2252 J. Opt. Soc. Am. A/Vol. 12, No. 10 /October 1995 Bichao et al.

Disability glare: effects of temporal characteristicsof the glare source and of the

visual-field location of the test stimulus

Isabel Cristina Bichao, Dean Yager, and Jeanette Meng

State College of Optometry, State University of New York, 100 East 24th Street, New York, New York 10010

Received February 2, 1995; revised manuscript received March 29, 1995; accepted April 3, 1995

One of the main early complaints of cataract patients, even when these patients exhibit only mild glareproblems as measured by standard tests, is that glare impairs their night driving. To provide a bettermeasure of the patients’ impairment, glare tests should include measurements of the glare effect in condi-tions more similar to those found in night driving. During night driving the ambient light is very low, andoncoming headlights present a transient temporal pattern. Furthermore, the objects of interest often ap-pear initially in the peripheral visual field. Thus three important characteristics of glare in night drivingare that the ambient illuminance is in the scotopic–mesopic range, the detection stimulus is in the periph-ery, and the glare source is transient. Most of the current glare testers measure glare only at photopiclevels, and all the glare tests that we know of use only steady sources of glare with foveal discrimina-tions. All these conditions are dealt with. The transient glare source raised thresholds by 0.5–0.75 logunit more than the steady glare source, and the transient glare effect was more pronounced and morelong lasting in the periphery. Standard glare testers seriously underestimate disability glare effects ineveryday life.

Key words: disability glare, driving, cataract, transient adaptation.

1. INTRODUCTION

Disability glare refers to the impairment in visual func-tion in a given retinal area caused by an intense lightsource whose image is not directed to that area.1 In-traocular forward scatter produces a veiling retinal illu-minance from the glare source, such as a bare light bulb oroncoming automobile headlights, which is at some angu-lar distance from a stimulus of interest, such as a printedpage or a traffic sign. The retinal contrast of the stimu-lus is thus reduced, and a quantifiable disability glare ef-fect may be measured.

A complication is introduced by the fact that in real-life situations potential glare sources are often constantlychanging. Research has shown that the sensitivity-lowering effect of transient adapting sources is muchgreater than that of steady sources. (For a review ofthese effects, see Hood and Finkelstein.2) None of thepresent tests of glare disability takes this factor into ac-count, nor do they control for the light-adaptation effectof the glare source.

The dynamic nature of the response of the visual sys-tem has been ignored by glare-tester designers and users.Typically, a visual function is measured with and with-out a steady glare source, and the difference between thetwo measurements is taken as an estimate of disabilityglare susceptibility. One of the main early complaints ofa substantial proportion of cataract patients, even whenthese patients present only mild glare problems, is thatglare impairs their night driving. Present-day tests arenot sensitive enough to measure some of the patients’deficits in visual function related to glare disability duringnight driving. Sometimes they fail to reveal any abnor-

0740-3232/95/102252-07$06.00

mal value for glare effects in patients who complain aboutbeing blinded by headlights of oncoming vehicles.

To provide a better measure of the patients’ impair-ment, glare tests should include measurements of theglare effect in conditions similar to the ones found innight driving. During night driving the ambient light isvery low, and the main sources of glare—oncoming head-lights and occasionally overhead lights—present a tran-sient temporal pattern. In other words, two importantcharacteristics of glare during night driving are that theambient illuminance is in the scotopic–mesopic range andthat the glare source is transient. However, most of thecurrent glare testers measure glare only at photopic levels[e.g., the Brightness Acuity Tester (Mentor Corporation,Norwell, Mass.)], and all the glare tests that we knowof use only steady sources of glare (for reviews, see, e.g.,Refs. 3 and 4).

The mechanisms underlying disability glare havebeen studied widely.1 Basically, two major hypothet-ical questions were considered: (1) Is the mechanisman inhibitory neuronal interaction across the retina, or(2) is it that the scattered light in the ocular media actsas a veiling luminance, lowering the contrast of the reti-nal image (or do both mechanisms act simultaneously)?Boynton et al.5 examined these questions; they studiedthe effect of direct and indirect adapting fields (glare), ondetection thresholds of a test flash at different adaptationintervals. The direct condition consisted of a central cir-cular field subtending 7 deg, in the center of which thetest spot was flashed. In the indirect (glare) conditionthe adapting field was a 5.5-deg spot at 18 deg in theperiphery, again with the test spot flashed in the fovea.They measured the detection threshold for the test spot

1995 Optical Society of America

Bichao et al. Vol. 12, No. 10 /October 1995 /J. Opt. Soc. Am. A 2253

as the time between onset of the adapting field and on-set of the test flash was varied and then adjusted theintensity of the direct and the indirect fields so that thedetection threshold was the same in both conditions whenthe onsets of the adapting field and the test flash weresimultaneous. The results showed that, when the lumi-nances of the conditions were adjusted in this way, thecurves (log threshold luminance versus adaptation inter-val) obtained were the same for both conditions. Thisresult held true when the luminances of the adaptingfields were changed. These results are in accordancewith the predictions of the scattered-light hypothesis andrender the neuronal interaction improbable because, inthe latter, some form of delay as well as some degree ofvariation should be expected, depending on the intensityof the adapting field. At present it is accepted that theglare effect is due mainly to the scatter of light producedin the intraocular media.

When a steady source of glare is used the scatteredlight acts not only as a veiling luminance but also as anadapting luminance, altering the state of light adaptationof the eye. When the visual system adapts to light thereis a loss in detection sensitivity as the ambient light levelincreases, and there may be a gain in contrast sensitivityas the ambient light increases at low luminances.6 Athigher intensities one obtains performance that demon-strates Weber’s law, according to which contrast sensitiv-ity remains constant with changes in adaptation level.7

The time course of light adaptation has been ex-tensively studied.5,8,9 The increase in the detectionthreshold is highest immediately after the onset of theadapting field. Sensitivity recovers rapidly in the firstsecond after the onset, and it is almost at an asymptoticlevel after the first minute. If the adapting stimulus isvery intense, it may take up to 10 min to reach the steadystate slowly, and the increase in sensitivity may be a logunit or more.

It is well documented that transient backgrounds aremore efficient in raising detection thresholds than aresteady backgrounds of the same intensity.10 – 14 If theadapting field is transient, the eye does not adapt to thehigher luminances and will at all times be in the ini-tial region of the adaptation curve, where the thresh-olds are higher. As mentioned above, it was shown5

that this transient effect is also present with indirectadaptation (glare).

To return to the issue of night-driving problems, thedetection of hazards and signs very often depends on theinitial use of peripheral vision, which then leads to eyeand head movements to foveate the object of interest.To assess more accurately and completely the debilitat-ing effects of a glare source, disability glare effects mustbe measured with a peripheral detection stimulus as wellas a foveal stimulus. Our search of the literature so farhas uncovered no studies of disability glare with a pe-ripheral detection stimulus. Also, the effect of a flasheddirect adapting field in raising threshold compared with asteady adaptation field is stronger in the periphery thanit is in the fovea.12,15 In the present paper we report re-sults obtained by use of a peripheral detection stimuluswith an indirect adaptation field and results from a repli-cation of some aspects of the experiment conducted byBoynton et al.5

2. METHODS

A. Experiment 1: Effects of TransientGlare and Glare IlluminanceA schematic view of the apparatus is shown in Fig. 1,and the subject’s view of the stimuli is shown in Fig. 2.We used a Macintosh computer to present the stimuli onApple monochrome monitors, to control the experiment,and to collect the data. The glare source and the de-tection stimuli were presented on different screens toavoid internal scattering within the monitors. With thetwo monitors the images were superimposed by meansof a mirror, placed at 45 deg., that had a round holecut in the middle of it. The resulting elliptical aperturewas 1.26 deg horizontal and 1.78 deg vertical. The sub-ject viewed the stimuli from 1 m away, monocularly withthe right eye, with the head positioned on a chin rest.The stimuli consisted of a bright glare source occupyingthe whole screen (visual angle 12.2 deg horizontal and9.1 deg vertical), which was reflected off the mirror to-ward the viewing eye. The glare luminance was 140,70, or 17.5 cdym2, measured with a Pritchard photome-ter, and the glare illuminance at the cornea was 4.5, 2.25,or 0.563 lux, measured with a Tektronix J-16 photometer.The detection stimulus was a disk of light with an angularsubtense of 11.4 arcmin, 15 ms in duration, produced ona second monitor and viewed through the aperture in themirror. The stimulus could appear centered 11.25 min tothe left or to the right of center of the aperture. Four dim

Fig. 1. Schematic overhead view of the experimental apparatus(not drawn to scale).

Fig. 2. Subject’s view of the stimuli used in experiment 1 (notdrawn to scale).

2254 J. Opt. Soc. Am. A/Vol. 12, No. 10 /October 1995 Bichao et al.

Fig. 3. Subject’s view of the stimuli used in experiment 2. Thefixation point for eccentric viewing is shown to the left of thestimulus aperture (not drawn to scale).

Fig. 4. Two examples of the time difference between the onsetof the glare source and the presentation of the stimulus: SOA’s.

fixation points were positioned at the ends of the majorand the minor axes of the ellipse, and the subject fixatedthe center of the elliptical opening.

B. Experiment 2: Effect of Visual-FieldLocation of the Test StimulusWe used the same apparatus that was used inexperiment 1, with only slight modifications. Forthe foveal measurements the subject fixated the centerof the elliptical opening. For the peripheral measure-ments the subject fixated on a dim grain-of-wheat lamppositioned 2.85 deg to the left of the stimuli. Fixationstability was not monitored, but all the subjects werewell-practiced psychophysical observers (see Section 3).Glare illuminance was constant at 4.5 lux at the cornea.

Again, the detection stimuli were disks of light withan angular subtense of 11.4 arcmin. For the foveal mea-surements stimuli were 15 ms in duration; because oflower sensitivity in some conditions in the periphery andbecause of luminance limitations of the stimulus moni-tor, stimuli were 45 ms in duration of eccentric viewing.They could appear 37.5 min above or below the centerof the elliptical aperture in the mirror. This separationof the test stimuli, wider than that used in experiment 1,was necessary because of the reduced sensitivity to spatialposition with eccentric viewing. The vertical rearrange-ment of the detection stimuli was for the purpose ofmaintaining a constant eccentricity with the horizontallyplaced fixation point (see Fig. 3).

Three main types of conditions were used in both ex-periments: (1) Dark adapted, with the glare monitor

turned off. (2) Light adapted, with the glare sourceturned on continuously. (These two conditions are anexample of the standard paradigm for glare testing.)(3) Transient glare, for which the source was turned onthe 1 s and the stimuli were presented at different timesafter onset of the glare.

In the dark-adapted conditions thresholds were mea-sured after 10 min of dark adaptation. In the steady con-ditions the subjects adapted to the glare source for 10 minbefore they started the trials, and the glare source waskept turned on throughout the experiment.

In the transient conditions the subjects dark adaptedfor 10 min before starting the trials, the glare sourcewas kept turned on for 1000 ms on each trial, and thetest flash appeared, in different blocks of trials, at dif-ferent times during the glare presentation. These timeswere expressed as stimulus onset asynchronies (SOA’s;see Fig. 4). There was a 10-s period between trials.

On each trial the test flash would appear randomly inone of two possible positions, and the subject pressed akey to indicate the position in which the stimulus had ap-peared. We used a spatial two-alternative forced-choicemethod combined with the method of constant stimuli.

Fig. 5. Threshold as a function of SOA for two subjects. Theglare illuminance was 4.5 lux. Foveal viewing.

Bichao et al. Vol. 12, No. 10 /October 1995 /J. Opt. Soc. Am. A 2255

Fig. 6. Threshold as a function of SOA for four subjects. The glare illuminance was 4.5 lux (squares), 2.25 lux (circles), or 0.563 lux(triangles). Foveal viewing.

Psychometric functions were fitted with a three-parameter Weibull function to estimate thresholds.16,17

We used Quick’s18 approximation to the cumulative nor-mal distribution function to describe the probability ofdetecting a stimulus:

P sxd ­ 1 2 22jx/ajb , (1)

where x is the stimulus dimension (luminance), a is thestimulus value for P sxd ­ 0.5, and b is the slope of thefunction determined by stimulus-related (intrinsic) noise.

The probability of a correct response in the two-alternative forced-choice method, Rsxd, is given by

Rsxd ­ gP sxd 1 0.5f1 2 P sxdg , (2)

where g is the upper asymptote of the psychometricfunction, determined by performance-related (extrinsic)noise. Parameter estimates were based on a maximum-likelihood method (for a discussion of this method seeHarvey19). For a given stimulus x, the likelihood Lsxd

that correct responses are given on k out of n trials isgiven by

Lsxd ­ fn!yk!sn 2 kd!gfRsxdgkf1 2 Rsxdgn2k, (3)

and the likelihood of a complete data set for a giventest is the product of the likelihoods for all the stimuliused in that set.16 Likelihoods were computed for thethree-dimensional parameter space, and the a value forthe maximum-likelihood set was taken as the thresholdluminance.

3. SUBJECTSAll the subjects had visual acuities correctable to 20/20.Their ages ranged from 16 to 54. Three of the subjectswere the authors of this paper, and the others also werewell-practiced psychophysical observers.

2256 J. Opt. Soc. Am. A/Vol. 12, No. 10 /October 1995 Bichao et al.

Fig. 7. Threshold as a function of glare illuminance at twodifferent SOA’s. Foveal viewing. Mean values are given forfour subjects. The error bars are standard errors of the means.

Fig. 8. Threshold at SOA15, SOA500, light adapted (LA), anddark adapted (DA). Foveal viewing and eccentricity (ECC) of2.8 deg. The glare illuminance was 4.5 lux. Mean values aregiven for five subjects. The error bars are standard errors ofthe means.

4. RESULTS

A. Experiment 1Complete threshold versus SOA functions at the highestglare illuminance are shown in Fig. 5 for two subjects.Thresholds begin to rise slightly before the glare sourceis turned on, reach a rapid peak at glare onset, fall toapproximately the steady light-adapted level in 500 ms,and then return quickly to nearly the dark-adapted levelwhen the glare source is turned off. These curves arequalitatively similar to Crawford’s9 results with directadaptation and to the indirect adaptation (glare) resultsreported by Boynton et al.5

In Fig. 6 the results obtained from four subjects by useof three different glare levels are presented. As the glareintensity increases, the magnitude of the transient effectincreases. In Fig. 7 we have plotted the threshold (meandata of the four subjects) as a function of glare illuminanceat two different extreme temporal conditions: SOA of15 ms (SOA15; short exposure, transient), and SOA of500 ms (SOA500; long exposure, close to the steady state).The threshold increases as a function of illuminance inboth conditions, but there is a steeper rise for the tran-sient glare: as glare intensity increases, the differencebetween transient and steady conditions increases.

B. Experiment 2A subset of the conditions used in experiment 1 wastested: dark adapted, steady light adapted, SOA15between the onset of the glare source and the presen-tation of the test flash, and SOA500. Figure 8 showsthe detection thresholds for these conditions in the foveaand 2.8 deg in the periphery. The pattern of results forboth the fovea and near the periphery is similar to thoseshown in Figs. 5 and 6. Thresholds are most elevatedjust after onset of the glare source and then begin tofall toward the steady light-adapted level. (Note thatthe absolute thresholds in the two visual-field locationscannot be compared because different stimulus durationswere used.)

Fig. 9. Disability glare: log threshold sglared 2 log threshold(no glare). Two glare conditions: steady, and SOA15. Twovisual-field locations: fovea, and 2.8-deg eccentricity. Meanvalues are given for five subjects. The error bars are standarderrors of the means.

Bichao et al. Vol. 12, No. 10 /October 1995 /J. Opt. Soc. Am. A 2257

Fig. 10. Increase in threshold for transient glare over thresholdfor steady glare, at SOA’s where x ­ 15 ms and x ­ 500 ms.Two visual-field locations: fovea and 2.8-deg eccentricity.Mean values are given for five subjects. The error bars arestandard errors of the means.

Fig. 11. Expanded view of the data shown in Fig. 10, atSOA500. Two visual-field locations: fovea, and 2.8-deg eccen-tricity. Mean values are given for five subjects. The errorbars are standard errors of the means.

Two disability glare scores were calculated by substrac-tion of the log threshold with no glare (dark adapted)from the log threshold with steady glare and from the logthreshold with transient glare (SOA15). This was doneseparately for the fovea and for the periphery. The meanresults across subjects are shown in Fig. 9. As shownin experiment 1, the transient glare effect is significantlygreater than the steady glare effect. In addition, theseresults show that the disability glare effect for both con-ditions is enhanced in the periphery.

In Fig. 10 we show that, in the fovea, thresholds returnto nearly the steady glare level after 500 ms but in theperiphery thresholds are still elevated at this time; thiseffect is shown in isolation in Fig. 11.

5. DISCUSSIONThe main result obtained from experiment 1 is that a dis-ability glare effect in the fovea produced by the sudden

onset of a glare source was significantly greater than theeffect from a steady glare source of the same illuminance.Clearly, transient glare effects, such as those often en-countered in everyday life, are much stronger than thosetested with standard glare parameters. Also, in peoplewith increased intraocular scatter, this transient effectmay have a disproportionately greater disabling effect andmay be more likely to present a greater difference betweensteady and transient glare effects. This means that suchpeople would manifest a bigger discrepancy between con-ventional glare testing (steady glare source) and real-lifenight-driving disability (transient glare source).

In experiment 2 we also show that within the limitedvalues of the parameters investigated so far, disabilityglare effects are stronger when the observer must de-tect a stimulus that is not in his or her central visualfield. Because many stimuli in everyday life do not fallinitially onto the fovea (such as roadside hazards in driv-ing), it is apparent that glare testers that do not addressthis problem will fail to predict real-life glare disabilityeffects. Furthermore, this visual-field location enhance-ment of glare effects is exaggerated in both magnitudeand duration when the glare source is not steady, whichagain is closer to real-life circumstances.

The present results were obtained with a stimulus anda glare source that do not directly simulate a real-life situ-ation such as driving at night. Preliminary dataobtained with a point source of glare, more like an auto-mobile headlight, have confirmed the conclusions of themore artificial arrangement used in this experiment.20,21

Also, different discrimination tasks and different ar-rangements of glare and stimulus could lead to differentresults22 and must be investigated to increase the gen-erality of the conclusions. Furthermore, these transienteffects with glare must be examined in subjects withvisual impairments, such as cataract patients.

ACKNOWLEDGMENTSI. Bichao was supported by an Individual Post-DoctoralFellowship from the National Eye Institute. HughWilson supplied the original Pascal programs that wereadapted for conducting these experiments, and WilliamSwanson provided us with the program to analyzethe psychometric functions. Two anonymous reviewersprovided helpful comments for revision of the originalmanuscript.

REFERENCES1. J. J. Vos, “Disability glare—a state of the art report,” CIE

J. 3, 39–53 (1984).2. D. Hood and M. Finkelstein, “Sensitivity to light,” in Hand-

book of Perception and Human Performance, K. Boff, L.Kaufman, and J. Thomas, eds. (Wiley-Interscience, NewYork, 1986), Vol. 1, Chap. 5.

3. American Academy of Ophthalmology, “Contrast sensitivityand glare testing in the evaluation of anterior segment dis-ease,” Ophthalmology 97, 1233–1237 (1990).

4. A. Neumann, G. McCarty, J. Locke, and B. Cobb, “Glaredisability devices for cataractous eyes: a consumer’s guide,”J. Cataract Refract. Surg. 14, 212–216 (1988).

5. R. Boynton, W. Bush, and J. Enoch, “Rapid changes infoveal sensitivity resulting from direct and indirect adaptingstimuli,” J. Opt. Soc. Am. 44, 56–60 (1954).

2258 J. Opt. Soc. Am. A/Vol. 12, No. 10 /October 1995 Bichao et al.

6. H. Blackwell, “Contrast thresholds of the human eye,”J. Opt. Soc. Am. 36, 624–643 (1946).

7. R. Yuan, D. Yager, M. Guethlein, G. Oliver, N. Kapoor,and R. Zhong, “Controlling unwanted sources of thresholdchange in disability glare studies: a prototype apparatusand procedure,” Optom. Vision Sci. 70, 976–981 (1993).

8. H. Baker, “The course of foveal light adaptation measuredby the threshold intensity increment,” J. Opt. Soc. Am. 39,172–179 (1949).

9. B. Crawford, “Visual adaptation in relation to brief condi-tioning stimuli,” Proc. R. Soc. London Ser. B 134, 283–302(1947).

10. E. Rinalducci and A. Beare, “Visibility losses caused by tran-sient adaptation at low luminance levels,” in TransportationResearch Board Special Report 156 (National Academy ofSciences, Washington, D.C., 1975), pp. 11–22.

11. W. Geisler, “Adaptation, afterimages and cone saturation,”Vision Res. 18, 279–289 (1978).

12. M. Finkelstein and D. Hood, “Cone system saturation:more than one stage of sensitivity loss,” Vision Res. 21,319–328 (1981).

13. E. H. Adelson, “Saturation and adaptation in the rod sys-tem,” Vision Res. 32, 1299–1312 (1982).

14. M. Hayhoe, N. Benimoff, and D. Hood, “The time course ofmultiplicative and subtractive adaptation processes,” VisionRes. 27, 1981–1996 (1987).

15. V. Greenstein and D. Hood, “Foveal–parafoveal differences

in suprathreshold response,” Invest. Ophthalmol. Vis. Sci.Suppl. 29, 29 (1979).

16. W. Swanson and E. Birch, “Extracting thresholds from noisypsychophysical data,” Percept. Psychophys. 51, 409–422(1992).

17. D. Yager and B. Beard, “Age differences in spatial contrastsensitivity are not the result of changes in subjects’ crite-ria or psychophysical performance,” Optom. Vision Sci. 71,778–782 (1994).

18. R. F. Quick, “A vector-magnitude model of contrast detec-tion,” Kybernetik 16, 65–67 (1974).

19. L. Harvey, “Efficient estimation of sensory thresholds,” Be-hav. Res. Methods Instrum. Comput. 18, 623–632 (1986).

20. J. Meng, C. Bichao, D. Yager, H. Zhan, and J. Cheung,“Disability glare with retinal position of the detection stim-ulus and the temporal relations between the glare sourceand detection stimulus as variables,” in Vision Science andIts Applications, Vol. 1 of 1995 OSA Technical Digest Se-ries (Optical Society of America, Washington, D.C., 1995),pp. 171–174.

21. C. Bichao, J. Meng, and D. Yager, “Temporal and spatialaspects of disability glare: simulation of nighttime driving,”Invest. Ophthalmol. Visual Sci. Suppl. 36, 940 (1995).

22. T. Frumkes, “Suppressive rod–cone interaction,” in Prin-ciples and Practice of Clinical Electrophysiology of Vision,J. Heckenlively and G. Arden, eds. (Mosby, St. Louis, Mo.,1991), pp. 469–474.