Useful Visual Field Reduction as a Function of Age and Risk of Accident in Simulated Car Driving

Abstract

purpose. To study the relationships between the reduction of the useful visual field, age, and driving performance.

methods. Forty-eight subjects, aged from 23 to 77 years performed a test to evaluate the size of their useful visual fields. The test involved the detection and localization of peripheral signals that could appear in an area of 70° of visual angle. The subjects then performed a simulated car-driving task involving the management of a situation that could lead to an accident.

results. The analysis of the data revealed that the ability to process peripheral signals and simulated driving performance (vehicle speed) deteriorate with age. Simulated driving performance and useful visual field measurement have been analyzed jointly. The results indicate that the reduction of the useful visual field, estimated using a target-localization task, is related to the individual’s ability to manage the simulated driving situation (correlation coefficient with speed = −0.43 and with reaction time for avoidance of a mobile obstacle = +0.30) and the deterioration of the useful visual field estimated using a target detection task is related only to vehicle speed (correlation coefficient = −0.32).

conclusions. The adoption of a lower speed by the drivers with a reduced visual field (the elderly ones) is probably an adaptation strategy to process the peripheral information. All useful visual field measurements do not seem to be equivalent to estimate the ability to process information relative to the mobile obstacle. The risk of collision should be estimated only with a useful visual field test using a target localization task.

The spatial area of the useful visual field can be measured by asking the subject to stare at a central spot while detecting the display of a peripheral signal at the same time. The farthest eccentricity at which the subject can detect the signal determines his or her visual field. ¹  

The useful visual field is smaller than the peripheral visual field. ^{2 3} It corresponds to the area around the fixation point from which information can be located and extracted quickly during a visual task. ⁴ This field is also called the functional or working visual field. ⁵ It can generally be measured by instructing the subject to perform a dual task: a central task and a peripheral task. The useful visual field corresponds to the part of the peripheral visual field inside which the sources of information can be processed in a visual fixation—that is, without any movement of the eyes or the head. ^{6 7 8} Some investigators assess the useful visual field by simply instructing the subjects to detect the presence of a peripheral signal and identify it, ^{4 9 10 11 12 13 14 15} whereas others demand localization. ^{2 7 16 17 18 19 20 21 22} Ball et al. ⁵ proposes that the limit of the visual field depends on the subject’s ability to locate peripheral signals. They defined the field by locating the farthest eccentricity from which an observer can locate at least 50% of the peripheral targets presented. ⁵ In the present study, the size of the useful visual field was evaluated through a double activity: the detection and then the localization of the peripheral signals. A dual visual task was proposed, first involving detection and then localization of signals appearing in the visual field. The size of the useful visual field is not constant. It depends mainly on the nature of the perception task and on the situation. The characteristics of the visual test also affect the size of the useful visual field. For example, central task complexity, peripheral task complexity, and priority given to the central task, as opposed to the peripheral task, can reduce the useful visual field. ^{6 14 16 21 23 24 25 26 27} It can also depend on individual characteristics, such as age. 

During recent years, several visual indices noted in drivers, such as acuity, contrast sensitivity, color discrimination, peripheral vision, and visual attention have been studied. The problems related to these indices have been associated with older drivers’ involvement in car accidents. ^{5 28 29} Among the different visual indices, the measurement of the useful visual field seems to be the factor most related to road accidents in older drivers. ^{28 30 31 32 33} Ball et al. ⁵ developed a test: the UFOV (useful field of view) Visual Attention Analyzer, or UFOV test. ⁵ It is used to assess three aptitudes for processing (divided attention, selective attention, and speed of processing), which are studied using spatial localization tasks. 

Several researchers have noted that the risk of accident while driving is significantly higher in older people when the useful visual field is reduced by 40% or more. ^{5 29 34} Until now, the experimental approach has consisted of a study of the relationship between the reduction of the useful visual field and the number of accidents in real situations, recorded over a period of several years before ^{2 5 28 29 34} or after ^{28 31} the UFOV test. Only Myers et al. ³³ proposed that older drivers undergo a real driving test, marked by two examiners in addition to a UFOV test, for comparison. Poor performance on the UFOV test was associated with a high number of driving errors (failing to stop at a stop sign, missing important road signs, making errors of judgment or taking a wrong position on the road). It should be noted that this evaluation took place under similar driving conditions for all drivers. Furthermore, the examiners took into account several types of errors in their assessments, with the result that the decrease of the useful visual field could be related to one, several, or all aspects of the driving task. This is why we studied only one aspect of driving: the ability to control the trajectory when avoiding a mobile obstacle (a truck that slowly entered the roadway in front of the driver). We propose that there is a relationship between reduction in the extent of useful visual field and performance in avoiding peripheral objects while driving. Driving was performed in a simulator, to have the same conditions for each subject (identical road and lighting, same trajectory of the truck at the same time). Furthermore, the situation did not involve any danger for the subject in the case of a failure. 

Methods

Subjects

The sample consisted of 48 adults aged between 23 and 77 years, arranged in three groups: 17 younger people aged between 23 and 39 years (mean ± SD, 29.7 ± 4.4), 15 middle-aged adults between 45 and 51 years (mean, 49.4 ± 3.3), and 16 elderly persons aged between 65 and 77 years (mean, 69.5 ± 3.3). Their sight was normal or corrected to normal (binocular visual acuity > 8/10). A manual perimeter, Gamb’s test, ³⁵ was used to check that the visual field of the subjects was normal (that is to say, whether the subject was able to perceive a signal on the whole of his or her visual field). In this equipment, the subject focused on a point in front of him or her while detecting a spot (eccentricity: 0.21°; contrast: 0.32) moving (speed of motion: 5°/s) along different vertices (0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, and 330). The research adhered to the tenets of the Declaration of Helsinki, received the approval of COPE (French Ethics Operational Committee which is the ethic committee of the French National Centre of Scientific Research), and was accepted by the French Ministry of Health. All subjects gave informed consent before participating in the research after explanation of the nature and possible consequences of the study. 

Device and Variables

The useful visual field test was performed on the Useful Field Measuring Device (PECVU; Poste d’Etude du Champ Visuel Utile). ³⁶ The subject was seated in the center of a hemicylindrical panel with a monitor screen in the middle. The luminance of the panel was 1.8 cd/m². The central task was performed on the monitor screen (luminosity of the background: 1.7 cd/m²). Eight spots were displayed in a circle (eccentricity: 1°) for 100 ms, and from time to time (in 10% of the trials), a spot more luminous than the seven others appeared. The subject had to detect and report the presence of this more luminous spot, as quickly and accurately as possible. The peripheral signals appeared on the 2-m high hemicylindrical panel encompassing 140° horizontally. This panel was covered with 108 green-light-emitting diodes (0.5 cm in diameter, eccentricity: 0.29°) distributed along arcs of circles whose center was the same as that of the monitor screen. The LEDs could be turned on for 100 ms randomly in time. A peripheral signal never occurred simultaneously with a critical signal in the central task. When the subject perceived a luminous peripheral signal or a central signal, he or she had to press the buttons on a joystick, one of which was assigned to the central task and the other to the peripheral task. If the subject perceived a peripheral signal, he had to locate it using a light-pointer, which was detected by a small photocell beside each LED. Then, 3 seconds later, the test started again. The instruction was (1) to detect the signal as quickly and accurately as possible and (2) to locate it precisely. No priority was given in the instructions to the central task or to the peripheral task. 

To assess the reduction of the useful visual field, we used the same criteria as Ball et al. ⁵ The farthest eccentricity from which the subject perceived at least 50% of the signals was considered to be the limit of the useful visual field. If the limit was located at 70° (horizontal extent was 140°), the subject’s visual field was complete. If the limit was located at 60°, 50°, 40°, 30°, or 20°, then the restriction was, respectively, 14.3%, 28.6%, 42.9%, 57.1%, or 71.4%. We define a detection useful visual field as when the reduction is assessed from the percentage of signals detected and a localization useful visual field as when the reduction is estimated from the percentage of signals localized (correct meridian and correct eccentricity). 

The subjects also performed a simulated car-driving task on the Vigilance Analysis Driving Simulator (PAVCAS [Poste d’Analyze de la Vigilance en Conduite Automobile Simulée]). This consists of a car cabin placed on a mobile base that enables longitudinal, vertical, pitching, and rolling movements linked to an interactive display unit. The road image was placed 3.50 m away from the driver and covered 45° of the horizontal and 25° of the vertical visual field. The instruction was to drive and respect the highway code. The circuit consisted of a 50-km straight and curved roadway. It included several critical areas that reproduced scenarios often met on a roadway. The subject had to follow another vehicle, overtake a truck that had stopped on the emergency lane, or avoid a truck entering the roadway. This last event was analyzed in the study. A truck stopped on a lay-by (located on the right side of the road), moved off, and entered the roadway at the moment when the subject arrived. The truck accelerated to reach a speed of 110 km/h which allowed considerable time for the subject to overtake it. Its trajectory and its speed were strictly the same for all the subjects. This event seems to be a situation that could lead to an accident if the visual sources of information related to the movement of the mobile obstacle are not quickly processed. 

The first driving index was the time necessary for the driver to avoid a potential collision with the truck. This reaction is shown by an increase in the lateral deviation in his or her trajectory (Fig. 1) . The second index was the time to collision calculated at the moment of the avoidance reaction. It corresponds to the time taken by the driver to reach the kilometric entry point of the truck on the roadway if the vehicle maintains its speed and a constant course. The third index was the speed of the subject’s vehicle calculated according to the data recorded during the 5 seconds before the truck’s departure. 

Procedure

Before driving, the subject performed the useful visual field test on the Useful Visual Field Measuring Device. He or she underwent a training period for the central task for 3 minutes, for the peripheral LED-detection activity for 5 minutes, and for both tasks at the same time for another 5 minutes. During this last period he or she also had to locate the signal perceived. Then, he or she performed the 20-minute test that included both the detection and the localization activities. After the test on the Useful Visual Field Measuring Device, the subject performed a simulated car-driving task. The situation that could lead to an accident (truck entering the roadway) occurred about 10 minutes after the start of his or her driving session. 

A Kruskal-Wallis nonparametric analysis of variance was performed for the reduction of the detected useful visual field and the reduction of the localized visual field, because the collected data were not normally distributed. The analysis of the reduction of useful visual field takes age into account (three groups: young subjects, middle-aged subjects, and older subjects). The comparisons between means were performed with the Siegel method. ³⁷ An analysis of variance was performed for each dependent variable related to driving in the simulator: elapsed time when avoidance began, time to collision, and speed. The corresponding data were normally distributed (all Kolmogorov-Smirnov test results were not significant). The analysis of these dependent variables takes age into account (three groups: younger subjects, middle-aged subjects, and older subjects). The comparisons between means were performed with the Newman-Keuls test. Spearman correlation coefficients were calculated between driving performances (time of the avoidance reaction, time to collision, and speed), the reduction in the useful visual field (measured with detection and localization tasks), and age (processed as a continuous variable). All these statistical analyses were conducted on computer (Statisca ver. 5.5; StatSoft). Comparisons between means and correlation coefficients were considered to be significantly different when the probability of error was ≤0.05. 

Results

Analysis of the Data Obtained in the PECVU Test: Measurement of the Reduction of the Useful Visual Field

Age had a significant effect on the reduction of the useful-detection visual field (χ² ₍₂₎ = 26.34; P < 0.001). This reduction was less significant in the younger and the middle-aged adults (means ± SD, 26.9% ± 11.2% and 30.5% ± 14.1%, respectively) than in the older adults (mean, 59.8% ± 5.8%). Age also had a significant effect (χ² ₍₂₎ = 30.78; P < 0.001) on the reduction of the localization useful visual field. The younger adults had a less-reduced field (mean, 30.3% ± 9.9%) than the middle-aged ones (mean, 41.9% ± 13.7%). Also, the field of the younger adults and the middle aged is less reduced than the field of the older adults (mean, 58.9% ± 4.9%). 

The correlation between the reduction of the detection useful visual field, the reduction of the localization useful visual field, and the age of the subjects are presented in Table 1 . 

The reduction of the detection useful visual field significantly correlated with the reduction of the localization useful visual field. The more significant the reduction of the detection useful visual field, the more limited the localization useful visual field (Table 1 , second line, second column). Furthermore, age correlated with the size of the detection useful visual field and with the size of the localization useful visual field. The older the individual, the more significant the visual reduction (Table 1 , third and fourth lines, second column). 

Analysis of the Data Obtained in the Simulated Driving Task

One older subject could not undergo the driving test for technical reasons related to the simulator. A second older subject had to be excluded from the driving data analysis, because he began to follow the truck instead of moving in the other lane to overtake it. Therefore, he could not perform the avoidance maneuver. The analysis presented herein was performed with a sample of 46 subjects. 

Reaction Time to Avoid the Truck, Time to Collision, and Vehicle Speed.

The analysis of the reaction time necessary to avoid the truck indicates that age had no effect. The reaction time was 3.970 ± 2.25 seconds (SD) for the younger subjects, 5.752 ± 4.28 seconds for the middle-aged subjects, and 7.054 ± 4.25 seconds for the older subjects. Age had no significant effect on time to collision. This time was 4.3 ± 2.18 seconds for the younger subjects, 2.8 ± 4.18 seconds for the middle-aged subjects, and 1.6 ± 4.30 seconds for the older subjects. We also note that the reaction time increased with age and time to collision decreased with age, even though the effect of age was not statistically significant on these data. Age had a significant effect on vehicle speed (F_(1,43) = 7.11; P = 0.002). The younger subjects drove faster (mean, 37.4 ± 1.98 m/s), than the middle-aged and the older subjects (means 35.03 ± 2.71 and 34.61 ± 2.02 m/s, respectively). 

Joint Analysis of the Useful Visual Field Reduction, Simulated Driving Indices, and Age.

The reduction of the detection and localization useful visual field was related to each driving index. The correlation coefficients obtained and the error probability threshold are presented in Table 2 . 

The more significant the useful visual field reduction, the slower the subjects drove. This relationship was found whether the useful visual field was measured using a detection or a localization task (Table 2 : second line, second and third columns). The more significant the useful visual field reduction, the longer the time to perform the avoidance maneuver (Table 2 : third line, third column). The more significant the drivers’ useful visual field reduction, the slower they are to implement an action that would enable avoidance of an accident (collision). It should be noted that this result was only observed when the useful visual field reduction was measured by using a localization task. The collision time did not correlate with the useful visual field reduction (Table 2 : fourth line, second and third columns). Finally, age correlated with all driving parameters. The older the individual, the lower his or her speed, the longer his or her reaction time to avoid the truck, and the shorter his or her time to collision (Table 2 : fourth column). 

The degradation of the driving performance (speed and reaction time for avoidance) can be explained by the reduction of the useful visual field, which would be the factor of risk (Figs. 2 3) . The relationship between the reduction of the useful visual field and these driving performances can also be explained by the age of the drivers (which would then be a confounding factor). Indeed, age and driving performance covaried. 

To determine whether age is a confounding factor, we performed a stratified analysis of the data. The age, the reduction of the useful visual field, and the driving indices were transformed into dichotomous qualitative variables to calculate the Mantel-Haenszel test (odds ratio and χ²). The two strata for age were the youngest drivers (age ≤ 48 years, median of the data of the experimental group) and the oldest ones (age > 48 years). The driver had a correct useful visual field if the reduction of the field was <28.6% and a reduced field if the reduction was ≥28.6%. Finally, subjects with good driving performances took up a speed ≥36 m/s (maximum speed limit authorized in France) and reacted rapidly to avoid the obstacle (reaction time for avoidance < 4.4 seconds, median of the data of the experimental group). The specific effect of the reduction of the useful visual field on the quality of driving is calculated for each stratum of the age variable as well as the relative common risk, adjusted according to age (Table 3) . 

There is therefore a relationship between the reduction of the useful visual field and speed, and age is not a confounding factor (significant χ² with an error probability of 0.05 for the reduction of the useful visual field and significant χ² with an error probability of 0.07 for the reduction of the localization useful visual field). However, the relationship between the reduction of the useful visual field and the speed of the avoidance reaction to the truck is not independent from age, which is a confounding factor (not significant χ²). 

Discussion

Age correlated with the reduction of the detection useful visual field and with that of the localization useful visual field. Furthermore, the significant correlation between the reduction of both fields led to the conclusion that the assessment of the useful visual field can be performed equally with either a detection task or a localization task. 

The joint analysis of the simulated driving data and of the data relative to the reduction of useful visual field gave an interesting result that was observed only when the reduction of the useful visual field was assessed by using a localization task. The more significant the reduction of the useful visual field, the longer it took the driver to avoid the truck. This result is consistent with our hypothesis: There is a relationship between reduction in the extent of useful visual field and performance in avoiding peripheral objects while driving. A first explanation for this result can be proposed. The avoidance reaction may come later in subjects with a reduced visual field, simply because when they drive slower, the risk of a collision with the truck is lower. In other words, the drivers with a reduced useful visual field would benefit from more time to reach the point of collision. This explanation must be rejected, however, because the collision time does not correlate significantly with the reduction of the useful visual field. A second explanation can be proposed. The drivers with a reduced visual field may wrongly analyze the movement of the truck with a trajectory that is tangent to theirs. As long as they do not swerve into the other lane, they remain in a situation that can lead to an accident (for example, a collision caused by the truck’s braking suddenly). It should be noted that the relationship between the reaction of avoidance and the useful visual field is only found if the measurement of this field is made with a localization task and is not independent of the age of the driver according the results obtained with the stratified analysis. The movement of the truck (speed, acceleration, direction) and the anticipation of its trajectory may require a precise and rapid localization of signals in peripheral vision. 

The driver’s speed is not independent of the reduction of his or her useful visual field. The more reduced the field, the slower he or she drives. The position assumed by Mackworth ³⁸ enables us to propose an interpretation for this result. According to Mackworth, ^{4 38} the size of the useful visual field depends directly on the quantity of information to be processed. The larger the quantity of information, the larger the shrinkage of the useful visual field. While one drives, the quantity of information to be processed for each unit of time depends directly on the driving speed. The higher the speed, the larger the quantity of information to be processed for each unit of time. We can therefore consider the adoption of a lower speed by the individuals with a reduced visual field as an adaptation strategy, and this strategy is independent of age according the results obtained with the stratified analysis. These drivers seem to slow down to be able to process properly and quickly the peripheral visual sources of information. 

We had not proposed any hypothesis about a relationship between age and the driving indices. However, we have noted that the time taken to perform the avoidance maneuver increased with age, and the time of collision with the entry point of the truck decreased with age. A short collision time and delay in performing the avoidance maneuver naturally exposed the older drivers to a high risk of collision. If the avoidance reaction in real driving is systematically slower in occurring in older drivers, they are necessarily more often exposed to situations that can lead to an accident than others. In that case, it is logical to observe a higher rate of accidents in that population, a result often cited by Ball et al. ⁵ and Owsley et al. ³¹  

Conclusion

The whole of the results obtained in this experiment are consistent with a greater involvement of older drivers in road accidents. The relationship between the reduction of the useful visual field and the simulated driving indices shows the relevance of a test measuring this field in drivers. A complete useful visual field seems indeed to be a major element in the driver’s ability to control his or her trajectory in a simulated situation that can lead to a collision. All useful visual field measurements do not seem to be equivalent. The localization useful visual field is only related to the driver’s ability to process the visual information relative to the movement of a vehicle that arrives sideways. As to the ability to process static visual information in the road scene, one wonders whether it is related to the detection useful visual field. 

 Submitted for publication May 14, 2004; revised July 13 and September 28, 2004; accepted October 5, 2004.

 Disclosure: J. Rogé, None; T. Pébayle, None; A. Campagne, None; A. Muzet, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Lateral position of an older subject’s vehicle as a function of time. The subject was driving normally in a right lane. In this lane, his lateral position could vary from 0 to –3.6 m. The x-axis represents the central line. The direction is defined as the angle between the vehicle’s trajectory and the x-axis. Therefore, if the vehicle’s trajectory is parallel to the x-axis, its direction is zero. (A) The truck appeared in the on-ramp at this time index. (B) The subject was going toward the left lane at this time index, which is determined by the maximum value of the vehicle’s direction before the crossing of the central line. (C) The subject started avoiding the truck at this time index, which is defined as follows. We evaluated the mean of directions for each period of two seconds preceding (B). The start of avoidance was defined as the time index. This mean was below a threshold (one third of the highest vehicle’s direction before crossing the central line). Hence, the reaction time to avoid the truck is defined as C − A.

Table 1.

 

Correlation Coefficients between the Reduction of the Detection Useful Visual Field, the Reduction of the Localization Useful Visual Field, and Age

Correlation	Correlation Coefficient (95% CI)
Of the reduction of the detection UVF and the reduction of the localization UVF	+0.88 to (+0.80 to +0.93)
Of the reduction of the detection UVF and age	+0.67 to (+0.48 to +0.80)
Of the reduction of the localization UVF and age	+0.74 to (+0.58 to +0.84)

 UVF, useful visual field.

Table 2.

 

Correlation Coefficients between Useful Visual Field Reduction, age, and Driving Indices

	Reduction of Detection UVF (95% CI)	Reduction of Localization UVF (95% CI)	Age (95% CI)
Speed	−0.32 to (−0.05 to −0.54)	−0.43 (−0.17 to −0.63)	−0.54 (−0.31 to −0.71)
Reaction time for avoidance	NS	+0.30 (+0.03 to +0.53)	+0.39 (+0.13 to +0.60)
Time to collision	NS	NS	−0.30 (−0.03 to −0.53)

 UVF, useful visual field.

Figure 2.

 

Speed as a function of the reduction of detection useful visual field (UVF). (□) Older drivers (age > 48 years); (▪) younger drivers (age ≤ 48 years).

Figure 3.

 

Avoidance reaction as a function of the reduction of localization useful visual field (UVF). (□) Older drivers (age > 48 years); (▪) younger drivers (age ≤ 48 years).

Table 3.

 

Odds Ratio Adjusted for Age and 95% CI (Woolf Method) for the Relationship between Reduction of the Useful Visual Field and the Driving Experience

Driving Parameters	Younger Drivers			Older Drivers			Common Relative Risk
		Odds Ratio	95% CI	Odds Ratio	95% CI	Odds Ratio	χ2 (1)	P
Reduction of detection UVF speed	0.14	0.02, 1.06	0.48	0.07, 3.16	0.28	3.6	0.05
Reduction of localization UVF speed	0.14	0.02, 0.94	0.92	0.07, 12.4	0.28	3.1	0.07
Reduction of localization UVF RT for avoidance	0.60	0.11, 3.34	0.23	0.02, 3.06	0.45	0.45	0.26

 UVF, useful visual field; RT, reaction time.