Perception, 2014, volume 43, pages 1203 – 1213
doi:10.1068/p7814
The accuracy of drivers’ judgments of the effects of headlight glare on their own visual acuity Ashley A Stafford Sewall1, Stephanie A Whetsel Borzendowski2, Richard A Tyrrell1 1
Department of Psychology, [2 Applied Building Science, Inc.], Clemson University, 418 Brackett Hall, Clemson, SC 29634-1355, USA; e-mail: [email protected] Received 11 July 2014, in revised form 20 October 2014
Abstract. Drivers’ judgments of the magnitude of disability glare caused by high-beam headlights may not match actual declines in visual performance. This study investigated younger and older drivers’ beliefs about their own visual performance in the presence of headlight glare. Eleven older drivers and seventeen younger drivers judged the distance at which they would just be able to recognize the orientation of a white Landolt C if it were present adjacent to the headlamps of a stationary opposing vehicle at night. The younger participants were generally accurate in their estimates of the recognition distance of the stimulus, while older participants significantly overestimated both their own acuity and the effect of glare on their vision. From this study, we see that older drivers’ judgments about the disabling effects of oncoming headlights may be systematically inaccurate. These misperceptions about headlight glare may help explain why drivers tend to underuse high beams. Keywords: headlight glare, night driving, visual acuity
1 Introduction Most fatal traffic crashes occur at night, and degradations in drivers’ visual performance at night are thought to be a contributing factor in the occurrence of these crashes. For example, pedestrian fatalities increase significantly at night even when controlling for factors such as fatigue or alcohol consumption (Owens & Sivak, 1996; Sullivan & Flannagan, 2002). The National Highway Traffic Safety Administration (NHTSA, 2012) reports that as of the year 2012 more than 4700 pedestrians were killed by traffic crashes in the United States and 70% of these pedestrian fatalities occurred at night. Although all drivers are visually impaired at night, there is evidence to suggest that the visual problems that result from low levels of illumination are particularly severe for older drivers (eg Owsley, Sekuler, & Siemsen, 1983; Shinar & Schieber, 1991). In this paper we examine the impact of headlight glare on both younger and older drivers’ ability to see at night. In particular, we measure these drivers’ actual ability to discriminate fine detail of an object positioned on the roadway when facing an oncoming glare vehicle in comparison with their estimates of their visual performance in the presence of glare. The use of high-beam headlamps can improve the visual conditions faced by drivers at night. Wood, Tyrrell, and Carberry (2005) found that, when drivers used their high beams, the mean distance at which they responded to a pedestrian was 2.6 times greater (white clothes) and 3.5 times greater (black clothes) than when they relied on their low beams. Additionally, researchers have concluded that the visibility afforded by low-beam headlamps can be insufficient for travel speeds above 32 km h–1 (20 mph) (Leibowitz, Owens, & Tyrrell, 1998). Importantly, in conditions of reduced luminance the stopping distance necessary to avoid a collision with an object (eg pedestrians) on or along the roadway is often significantly longer than the visibility distance of that object (Olson & Sivak, 1983; Shinar, 1984). In one study, in 45% of trials involving visually healthy younger drivers, the visibility distance of a pedestrian wearing dark clothing was less than the stopping distance required to avoid a collision with that pedestrian (Olson & Sivak, 1983). In 83% of trials involving visually healthy, older drivers, stopping distance was greater than visibility distance. Dimly illuminated objects like
1204
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
pedestrians often go unnoticed until it is too late for the driver to avoid a collision. Although research has found that pedestrians who place retroreflective markings on their major joints (in configurations that present biological motion information to approaching drivers) can drastically increase their own conspicuity (eg Balk, Tyrrell, Brooks, & Carpenter, 2008; Owens, Antonoff, & Francis, 1994; Wood et al., 2005), the use of biological motion in this context remains relatively infrequent. While drivers should remain vigilant for the sudden appearance of inconspicuous pedestrians and other objects at night, the use of high-beam headlights can significantly improve drivers’ ability to recognize inconspicuous pedestrians along the roadway. Despite the value of high-beam headlamps at night, existing evidence indicates that drivers tend to underutilize their high beams. In a study by Mefford, Flannagan, and Bogard (2006), participants who drove instrumented vehicles for 7–21 days relied on low beams for 75% of the time that high beams could have been safely used. In other observational studies, drivers used their high beams only 10–50% of the time that high beams could have been used (conditions in which there is no opposing vehicles present, no lead vehicle, and on rural roads) (Hare & Hemion, 1968; Sullivan, Adachi, Mefford, & Flannagan, 2004). Though this tendency to underuse high-beam headlamps has been documented repeatedly, we know less about why drivers choose not to use high beams even in ideal high-beam-use conditions. One possible explanation is that drivers typically believe that the illumination provided by their low beams is sufficient to support safe driving. Indeed, at night, drivers receive continuous feedback that they can see well enough to maintain their vehicle’s lane position but rarely receive feedback about their inability to see low-contrast objects on the roadway. It has been proposed that this ‘selective degradation’ of visual functions at night results in drivers being overconfident at night (Brooks, Tyrrell, & Frank, 2005; Leibowitz & Owens, 1977; Owens & Tyrrell, 1999). There is evidence to suggest that drivers have a limited understanding of their visual deficits in low-illumination conditions (eg Olson & Sivak, 1983; Shinar, 1984). Another possible explanation for drivers underusing their high beams is that they may be reluctant to ‘blind’ (ie be a disabling glare source for) other drivers. Glare can indeed have negative effects on visual performance (ie disability glare) and on driving behaviors (Leibowitz, Tyrrell, Andre, Groetzinger-Eggers, & Nicholson, 1993; Theeuwes, Alferdinck, & Perel, 2002). Headlight glare can create intraocular light scattering, which can decrease drivers’ contrast sensitivity and visibility distance. Intraocular light scattering is more pronounced in older drivers, and as such these drivers may have greater visual difficulties (NHTSA, 2007). Additionally, of course, glare can result in an unpleasant subjective experience (ie discomfort glare). There have been numerous consumer complaints about the discomfort created by encounters with high intensity discharge (HID) headlamps or improperly aimed headlamps (NHTSA, 2001). In particular, HID headlamps have been described by consumers as “too bright”, “painful”, and “blinding”. In a study investigating the subjective experience of glare, 31% of drivers aged 65–74 and 26% of drivers aged 18–25 found oncoming glare to be “disturbing” (NHTSA, 2003). Yet, drivers’ subjective impressions that they are visually disabled by glare may sometimes be inaccurate. In Theeuwes et al. (2002) drivers performed a pedestrian detection task while exposed to headlight glare, and indicated the discomfort they experienced as a result of the glare source. Here, drivers’ subjective ratings of discomfort were not predictive of their performance on the pedestrian detection task. It seems possible that drivers’ subjective experience of discomfort caused by a glare source may not be tightly correlated with objective measures of the degree to which the glare source degrades their visual performance. Drivers who experience discomfort glare may incorrectly assume that their visual performance has been significantly impaired, and the link between measures of discomfort glare and measures of disability glare may vary as a function of the age of the driver.
The accuracy of drivers’ judgments of the effects of headlight glare
1205
To attempt to clarify the relationships among discomfort glare, disability glare, and driving behavior, research has begun to examine the accuracy of drivers’ judgments of their own acuity in the presence of glare. In a lab-based psychophysical study Balk and Tyrrell (2011a) had college-age participants (18–25 years) estimate their own visual acuity (by providing both manual and verbal estimates of the size of a barely discernible acuity target) when facing a glare source. In that study, as the glare intensity increased, observers’ estimates of their acuity significantly worsened. Yet, the glare source that was used was not sufficiently intense to affect their actual acuity. Thus, while experiencing the discomfort of a glare source, the observers exaggerated its disabling effect. In a follow-up study that was conducted on a closed road, an experimenter drove college-aged participants slowly toward a stationary opposing vehicle (the ‘glare vehicle’) and asked the participants to estimate the distance at which they would just be able to recognize a retroreflective Landolt C stimulus that had earlier been visible adjacent to the headlamps of the glare vehicle (Balk & Tyrrell, 2011b). Participants estimated significantly shorter recognition distances when facing the opposing vehicle’s high beams than when the opposing vehicle used low beams. However, the participants’ actual ability to recognize the stimulus was unaffected by the beams of the glare vehicle. Thus, like the observers in the Balk and Tyrrell lab study, these participants exaggerated the disabling effect that the glare source would have on their vision. Together, these two studies provide new perspective on drivers’ judgments about their ability to see at night and new methodologies by which these judgments can be investigated. The extent to which the results of the Balk and Tyrrell experiments generalize to situations in which a driver facing headlight glare encounters an object in the roadway at night is unknown. One reason why headlight glare may not have affected their drivers’ visual performance is that the acuity stimulus featured an artificially high level of contrast as a result of it having been retroreflective and illuminated by headlamps. The retroreflective stimulus may have provided sufficient contrast to remain relatively robust to the negative effects of glare. For this reason, the experiment presented here measured observers’ judgments of their ability to recognize a nonretroreflective stimulus (lower contrast and lower luminance) and actual recognition distance of this stimulus in the presence of glare. Additionally, Balk and Tyrrell tested only college-aged drivers. Because older drivers’ reactions to headlight glare may differ from that of younger drivers, we sought to quantify the accuracy of both younger and older drivers’ judgments of the visual effects of headlamp glare. 2 Materials and method 2.1 Participants Eleven older drivers (M = 71 years, range = 65–80 years; four males) and seventeen younger drivers (M = 19 years, range = 18–21 years; nine males) participated in this experiment. Participants were required to achieve a 20/40 (6/12) or better binocular visual acuity (Bailey– Lovie chart under room illumination), and achieve a contrast sensitivity of 1.65 (Pelli–Robson letter sensitivity chart under room illumination), and report having no known visual pathologies other than those being corrected by lenses that were worn during the experiment. All participants gave written informed consent prior to participation. This study, which adhered to the tenets of the Declaration of Helsinki, was approved by the local Institutional Review Board. 2.2 Test site and vehicles The test site was an unilluminated one-lane utility roadway that was 3.05 m wide and that included a section of straight, level roadway that was 230 m long. The asphalt road was without pavement markings and streetlights, and access to the test site was closed during data collection. Participants were tested at least one hour after sunset and only on nights free of precipitation and fog.
1206
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
The test site is depicted in figure 1. The low beams (xenon) and high beams (low beam xenon + halogen) of a 2008 Infiniti EX35 served as the glare source, and throughout the experiment this vehicle remained parked on one end of the straight portion of road facing the participant vehicle. The glare vehicle was parked 213 m from the starting position of the participant vehicle (a 2004 Nissan Sentra with halogen low beams and high beams). Participants were tested individually, and each sat in the front passenger seat of the participant vehicle. The engines of both vehicles were running throughout data collection. The lateral position of the two vehicles was offset as depicted in figure 1.
Stationary glare vehicle Landolt C stimulus headlight beam Cones used as baseline/visual anchor for participants
headlight beam
Moving participant vehicle
Figure 1. [In color online, see http://dx.doi.org/10.1068/p7814] Configuration of experimental vehicles and Landolt C stimulus.
2.3 Visual stimulus The visual stimulus used in this experiment was a nonretroreflective Landolt C (21.5 cm in diameter and 3.5 cm stroke and gap width) constructed from white paper and mounted on a tripod. The legs of the tripod were masked by black felt. From the participant’s perspective, the center of the Landolt C stimulus was located 80.5 cm above the ground, 1.52 m to the right of the right-most edge of the glare vehicle, and facing the participant at a longitudinal distance that was aligned with the headlamps of the glare vehicle. The lateral position of the two vehicles was such that if the participant vehicle had continued to move forward, it would have passed to the right of the glare vehicle with less than 0.2 m separation. 2.4 Experimental design This experiment featured a 2 (low or high beams of the participant vehicle) × 2 (low or high beams of the glare vehicle) × 2 (response type: estimated or actual recognition) withinsubjects design for younger participants and a separate 2 × 2 × 2 within-subjects design for older participants. To prevent participants’ judgments of recognition distance from being influenced by the measurement of actual recognition distances, estimated recognition distances were always measured before actual recognition distances. 2.5 Estimated recognition distance To establish the baseline distance at which the participant could determine the orientation of the stimulus in the absence of any headlight glare, the participant vehicle was first driven slowly (with engine idling) toward the Landolt C in a no‑glare situation (low beams on
The accuracy of drivers’ judgments of the effects of headlight glare
1207
the participant vehicle and only parking lights on the glare vehicle). When the participant indicated that he or she could discern the orientation of the Landolt C, the vehicle was stopped (all participants accurately reported the orientation of the stimulus at this distance). This baseline distance was marked by orange (nonretroreflective) cones on either side of the roadway. These cones remained in place throughout the experiment to indicate to the participants when they were able to see the stimulus with no glare present. This process also familiarized the participant with the size, location, and appearance of the Landolt C stimulus. After measuring and marking the baseline distance, the Landolt C stimulus and tripod were covered by a large black cloth. Despite the fact that the Landolt C was no longer visible, participants were asked to imagine that it was, and over the course of 8 trials (4 headlamp combinations × 2 travel directions) participant’s estimates of the distance at which he or she would be able to just determine the orientation of the Landolt C stimulus were measured. A Latin square was used to determine the order of the 4 headlamp combinations for each participant, and direction of travel was counterbalanced across all participants. Each trial began by driving the participant to the starting point, which was either 213 m (forward trials) or 8 m (reversing trials) from the stimulus and glare vehicle. The headlamps of the two vehicles were then set to the appropriate setting (low vs high beam) for the trial and then turned on. The participant was reminded that their task was to estimate the distance at which he or she would be able to just determine the orientation of the Landolt C stimulus (if it were still visible). The experimenter then drove the participant slowly (with the brakes off and engine idling) toward or away from the stimulus, and the participant marked each estimated recognition distance by dropping a numbered bean bag out of the passenger window. For reverse-moving trials the participant was asked to drop the bean bag at the point at which he or she estimated that the orientation of the stimulus would no longer be discernible if it were still visible. The next trial began approximately two minutes later; during this delay, the headlamps of the glare vehicle were kept off and the participant vehicle was moved to the appropriate starting point. 2.6 Actual recognition distance After all estimated recognition distances were measured, actual recognition distances were measured for the 4 headlamp combinations and 2 directions of travel. For each participant the sequence of the 8 trials matched the sequence used in the measurement of that participant’s estimated recognition distances. The black cloth was removed from the Landolt C stimulus. Participants were again driven slowly toward or away from the stimulus and this time asked to indicate the point at which they were just able to determine (forward trials) or no longer able to determine (reverse trials) the orientation of the Landolt C stimulus. The participant dropped bean bags from the passenger window to mark the locations. The participant was then asked to indicate the orientation of the C by pointing to a chart that presented the 8 possible orientations (0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315°). A delay of approximately two minutes elapsed between adjacent trials; during this delay the headlamps of the glare vehicle were kept off, the participant’s vehicle was moved to the appropriate starting point for the next trial, and the orientation of the C was changed (to a predetermined random orientation). One of the younger drivers reported an incorrect orientation for one of the trials. This trial was repeated after the eighth trial. All other participants reported correct orientations for all trials. Following all measurements, the experimenters measured the distance of each bean bag from the Landolt C using a measuring wheel. 2.7 Subjective ratings of discomfort glare After each estimated and actual recognition trial, the participant rated the intensity of the discomfort caused by the headlights of the glare vehicle using the de Boer scale (scale ranging from 9, “unnoticeable”, to 1, “unbearable”).
1208
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
3 Results The primary purpose of this experiment was to investigate the accuracy of older and younger drivers’ judgments of the effect of glare on their visual performance. Estimated and actual recognition distances were measured from twenty-eight participants (eleven older participants and seventeen younger participants). To examine the accuracy of their judgments, the data were analyzed separately for the two age groups using separate 2 (low or high beams for the participant vehicle) × 2 (low or high beams for the glare vehicle) × 2 (response type: estimated or actual recognition) repeated-measures ANOVAs for each age group. Additionally, for each age group a separate ANOVA analyzed the effects of the beam manipulations on measures of estimated and actual recognition distances. 3.1 Younger participants’ recognition distances The younger participants’ mean initial baseline recognition distance (the position of the cones that marked the distance at which each participant initially recognized the orientation of the stimulus when their vehicle approached using low beams and the glare vehicle used only parking lights) was 47.7 m (standard deviation = 10.7 m). A 2 (beam setting of the participant vehicle) × 2 (beam setting of the glare vehicle) × 2 (response type: estimated and actual recognition distance) repeated-measures ANOVA explored the accuracy of the younger participants’ judgments of recognition distance (see figure 2). On average, estimated recognition distance (M = 44.4 m) was 12% longer than actual recognition distance (M = 39.6 m). However, this main effect was not significant (F1, 16 = 1.9, p > 0.05, hp2 = 0.11); and, further, there were no significant interactions involving the glare vehicle, participant vehicle, or response type on recognition distance ( p > 0.05). This pattern indicates that the younger participants’ estimates of recognition distance were, on the whole, relatively accurate. To analyze the data from the younger group more deeply, separate preplanned 2 (beam setting of participant vehicle) × 2 (beam setting of glare vehicle) repeated-measures ANOVAs analyzed the participants’ estimates of recognition distances and, separately, the measures of the participants’ actual recognition distances. When looking first at estimated recognition, participants’ estimates of recognition distance significantly differed as the glare vehicle’s 80
estimated actual
70
Recognition distance/m
60 50 40 30 20 10 0
glare: low participant: low
glare: low participant: high
glare: high participant: low
glare: high participant: high
Figure 2. Mean estimated and actual recognition distances at each of the four headlamp combinations for younger participants. The horizontal dashed line represents the initial baseline recognition distance (participant vehicle using low beams; glare vehicle using only parking lights).
The accuracy of drivers’ judgments of the effects of headlight glare
1209
headlights changed (F1, 16 = 63.8, p < 0.001, hp2 = 0.80) and as the participant vehicle’s headlights were changed (F1, 16 = 28.5, p 0.05). The actual recognition distances also changed significantly when the headlamps of the glare vehicle were switched to high beams (F1, 16 = 65.2, p
doi:10.1068/p7814
The accuracy of drivers’ judgments of the effects of headlight glare on their own visual acuity Ashley A Stafford Sewall1, Stephanie A Whetsel Borzendowski2, Richard A Tyrrell1 1
Department of Psychology, [2 Applied Building Science, Inc.], Clemson University, 418 Brackett Hall, Clemson, SC 29634-1355, USA; e-mail: [email protected] Received 11 July 2014, in revised form 20 October 2014
Abstract. Drivers’ judgments of the magnitude of disability glare caused by high-beam headlights may not match actual declines in visual performance. This study investigated younger and older drivers’ beliefs about their own visual performance in the presence of headlight glare. Eleven older drivers and seventeen younger drivers judged the distance at which they would just be able to recognize the orientation of a white Landolt C if it were present adjacent to the headlamps of a stationary opposing vehicle at night. The younger participants were generally accurate in their estimates of the recognition distance of the stimulus, while older participants significantly overestimated both their own acuity and the effect of glare on their vision. From this study, we see that older drivers’ judgments about the disabling effects of oncoming headlights may be systematically inaccurate. These misperceptions about headlight glare may help explain why drivers tend to underuse high beams. Keywords: headlight glare, night driving, visual acuity
1 Introduction Most fatal traffic crashes occur at night, and degradations in drivers’ visual performance at night are thought to be a contributing factor in the occurrence of these crashes. For example, pedestrian fatalities increase significantly at night even when controlling for factors such as fatigue or alcohol consumption (Owens & Sivak, 1996; Sullivan & Flannagan, 2002). The National Highway Traffic Safety Administration (NHTSA, 2012) reports that as of the year 2012 more than 4700 pedestrians were killed by traffic crashes in the United States and 70% of these pedestrian fatalities occurred at night. Although all drivers are visually impaired at night, there is evidence to suggest that the visual problems that result from low levels of illumination are particularly severe for older drivers (eg Owsley, Sekuler, & Siemsen, 1983; Shinar & Schieber, 1991). In this paper we examine the impact of headlight glare on both younger and older drivers’ ability to see at night. In particular, we measure these drivers’ actual ability to discriminate fine detail of an object positioned on the roadway when facing an oncoming glare vehicle in comparison with their estimates of their visual performance in the presence of glare. The use of high-beam headlamps can improve the visual conditions faced by drivers at night. Wood, Tyrrell, and Carberry (2005) found that, when drivers used their high beams, the mean distance at which they responded to a pedestrian was 2.6 times greater (white clothes) and 3.5 times greater (black clothes) than when they relied on their low beams. Additionally, researchers have concluded that the visibility afforded by low-beam headlamps can be insufficient for travel speeds above 32 km h–1 (20 mph) (Leibowitz, Owens, & Tyrrell, 1998). Importantly, in conditions of reduced luminance the stopping distance necessary to avoid a collision with an object (eg pedestrians) on or along the roadway is often significantly longer than the visibility distance of that object (Olson & Sivak, 1983; Shinar, 1984). In one study, in 45% of trials involving visually healthy younger drivers, the visibility distance of a pedestrian wearing dark clothing was less than the stopping distance required to avoid a collision with that pedestrian (Olson & Sivak, 1983). In 83% of trials involving visually healthy, older drivers, stopping distance was greater than visibility distance. Dimly illuminated objects like
1204
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
pedestrians often go unnoticed until it is too late for the driver to avoid a collision. Although research has found that pedestrians who place retroreflective markings on their major joints (in configurations that present biological motion information to approaching drivers) can drastically increase their own conspicuity (eg Balk, Tyrrell, Brooks, & Carpenter, 2008; Owens, Antonoff, & Francis, 1994; Wood et al., 2005), the use of biological motion in this context remains relatively infrequent. While drivers should remain vigilant for the sudden appearance of inconspicuous pedestrians and other objects at night, the use of high-beam headlights can significantly improve drivers’ ability to recognize inconspicuous pedestrians along the roadway. Despite the value of high-beam headlamps at night, existing evidence indicates that drivers tend to underutilize their high beams. In a study by Mefford, Flannagan, and Bogard (2006), participants who drove instrumented vehicles for 7–21 days relied on low beams for 75% of the time that high beams could have been safely used. In other observational studies, drivers used their high beams only 10–50% of the time that high beams could have been used (conditions in which there is no opposing vehicles present, no lead vehicle, and on rural roads) (Hare & Hemion, 1968; Sullivan, Adachi, Mefford, & Flannagan, 2004). Though this tendency to underuse high-beam headlamps has been documented repeatedly, we know less about why drivers choose not to use high beams even in ideal high-beam-use conditions. One possible explanation is that drivers typically believe that the illumination provided by their low beams is sufficient to support safe driving. Indeed, at night, drivers receive continuous feedback that they can see well enough to maintain their vehicle’s lane position but rarely receive feedback about their inability to see low-contrast objects on the roadway. It has been proposed that this ‘selective degradation’ of visual functions at night results in drivers being overconfident at night (Brooks, Tyrrell, & Frank, 2005; Leibowitz & Owens, 1977; Owens & Tyrrell, 1999). There is evidence to suggest that drivers have a limited understanding of their visual deficits in low-illumination conditions (eg Olson & Sivak, 1983; Shinar, 1984). Another possible explanation for drivers underusing their high beams is that they may be reluctant to ‘blind’ (ie be a disabling glare source for) other drivers. Glare can indeed have negative effects on visual performance (ie disability glare) and on driving behaviors (Leibowitz, Tyrrell, Andre, Groetzinger-Eggers, & Nicholson, 1993; Theeuwes, Alferdinck, & Perel, 2002). Headlight glare can create intraocular light scattering, which can decrease drivers’ contrast sensitivity and visibility distance. Intraocular light scattering is more pronounced in older drivers, and as such these drivers may have greater visual difficulties (NHTSA, 2007). Additionally, of course, glare can result in an unpleasant subjective experience (ie discomfort glare). There have been numerous consumer complaints about the discomfort created by encounters with high intensity discharge (HID) headlamps or improperly aimed headlamps (NHTSA, 2001). In particular, HID headlamps have been described by consumers as “too bright”, “painful”, and “blinding”. In a study investigating the subjective experience of glare, 31% of drivers aged 65–74 and 26% of drivers aged 18–25 found oncoming glare to be “disturbing” (NHTSA, 2003). Yet, drivers’ subjective impressions that they are visually disabled by glare may sometimes be inaccurate. In Theeuwes et al. (2002) drivers performed a pedestrian detection task while exposed to headlight glare, and indicated the discomfort they experienced as a result of the glare source. Here, drivers’ subjective ratings of discomfort were not predictive of their performance on the pedestrian detection task. It seems possible that drivers’ subjective experience of discomfort caused by a glare source may not be tightly correlated with objective measures of the degree to which the glare source degrades their visual performance. Drivers who experience discomfort glare may incorrectly assume that their visual performance has been significantly impaired, and the link between measures of discomfort glare and measures of disability glare may vary as a function of the age of the driver.
The accuracy of drivers’ judgments of the effects of headlight glare
1205
To attempt to clarify the relationships among discomfort glare, disability glare, and driving behavior, research has begun to examine the accuracy of drivers’ judgments of their own acuity in the presence of glare. In a lab-based psychophysical study Balk and Tyrrell (2011a) had college-age participants (18–25 years) estimate their own visual acuity (by providing both manual and verbal estimates of the size of a barely discernible acuity target) when facing a glare source. In that study, as the glare intensity increased, observers’ estimates of their acuity significantly worsened. Yet, the glare source that was used was not sufficiently intense to affect their actual acuity. Thus, while experiencing the discomfort of a glare source, the observers exaggerated its disabling effect. In a follow-up study that was conducted on a closed road, an experimenter drove college-aged participants slowly toward a stationary opposing vehicle (the ‘glare vehicle’) and asked the participants to estimate the distance at which they would just be able to recognize a retroreflective Landolt C stimulus that had earlier been visible adjacent to the headlamps of the glare vehicle (Balk & Tyrrell, 2011b). Participants estimated significantly shorter recognition distances when facing the opposing vehicle’s high beams than when the opposing vehicle used low beams. However, the participants’ actual ability to recognize the stimulus was unaffected by the beams of the glare vehicle. Thus, like the observers in the Balk and Tyrrell lab study, these participants exaggerated the disabling effect that the glare source would have on their vision. Together, these two studies provide new perspective on drivers’ judgments about their ability to see at night and new methodologies by which these judgments can be investigated. The extent to which the results of the Balk and Tyrrell experiments generalize to situations in which a driver facing headlight glare encounters an object in the roadway at night is unknown. One reason why headlight glare may not have affected their drivers’ visual performance is that the acuity stimulus featured an artificially high level of contrast as a result of it having been retroreflective and illuminated by headlamps. The retroreflective stimulus may have provided sufficient contrast to remain relatively robust to the negative effects of glare. For this reason, the experiment presented here measured observers’ judgments of their ability to recognize a nonretroreflective stimulus (lower contrast and lower luminance) and actual recognition distance of this stimulus in the presence of glare. Additionally, Balk and Tyrrell tested only college-aged drivers. Because older drivers’ reactions to headlight glare may differ from that of younger drivers, we sought to quantify the accuracy of both younger and older drivers’ judgments of the visual effects of headlamp glare. 2 Materials and method 2.1 Participants Eleven older drivers (M = 71 years, range = 65–80 years; four males) and seventeen younger drivers (M = 19 years, range = 18–21 years; nine males) participated in this experiment. Participants were required to achieve a 20/40 (6/12) or better binocular visual acuity (Bailey– Lovie chart under room illumination), and achieve a contrast sensitivity of 1.65 (Pelli–Robson letter sensitivity chart under room illumination), and report having no known visual pathologies other than those being corrected by lenses that were worn during the experiment. All participants gave written informed consent prior to participation. This study, which adhered to the tenets of the Declaration of Helsinki, was approved by the local Institutional Review Board. 2.2 Test site and vehicles The test site was an unilluminated one-lane utility roadway that was 3.05 m wide and that included a section of straight, level roadway that was 230 m long. The asphalt road was without pavement markings and streetlights, and access to the test site was closed during data collection. Participants were tested at least one hour after sunset and only on nights free of precipitation and fog.
1206
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
The test site is depicted in figure 1. The low beams (xenon) and high beams (low beam xenon + halogen) of a 2008 Infiniti EX35 served as the glare source, and throughout the experiment this vehicle remained parked on one end of the straight portion of road facing the participant vehicle. The glare vehicle was parked 213 m from the starting position of the participant vehicle (a 2004 Nissan Sentra with halogen low beams and high beams). Participants were tested individually, and each sat in the front passenger seat of the participant vehicle. The engines of both vehicles were running throughout data collection. The lateral position of the two vehicles was offset as depicted in figure 1.
Stationary glare vehicle Landolt C stimulus headlight beam Cones used as baseline/visual anchor for participants
headlight beam
Moving participant vehicle
Figure 1. [In color online, see http://dx.doi.org/10.1068/p7814] Configuration of experimental vehicles and Landolt C stimulus.
2.3 Visual stimulus The visual stimulus used in this experiment was a nonretroreflective Landolt C (21.5 cm in diameter and 3.5 cm stroke and gap width) constructed from white paper and mounted on a tripod. The legs of the tripod were masked by black felt. From the participant’s perspective, the center of the Landolt C stimulus was located 80.5 cm above the ground, 1.52 m to the right of the right-most edge of the glare vehicle, and facing the participant at a longitudinal distance that was aligned with the headlamps of the glare vehicle. The lateral position of the two vehicles was such that if the participant vehicle had continued to move forward, it would have passed to the right of the glare vehicle with less than 0.2 m separation. 2.4 Experimental design This experiment featured a 2 (low or high beams of the participant vehicle) × 2 (low or high beams of the glare vehicle) × 2 (response type: estimated or actual recognition) withinsubjects design for younger participants and a separate 2 × 2 × 2 within-subjects design for older participants. To prevent participants’ judgments of recognition distance from being influenced by the measurement of actual recognition distances, estimated recognition distances were always measured before actual recognition distances. 2.5 Estimated recognition distance To establish the baseline distance at which the participant could determine the orientation of the stimulus in the absence of any headlight glare, the participant vehicle was first driven slowly (with engine idling) toward the Landolt C in a no‑glare situation (low beams on
The accuracy of drivers’ judgments of the effects of headlight glare
1207
the participant vehicle and only parking lights on the glare vehicle). When the participant indicated that he or she could discern the orientation of the Landolt C, the vehicle was stopped (all participants accurately reported the orientation of the stimulus at this distance). This baseline distance was marked by orange (nonretroreflective) cones on either side of the roadway. These cones remained in place throughout the experiment to indicate to the participants when they were able to see the stimulus with no glare present. This process also familiarized the participant with the size, location, and appearance of the Landolt C stimulus. After measuring and marking the baseline distance, the Landolt C stimulus and tripod were covered by a large black cloth. Despite the fact that the Landolt C was no longer visible, participants were asked to imagine that it was, and over the course of 8 trials (4 headlamp combinations × 2 travel directions) participant’s estimates of the distance at which he or she would be able to just determine the orientation of the Landolt C stimulus were measured. A Latin square was used to determine the order of the 4 headlamp combinations for each participant, and direction of travel was counterbalanced across all participants. Each trial began by driving the participant to the starting point, which was either 213 m (forward trials) or 8 m (reversing trials) from the stimulus and glare vehicle. The headlamps of the two vehicles were then set to the appropriate setting (low vs high beam) for the trial and then turned on. The participant was reminded that their task was to estimate the distance at which he or she would be able to just determine the orientation of the Landolt C stimulus (if it were still visible). The experimenter then drove the participant slowly (with the brakes off and engine idling) toward or away from the stimulus, and the participant marked each estimated recognition distance by dropping a numbered bean bag out of the passenger window. For reverse-moving trials the participant was asked to drop the bean bag at the point at which he or she estimated that the orientation of the stimulus would no longer be discernible if it were still visible. The next trial began approximately two minutes later; during this delay, the headlamps of the glare vehicle were kept off and the participant vehicle was moved to the appropriate starting point. 2.6 Actual recognition distance After all estimated recognition distances were measured, actual recognition distances were measured for the 4 headlamp combinations and 2 directions of travel. For each participant the sequence of the 8 trials matched the sequence used in the measurement of that participant’s estimated recognition distances. The black cloth was removed from the Landolt C stimulus. Participants were again driven slowly toward or away from the stimulus and this time asked to indicate the point at which they were just able to determine (forward trials) or no longer able to determine (reverse trials) the orientation of the Landolt C stimulus. The participant dropped bean bags from the passenger window to mark the locations. The participant was then asked to indicate the orientation of the C by pointing to a chart that presented the 8 possible orientations (0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315°). A delay of approximately two minutes elapsed between adjacent trials; during this delay the headlamps of the glare vehicle were kept off, the participant’s vehicle was moved to the appropriate starting point for the next trial, and the orientation of the C was changed (to a predetermined random orientation). One of the younger drivers reported an incorrect orientation for one of the trials. This trial was repeated after the eighth trial. All other participants reported correct orientations for all trials. Following all measurements, the experimenters measured the distance of each bean bag from the Landolt C using a measuring wheel. 2.7 Subjective ratings of discomfort glare After each estimated and actual recognition trial, the participant rated the intensity of the discomfort caused by the headlights of the glare vehicle using the de Boer scale (scale ranging from 9, “unnoticeable”, to 1, “unbearable”).
1208
A A Stafford Sewall, S A Whetsel Borzendowski, R A Tyrrell
3 Results The primary purpose of this experiment was to investigate the accuracy of older and younger drivers’ judgments of the effect of glare on their visual performance. Estimated and actual recognition distances were measured from twenty-eight participants (eleven older participants and seventeen younger participants). To examine the accuracy of their judgments, the data were analyzed separately for the two age groups using separate 2 (low or high beams for the participant vehicle) × 2 (low or high beams for the glare vehicle) × 2 (response type: estimated or actual recognition) repeated-measures ANOVAs for each age group. Additionally, for each age group a separate ANOVA analyzed the effects of the beam manipulations on measures of estimated and actual recognition distances. 3.1 Younger participants’ recognition distances The younger participants’ mean initial baseline recognition distance (the position of the cones that marked the distance at which each participant initially recognized the orientation of the stimulus when their vehicle approached using low beams and the glare vehicle used only parking lights) was 47.7 m (standard deviation = 10.7 m). A 2 (beam setting of the participant vehicle) × 2 (beam setting of the glare vehicle) × 2 (response type: estimated and actual recognition distance) repeated-measures ANOVA explored the accuracy of the younger participants’ judgments of recognition distance (see figure 2). On average, estimated recognition distance (M = 44.4 m) was 12% longer than actual recognition distance (M = 39.6 m). However, this main effect was not significant (F1, 16 = 1.9, p > 0.05, hp2 = 0.11); and, further, there were no significant interactions involving the glare vehicle, participant vehicle, or response type on recognition distance ( p > 0.05). This pattern indicates that the younger participants’ estimates of recognition distance were, on the whole, relatively accurate. To analyze the data from the younger group more deeply, separate preplanned 2 (beam setting of participant vehicle) × 2 (beam setting of glare vehicle) repeated-measures ANOVAs analyzed the participants’ estimates of recognition distances and, separately, the measures of the participants’ actual recognition distances. When looking first at estimated recognition, participants’ estimates of recognition distance significantly differed as the glare vehicle’s 80
estimated actual
70
Recognition distance/m
60 50 40 30 20 10 0
glare: low participant: low
glare: low participant: high
glare: high participant: low
glare: high participant: high
Figure 2. Mean estimated and actual recognition distances at each of the four headlamp combinations for younger participants. The horizontal dashed line represents the initial baseline recognition distance (participant vehicle using low beams; glare vehicle using only parking lights).
The accuracy of drivers’ judgments of the effects of headlight glare
1209
headlights changed (F1, 16 = 63.8, p < 0.001, hp2 = 0.80) and as the participant vehicle’s headlights were changed (F1, 16 = 28.5, p 0.05). The actual recognition distances also changed significantly when the headlamps of the glare vehicle were switched to high beams (F1, 16 = 65.2, p
