidt), email@example.com (M. Schrauf), firstname.lastname@example.org (M. Simon), email@example.com (A. Buchner), firstname.lastname@example.org (W.E. Kincses).
1369-8478/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.trf.2011.01.005 252 E.A. Schmidt et al. / Transportation Research Part F 14 (2011) 251–260 and passenger, the presented ﬁndings also allow inferring its potential as a countermeasure to a vigilance decrement under monotonous daytime driving conditions.
1.1. Terminology: fatigue and sleepiness Sleepiness and fatigue are often used synonymously. For clariﬁcation May and Baldwin (2009) introduced their model of fatigue, distinguishing between task-related active as well as passive fatigue and sleep-related fatigue, which most authors refer to as sleepiness or the difﬁculty in remaining awake (Philip et al., 2005). May and Baldwin state that the performance decrement – and therefore the task of detecting it – might be similar for all types of fatigue whereas countermeasures might only work for one or the other. Given that we investigated monotonous daytime driving with well-rested drivers, our focus is on task-related passive fatigue, although strictly speaking a minor inﬂuence of sleep-related fatigue could not be ruled out because our participants partly drove during their circadian afternoon dip.
1.2. Subjective assessment of driver state
Various single- and multi-item questionnaire measures of sleepiness have been developed and have been validated against both physiological as well as performance measures (e.g., Stanford Sleepiness Scale ‘‘SSS’’: Hoddes, Zarcone, Smythe, Phillips, & Dement, 1973; Karolinska Sleepiness Scale ‘‘KSS’’: Kaida et al., 2006; Åkerstedt & Gillberg, 1990). Apart from their limited objectivity, subjective questionnaire measures of sleepiness imply a tradeoff between the number of measurements in a given time period and the possibility of interfering with the process under investigation because answering questions may have activating effects and may thus reduce sleepiness. According to Kecklund et al. (2006) the measurement frequency of the KSS varies from once every 3–5 min down to just a few assessments per hour. Kaida et al. (2007) found that indeed the repeated rating of sleepiness reduced subjective post-test sleepiness, reduced EEG alpha power, and improved the subjective perception of performance. Interestingly, task performance did not improve. This led the authors to conclude that the effects of sleepiness were underestimated in the sleepiness ratings. Kaida et al. were interested in the long-term effects of sleepiness ratings for an entire experimental session. Therefore, conclusions about the immediate inﬂuence of the verbal sleepiness assessments were not obtained.
1.2.1. Indications from studies on countermeasures Although little is known about the short-term effects of a subjective state assessment in a monotonous situation, related research on fatigue countermeasures might shed a light on this issue. A variety of experimental studies systematically evaluating the effects of potential countermeasures on sleep-related fatigue have been conducted by Reyner and Horne (1997, 1998). They concluded that the most effective and enduring countermeasure to sleepiness was a combination of caffeine intake and a short nap. In comparison, any positive effects of exposure to cold air and turning on the radio lasted only for about 15 min. However, questionnaire-based studies showed that in retrospective assessments between 25% and 35% of the drivers considered the engaging in a conversation with a passenger to be useful (Anund, Kecklund, Peters, & Åkerstedt, 2008; Maycock, 1997; Nordbakke & Sagberg, 2007). Therefore, at least some positive effects of investigator–participant communications are to be expected, if only for a brief post-communication interval.
1.2.2. Effect of conversations To our knowledge the potentially activating effects of a verbal communication under monotonous driving conditions have been investigated in simulator studies only (Chan & Atchley, 2009; Drory, 1985; Gershon, Ronen, Oron-Gilad, & Shinar, 2009;
Oron-Gilad, Ronen, & Shinar, 2008). Drory reported that long-haul truck drivers’ performance improved during a monotonous simulator drive when they were engaged in a short verbal task from time to time. Oron-Gilad et al. and Gershon et al. showed that a trivia game, carrying a verbal component, prevented deterioration of simulator driving performance and improved alertness assessed by physiological measures. This effect occurred immediately when engaged in the task but did not last longer than the task itself.
Despite this, studies on the distracting effects of cell phone and passenger conversations (Caird, Willness, Steel, & Scialfa, 2008; Drews, Pasupathi, & Strayer, 2008; Horrey & Wickens, 2006) have suggested that distraction might outweigh any positive effect of these types of conversations on driving performance. It is beyond the scope of this paper to disentangle these effects. We mainly focus on the effect following the communication.
1.3. Assessing drivers’ vigilance state
The vigilance measures applied in the research reported here can be classiﬁed as follows:
1.3.1. Continuous measures In accordance with our previous research we used the alpha spindle rate, a feature derived from the EEG alpha-band (7– 13 Hz), as a sensitive EEG-based measure of a vigilance decrement induced by a monotonous driving situation (Simon et al., in press; Schmidt et al., 2009; see also Kecklund & Åkerstedt, 1993; Papadelis et al., 2007; Tietze & Hargutt, 2001). Apart from its high speciﬁcity to changes in vigilance, the main reason for preferring this measure to the classic EEG power measures is its robustness against external and internal noise and artifacts.
We also measured eye-blink duration because several authors reported blink duration to be their most informative occulomotoric parameter when measuring fatigue (Cafﬁer, Erdmann, & Ullsperger, 2003; Ingre, Åkerstedt, Peters, Anund, & Kecklund, 2006). In a large sample study Schleicher, Galley, Briest, and Galley (2008) observed a gradual increase in blink duration with decreasing alertness.
Finally participants’ heart rate was recorded as an indicator of the physical activation level, which has shown to be a sensitive indicator of vigilance changes in the driving context (O’Hanlon & Kelly, 1977).
1.3.2. Discrete event-related measures In the context of a real driving situation, Laurell and Lisper (1978) demonstrated that an auditory secondary reaction time task was sensitive to changes in vigilance, and that it predicted brake reaction times. Slow reactions (as opposed to fast ones) seem to be particularly sensitive indicators of reduced vigilance (Graw, Kräuchi, Knoblauch, Wirz-Justice, & Cajochen, 2004;
Williams, Lubin, & Goodnow, 1959) even in the driving context (Schmidt et al., 2009). Accordingly we implemented a simple secondary auditory task that most likely would not interfere with the motor requirements of the driving task, assuming
that even minor reductions in vigilance should ﬁrst be reﬂected in secondary task performance (subsidiary task paradigm:
O’Donnell & Eggemeier, 1986). Participants were explicitly instructed to prioritize the primary task of driving and it was to be expected that the potentially high costs of major driving errors would also cause participants to give the highest possible priority to the driving task.
We further assessed the amplitude of the stimulus-induced P3 event-related potential (ERP, for a review see Polich, 2007) that has also been shown to be sensitive to changes in vigilance (Koelega et al., 1992; Schmidt et al., 2009) and can be interpreted as a measure of the processing depth of the auditory stimulus.
We hypothesized that (1) the drivers’ vigilance state should decrease with driving distance (main effect of distance). This should be reﬂected in an increase in subjective fatigue measures, alpha spindle rate, blink duration, and reaction times as well as a decrease in heart rate and P3 amplitude. We further hypothesized that (2) the verbal assessment of drivers’ state by the investigator should improve drivers’ vigilance state (main effect of communication). This should be reﬂected in a decrease in alpha spindle rate, blink duration, and reaction times and in an increase in heart rate and P3 amplitude.
2.1. Participants Twenty-six right-handed participants (20 male, six female; age: M = 26.6, range: 21–40) with extensive driving experience (mean annual driving distance of approximately 20,500 km, i.e. 12,800 miles per year) were recruited on a voluntary basis for an ‘‘in-car EEG driving study’’. Participants were screened for a variety of exclusion criteria (handedness, auditory and visual disabilities, various illnesses), instructed to sleep regularly the night before the experiment, and to refrain from consuming caffeine in the morning on the day of the experiment. For their participation they received a monetary compensation of €100. The size of the sample available for data analysis was reduced by three subjects who aborted the drive due to heavy fatigue after less than 2:30 h driving duration which would have resulted in too few data points for a reliable analysis.
As a result, 23 (18 male, ﬁve female; age: M = 26.7, range: 21–40) data sets containing all measures were available for statistical analysis.
2.2. Materials and procedure
Participants arrived at 9:00 am and signed an informed consent form. While the physiological recording equipment was applied, the participants completed the German versions of the morningness–eveningness–questionnaire (D-MEQ: Griefahn, Künemund, Bröde, & Mehnert, 2001) and the Edinburgh handedness inventory (Oldﬁeld, 1971). Prior to the test-drive in real trafﬁc the participants completed a monotonous simulated driving session of approximately 1.5 h duration starting at 10:00 am in a low-level ﬁxed base simulator (Lane Change Task: Mattes, 2003). The results of the simulator study will be published elsewhere and are not subject of the present paper. To control for possible circadian (Folkard, 1997; Lenné, Triggs, & Redman,