Eyes, Brains, and Autos

Vision disorders pose a driving safety risk and commonly arise at thelevel of the eye in cataract, glaucoma, macular degeneration, and diabeticretinopathy and at the level of the brain in advancing age, stroke, and Alzheimerdisease and related conditions. These disorders can increase driver safetyerrors because of reduced visual acuity, contrast sensitivity, and visualfields. Aging and brain lesions, especially, can also reduce the useful fieldof view in drivers with normal visual fields; increase the attentional blinkand change blindness; impair perception of structure and depth from visualmotion cues and motion parallax; decrease perception of heading from opticalflow and detection of impending collisions; and increase the chance of gettinglost. Better tools are needed for detecting and alerting visually impaireddrivers who are at greatest risk for a crash. These drivers can be assessedwith a state-administered road test, instrumented vehicles, and driving simulators.Emerging safety interventions include optical and electronic visual aids forvisually impaired drivers, coupled with new vehicle designs, onboard warningdevices, and reflective clothing that highlights the motion of pedestrians.

Vision is critically important for safe automobile driving.¹ Drivers must monitor multiple inputs from centraland peripheral vision and the other senses, allocate attention among onboardand roadway targets and distracters, and rapidly correct their safety errorswhile also monitoring personal fatigue or other incapacities that might compromisesafety. Drivers with impairments of visual perception, visual cognition, andvisual memory are liable to make poor decisions based on faulty inputs, increasingtheir risk of errors, crashes, and injuries.

In 1968, Burg² reported that impairmentsof visual acuity, glare recovery, and visual fields were associated with increasednumbers of crashes and traffic violations (in 3-year records) of 17 500California driver's license applicants. Yet, driving safety cutoff scoresfor common measures such as visual acuity and visual fields remain uncleardecades later. Visual functions such as attention, motion, and object perceptionand other aspects of visual cognition that may affect driver safety are seldommeasured with standard clinical visual assessment tools.

The American Academy of Ophthalmology (AAO), San Francisco, Calif, recognizesthat ophthalmologists must often assess visual functions for driving licensurein patients with cataract, diabetic retinopathy, glaucoma, macular degeneration,refractive error, and other visual disorders. Yet AAO policies approved inOctober 2001 found few reliable visual criteria for discriminating betweensafe and unsafe drivers.³ Practitioners andpatients face arbitrary state policies for driver restriction with visualacuity cutoffs ranging from 20/40 to 20/200. This article reviews currentpolicies, tests, and countermeasures in drivers with visual impairment dueto lesion of the eye or brain.

In many states, visual acuity is the only visual criterion used in issuinga driver's license. Letter acuity can be easily measured using the Early TreatmentDiabetic Retinopathy Study chart.⁴ This commontool for assessing pattern vision has been used in several major clinicaltrials sponsored by the National Eye Institute, Bethesda, Md. Many practitionerswould likely agree that visual impairments of 20/200 should preclude driving,but how should we judge driving fitness in patients with less severe visualimpairments? The 20/40 cutoff score in many states may be unfair and is nolonger supported by the AAO.

Spatial contrast sensitivity, which is the ability to perceive patterns(such as gratings or letters) presented at different contrasts, may be a betterpredictor of driving competency than visual acuity. Contrast sensitivity iseasy to measure (left eye or right eye or both eyes) using a wall chart⁵ that provides a measure of low to medium spatial frequencysensitivity (ie, near the peak of the human contrast-sensitivity function).Owsley et al⁶ studied police-reported crashrates in treated and untreated patients with cataracts in one or both eyesand a visual acuity of 20/40 or worse. The treated group had slightly lessthan half the crash rate of the untreated group on 4-year follow-up. Differencesin crash rates in treated and untreated patients with cataracts were predictedby contrast sensitivity scores but not visual acuity scores. Surprisingly,contrast sensitivity is not a licensing parameter in any state. Contrast sensitivitycan also be measured under standard photopic conditions and under low-visibilityconditions by viewing the contrast sensitivity chart through low-transmittancefilters. This approach may be useful for screening drivers who have troubleseeing in low-light conditions, but it remains a research issue.

Drivers with macular degeneration have visual field loss primarily inareas of high-detail vision around fixation and may be included in discussionsof drivers with abnormal visual acuity and contrast sensitivity. A study ofdrivers with Stargardt disease and cone-rod dystrophy affecting their centralvision showed an increased crash risk only for night driving when comparedwith control subjects.⁷ Drivers who are awareof their visual defect may compensate well enough to drive safely during thedaytime. Some states allow such drivers with reduced visual acuity the chanceto demonstrate driving competency in a road test.

Glare can impair road visibility on sunny days in the summer, in thewinter from reflections off the snow, and at night from the headlights ofoncoming cars. Glare discomfort is a complex measure that may or may not berelated to driver performance impairment. Glare disability is tied to performanceimpairment and is more directly relevant to driver safety. Laser assistedin situ keratomileusis surgery and radial keratotomy may increase glare discomfortand disability. Older drivers with cataracts may report rings of glare aroundglowing objects like lamps and headlights. Whether these phenomena affectdriver safety is a research question. These drivers may already have impairedlow-luminance vision, independent of glare. We recently found that glare disabilitycorrelated with visual attention impairments in older individuals, suggestinga decreased ability to separate visual signals from noise created by extraneousglare.⁸ Glare disability can be tested usinga commercially available instrument, the Mentor Brightness Acuity Tester (MarcoInc, Jacksonville, Fla),⁹ which floods a viewer'seye with light as the viewer tries to read a visual acuity chart.

Color cues allow us to parse information in scenes from chromatic boundaries.These cues increase the recognition of targets in natural scenes amid glare,shadows, and camouflage, which reduce conspicuity and mask object borders.^10,11 While these roles would seem to behighly relevant to the driving task, the AAO does not recommend color-visiontesting in driver assessments. This is because studies to date show no associationbetween color-vision impairment and reduced driving performance and color-impaireddrivers can use cues other than color. Note that traffic signals are oftenmounted in a standard vertical order so that a knowledgeable driver can inferstop (red on top), go (green on bottom), and caution (yellow in the middle)despite color-vision loss.¹² However, in someplaces, traffic lights are arranged horizontally. Future traffic devices mayuse single, large, energy-saving LED lamps that show yellow, red, or greenand eliminate position cues. Color-vision testing might then assume greaterimportance.

Visual field defects, which may arise at the level of the eye or thebrain, are a common reason for considering whether a patient is fit to drive.The effects on automobile driving depend on the location of the defect withinthe visual fields and the specific types of processes affected within theabnormal region. The many possible degrees of visual field loss correspondto different lesions in the visual pathways.

Some individuals with acquired visual field defects may experience a"hole" in their vision.¹³ The added task ofhaving to remember to search for critical information in the areas of impairedvision might create an extra cognitive load or interference, tantamount tothe burden of multitasking.

Briefly, the binocular visual fields normally subtend more than 180°across. The fovea subtends about 3° and has the highest visual acuity.The macula or parafovea spans about 10° and also participates in visualtasks that demand fine visual resolution such as reading maps, road signs,dials, and gauges. The peripheral visual fields extend beyond this and havelow visual acuity but good temporal resolution and motion detection. Few wouldargue that keyhole or tunnel vision with fields spanning 20° would bea contraindication to licensure, but there is a vast gap between this andfull monocular or binocular fields.

Wood and Troutbeck^14,15 foundthat young adult drivers whose binocular visual fields were acutely constrictedto 40° or less by wearing goggles had trouble identifying road signs andnavigating and needed more time to complete a test-track drive. It is unclearhow these acute findings apply to real patients with chronic visual defects.Council and Allen¹⁶ found no increased overallcrash risk in drivers with binocular visual field restrictions lower than140°, yet less than 1% of the drivers had restrictions lower than 120°.Field-restricted drivers had a higher proportion of side collisions. The studydid not include formal field mapping or correlation to the shape of the fieldrestriction. North¹⁷ thought that lack of correlationbetween visual field loss and driving problems reflected methodological flawsin previous studies or neural recovery and compensatory strategies by driverswith an acquired field loss.

Johnson and Keltner¹⁸ screened the visualfields of 10 000 volunteers (20 000 eyes). Drivers with binocularvisual field loss had up to twice the crash and traffic violation rates ofthose with normal visual fields. Drivers with monocular visual loss had crashand conviction rates equivalent to those of a control group. Most subjectswith visual field defects had glaucoma, retinal disorders, or cataracts (andthe number who had homonymous hemianopia was not specified).

Drivers with retinitis pigmentosa may have marked constriction of theperipheral visual fields causing them to be unable to detect objects approachingfrom the side. Fishman et al¹⁹ found a greaterlikelihood of crashes in 42 patients with progressive constriction of thevisual fields due to retinitis pigmentosa than in 87 control subjects of similarage. Szlyk et al²⁰ found more crashes bothin driving simulator tasks and state records during the preceding 5 yearsin 21 drivers with retinitis pigmentosa than in 31 healthy control subjects.Visual field size was the best predictor of real-world and simulator crashes.

In short, when it comes to visual field loss and driving, size matters.Severe binocular visual field loss elevates driver crash risk, but subtlevisual field impairment alone is unlikely to play a significant role.²¹

There is a paucity of research on driving performance and crash riskdue to cerebral visual field loss. Lesions of the primary visual cortex (locatedin Brodmann area 17, aka striate cortex or area V1) or white matter producedefects in the visual fields opposite the side of the lesion. These defectsare homonymous (they occupy the same hemifield ineach eye because of the reversal of real-world images by the lens and crossingof nasal fibers of the optic nerve) and congruent (meaningthe defects in the 2 eyes are nearly identical when superimposed).

Hemianopia refers to loss of half of the visual field. Drivers withhemianopia cannot see objects on one side of fixation. A visual field defectthat is restricted to the upper or lower quadrant of a hemifield is knownas a quadrantanopia. A lesion below the calcarine fissure results in an upperquadrantanopia. A lesion above the calcarine fissure causes a lower quadrantanopiaand may have a greater effect on automobile driving because the lower visualquadrants normally possess better attentional resolution than the upper²² and are better positioned for searching the roadwaypanorama, vehicle controls, and displays. Damage to the macular representationin V1 is troublesome because it may interfere with ocular fixation, visualscanning, and the ability to process visual spatial details (see "Visual Acuityand Contrast Sensitivity" subsection).

Because of a lack of evidence on driving performance in patients withhemianopia, some US states are unable to disclose the criteria they use forlicensing these drivers. Under many Australian and European rules, the diagnosisof hemianopia precludes holding any driver's license. Belgium has relied onmedical experts to decide on the licensure of drivers with cerebral visualloss.²³ Under such policies, a driver may comeunder scrutiny for reasons unrelated to a visual defect and face license lossdespite having driven safely for years or individuals with hemianopia sincechildhood may be denied a driving learner's permit despite having developedadequate compensatory mechanisms.

Paris et al²⁴ reviewed the records of60 Canadian drivers with homonymous hemianopia. At first, 57% did not meetthe minimum field requirements for an unrestricted license in any provinceor territory. Forty percent had some functional recovery, mostly during thefirst 6 months after initial diagnosis. Paris et al emphasized that varioustypes of homonymous hemianopia may have different driving risks and that healthcare personnel should advise the driver of the potential risks. The problemis that there are no consistent performance-based safety standards to guidethis advice.

Tant et al²⁵ studied driving performancein 28 drivers with chronic homonymous hemianopia (15 left, 13 right). Of these28 drivers, 21 had quit driving. Driving evaluators rated reduced steeringcontrol as the most noticeable effect of hemianopia, but some individualswith hemianopia appeared to be safe drivers. Patients with hemianopia mayadopt visual search strategies to successfully compensate for their defects.Hemianopia with macular involvement would seem to pose the greatest driversafety risk, but this remains a research issue.

Stroke, trauma, and tumor commonly cause the cerebral lesions that producevisual field defects. These lesions often extend into the prestriate cortex(Brodmann areas 18 and 19 or area V2/V3) and adjacent temporal lobe and parietallobe. The resulting defects tend to be incomplete and less well localizedthan those caused by V1 lesions. These defects have been explained using thesimple heuristic device of parallel processing in 2 visual systems originatingin V1: a "what" pathway and a "where" pathway. Lesions in these pathways canimpair visual processes important to automobile driving independent of V1-typevisual field defects.²⁶

Briefly, damage in the ventral occipital lobe and adjacent temporalregions along a "what" pathway is associated with defects of visual recognition(visual agnosia), color perception (cerebral achromatopsia), and reading (acquiredalexia). These conditions can impair driving performance, even in the absenceof a visual field defect. For example, drivers with visual agnosia may havedriving safety problems due to the inability to recognize the meaning of hazardousobjects, and drivers with alexia may have difficulty reading road signs andmaps.

Damage along the occipital-parietal "where" pathway is associated withdefects of visually guided eye and hand control and disordered visuospatialattention and impaired motion processing (cerebral akinetopsia). Patientswith Bálint syndrome, often associated with bilateral parietal lobelesions (due to stroke or a visual variant of Alzheimer disease), have severereductions of visual attention²⁷ and cannotdrive safely. Some of these patients are looking but not seeing²⁸ (Figure 1). Patients with hemineglect, a neurologicalsyndrome most often associated with a lesion of the right parietal lobe, oftenfail to attend to stimuli in the left hemifield, whether they have a lefthomonymous hemianopia, and also should not drive (see "Visual Attention andDriving" subsection).

Information from parallel pathways is also processed outside of thevisual cortex. Damage to the prefrontal cortex may impair mechanisms for "executiveattention" and working memory (see "Visual Attention and Driving" subsection)that briefly maintain visual information (such as the location and identityof other vehicles near the driver's car) so that it is available for use.²⁹ Damage to the cerebellum may impair neural mechanismsthat distinguish between image movement across the retina and self-movement,^30-32 which are importantfor perception of heading, collision detection, and related abilities.

Visual attention is of key importance to automobile driving. Standardperimetry tasks (such as Goldmann and Humphrey perimetry) minimize attentiondemands to gain maximal estimates of sensory ability, leading to overestimatesof functional ability in elderly individuals and persons with brain damageengaging in real-world tasks that demand peripheral vision.^33,34 Reductionin the useful field of view, the visual area from which information can beacquired without moving the eyes or head, correlates with increased vehiclecrash risk.^35,36 Useful fieldof view task performance depends on speed of processing, divided attention,and selective attention. The attended field of view is similar to the usefulfield of view except drivers are allowed to move their eyes and head.³⁷ The efficiency with which drivers can extract informationfrom a cluttered scene (such as a busy traffic intersection) begins to deteriorateby 20 years of age.³⁸

Change blindness is an example of a type of blindness that occurs evenin persons with normal vision and is the inability to detect critical changesin a scene because of a brief visual disruption. The disruptions can includesaccades, flickers, blinks, camera cuts,³⁹ orgradual image changes.⁴⁰ Change blindness probablydepends on visual working memory and spatial attention.⁴¹ Changeblindness is more likely when working memory is occupied by other informationor working memory capacity or duration is impaired (eg, because of aging,neurological disease, drugs, or fatigue), and it reduces the ability to perceivesalient changes in traffic-related scenes.^42,43

The attentional blink is another type of blindness that can occur inpeople with normal vision. When we identify a visual object, our ability toperceive a second object is impaired for several hundred milliseconds (becausevisual working memory is still occupied by the first object when the secondarrives). This period, known as the attentional blink, is not due to an eyeblink and can be measured in a laboratory setting using a rapid, serial visualpresentation of visual targets (often a sequence of letters) on a computermonitor. The attentional blink can increase pathologically because of reducedtemporal processing speed and working memory in patients with a variety ofbrain lesions.⁴⁴ Increased attentional blinkmay impair a driver's ability to perceive information from a continuous streamof signs, lights, roadway obstacles, and other vehicles.

Safe driving also requires executive attention to switch the focus ofattention among critical tasks such as tracking the road terrain; monitoringthe changing locations of neighboring vehicles; reading signs, maps, trafficsignals, and dashboard displays; and checking the mirrors.⁴⁵ Thisinvolves switching attention between disparate spatial locations, local andglobal object details, and different visual tasks and is thought to rely onmechanisms in the prefrontal areas.⁴⁶ Notethat cell-phone conversation diverts the brain's attention to an engagingcognitive context and interferes with visual demands of driving.⁴⁷ Amultitude of gauges, dials, radio controls, and "infotainment" displays inmodern vehicles provide similar interference.

There is little evidence that impairments of binocular stereopsis, whichmay affect up to 10% of the general population, affect driving safety. Thisis probably because information on object structure and depth is so criticalfor interacting with objects and obstacles that our brains use multiple redundantcues.⁴⁸ These cues include accommodation, convergence,binocular disparity, motion parallax, texture accretion/deletion, convergenceof parallels, position relative to the horizon, relative size, familiar size,texture gradients, edge interpretation, shading and shadows, and aerial perspective.⁴⁹ Understanding the role of these redundant cues inthe driving task is an active research topic.

Briefly, perception of structure-from-motion or kinetic depth is a real-worlduse of motion perception that may fail in patients with visual cortex lesionsdue to stroke⁵⁰ or early Alzheimer disease.⁵¹ Structure-from-motion deficits in drivers with brainlesions are associated with increased relative risk for safety errors andcar crashes in driving simulation scenarios.⁵²

Recovery of depth from motion relies on relative movements of retinalimages. For motion parallax, relative movement of objects is produced by movingthe head along the interaural axis. Impairments of motion parallax may bea factor in vehicle crashes involving drivers with cerebral impairments, whenthe drivers must make quick judgments with inaccurate or missing perceptualinformation regarding the location of surrounding obstacles, and may contributeto crashes involving alcohol intoxication.⁵³

Displacement of images across the retina during self-motion (ego-motion)produces optic flow patterns that can specify the trajectory of self-motionwith high accuracy.^54-56 Perceptionof heading from optical flow patterns can decline because of aging and drugssuch as marijuana (THC) and ecstasy (MDMA), presumably because of chroniceffects on cholinergic receptors (with THC) and serotoninergic/5-hydroxytryptamine-2receptors (with MDMA).⁵⁷ Processing of visualmotion cues also may be impaired in patients taking antidepressants, suchas nefazodone hydrochloride, that block serotonin reuptake.⁵⁸

Detecting and acting to avoid impending collision events require informationon the driver's vehicle and approaching objects. Objects on collision pathswith the driver maintain a fixed location in the driver's field of view, whereassafe objects will translate to the left or right side. Time to contact isestimated from the expanding retinal image of the approaching object. Olderdrivers are less accurate than younger drivers at detecting an impending collisionduring braking^59,60 and at determiningif an approaching object in a driving scene will crash into them.⁶¹ Performance is worse for longer time-to-contact conditions,possibly because of a greater difficulty in detecting the motion of smallobjects in the driver's field of view.⁶¹ Judgmentson time to contact can be measured in real life using radar detectors.⁶²

Navigating a route relies on visual perception, attention, spatial abilities,and memory. Drivers with topographical disorientation have navigation problemsdespite normal or near-normal visual sensory abilities.^63,64 Patientswith topographical disorientation may have lesions in inferotemporal regions,posterior parahippocampal regions, and the hippocampus, especially in theright hemisphere. Causes of topographical disorientation include stroke, trauma,and neurodegenerative diseases including Alzheimer disease.⁶⁵ Associatedproblems include visual agnosia and other memory-related disturbances. Aspectsof complex disorders of visual processing associated with driver navigationimpairments can be tested with the Trail-Making Test, Benton Visual RetentionTest, and the Complex Figure 1 Test.^45,66,67

The AAO recommends that drivers with vision loss of "intermediate" severitybe allowed to take a road test to assess if they can compensate for the measuredloss. Road tests assess driver performance under the supervision of trainedexperts and are often assumed to be the "gold standard" of driver fitness.The licensing authority may then grant an unrestricted license; restrict drivingto certain conditions, destinations, areas, or equipment; or deny licensure.Yet road tests were developed to test if novice drivers had learned to applythe rules of the road, not to test experienced drivers who may have becomeimpaired. Road testing carries the risk inherent in the real-world road environment.Test conditions vary depending on the weather, daylight, traffic, and drivingcourse. Driving experts may have different biases and scoring criteria, andthere are few data to show that road-test scores predict crash involvement.The use of driving simulation (Figure 1)and instrumented vehicles offer more controlled and less biased assessmentsof driver performance.^45,68

The AAO recommends testing visual sensory, motor, and cognitive abilitiesin at-risk drivers. If possible, training and interventions should be offeredto mitigate deficiencies. Drivers may wear glasses or contact lenses as neededto optimize visual acuity. Drivers with reduced contrast sensitivity may benefitby avoiding dawn, dusk, and nighttime driving and by wearing yellow-filterlenses. Some drivers find that glare is reduced by wearing yellow filtersand polarized lenses. The effect of reduced visual acuity and contrast sensitivityon driving safety may be decreased by restricting travel to familiar low-speedroads and avoiding rush hour, nighttime, and poor weather conditions likerain, snow, and ice.

Bioptic lenses are sometimes prescribed to allow drivers with low visionto see magnified images of traffic signals and signs through telescopes mountedon glasses.⁶⁹ Efficacy should be demonstratedcase by case. Safety benefits and drawbacks of these devices are researchissues. Overlap of the expanded image onto the background produces a ringscotoma; whether this actually increases risk for crashes is a research question.Other issues with optical and electronic techniques that inlay images of areasof interest in traffic scenes are user acceptance, training, adaptation, andcost. Some of the older drivers who might use these devices have difficultydividing attention between 2 sources of information and do not even checkthe gauges and mirrors. Because it takes several hundred milliseconds to directattention from the natural scene to the high-resolution inlay, sudden trafficconflicts might catch the driver unaware.

Reductions in the useful field of view can be partially reversed bytraining effects that can be maintained for at least 1 year⁷⁰;the transfer of this training to driving performance is under investigation.Rehabilitation strategies in homonymous hemianopia include compensation, adaption,and restoration.²⁵ Enlarged side and rearviewmirrors may aid some drivers with peripheral visual field defects. Dangersarising in the vehicle's blind spot can be detected with modern sensors thattrigger devices to warn the driver of the impending danger by haptic, voice,or visual cues.

Of note, while people have poor vision, cars have blind spots. The pillarsbetween the doors produce wedge-shaped areas of blindness trailing off obliquelyfrom the midsection of the vehicle. The posts at the end of the windshieldcreate blind spots going obliquely off the front part of the car. There areindustry efforts to mitigate these blind spots with the use of mirrors andcameras, although added displays may offer more interference and distractionfrom the road. Night driving uses photopic vision, made possible under theillumination provided by streetlights and a car's headlights. Modern cars(eg, some Cadillacs since 2000) have "night vision" systems to allow carsto "see" beyond the headlights in the dark. These displays are not meant tosupplant other driver cues but to augment these cues and to add informationthat can help the driver. Such a system is not a prosthetic or aid for thenight vision–impaired driver, and a night vision–impaired drivershould not expect to benefit from a car equipped with night vision.

Finally, about 78 000 pedestrians are struck and injured by carseach year in the United States; nearly 5000 of those struck are killed, andmore than 60% of these injuries occur at night.⁷¹ Useof reflectors around joints producing motion cues may be a wise preventionstrategy at night for pedestrians. Drivers are much better able to recognizepedestrians at night who wear reflective tape around the joint areas of clothing(wrists, elbows, shoulders, knees, ankles) as opposed to a reflective vestor white clothing.⁷² By the time an older driverdetects a pedestrian in black clothing, it is often too late.

Corresponding author: Matthew Rizzo, MD, Roy J. and Lucille A. CarverCollege of Medicine, Department of Neurology, Division of Neuroergonomics,University of Iowa, 200 Hawkins Dr, Iowa City, IA 52242-1053 (e-mail: matthew-rizzo@uiowa.edu).

Submitted for publication January 7, 2004; final revision received January7, 2004; accepted January 7, 2004.