Luminance, not brightness, determines temporal brightnessenhancement with chromatic stimuli

Richard W. BowenDepartment of Psychology, Loyola University of Chicago, Chicago, Illinois 60626

Mary Jo NissenEye Research Laboratories, The University of Chicago, Chicago, Illinois 60637

(Received 27 July 1978; revised manuscript received 6 December 1978)

Brightness-duration relations for chromatic stimuli were studied using three pulse-to-backgroundluminance relations: chromatic equal-luminance pulses (3.2 cd/M2) were presented as increments of0.3 or 1.0 log units above a lower luminance achromatic background, or were presented in huesubstitution, equated in luminance to the achromatic background, so that no spatio-temporal lumi-nance transients occurred during stimulus presentation. Incremental pulses produced temporalbrightness enhancement (the Broca-Sulzer phenomenon), but hue substitution pulses did not. Tem-poral brightness enhancement thus depends upon the occurrence of luminance transients and cannotbe produced by pulsed-to-background brightness differences associated solely with chromaticity dif-ferences.

The brightness of a chromatic stimulus is a joint functionof its luminance and its chromaticity. Spectral lights equatedfor luminance' differ substantially in brightness: brightnessis at a minimum near 570 nm (the wavelength of least satu-ration) and increases toward the spectral extremes by theequivalent of 0.5-0.6 log unit of luminance contrast. 2' 3 Thiseffect is thought to reflect a contribution to brightness byactivity in chromatic neural pathways for stimuli that activatethe achromatic or luminance pathways to an equal de-gree.2, 4 -6

The present study reports the effect of another variable,exposure duration, on the brightness of equal-luminancechromatic lights. With achromatic stimuli, brightness in-creases with duration for brief stimuli, is independent of du-ration for long stimuli, and is at a maximum value for stimuliof intermediate duration (50-150 ms).7' 8 The existence ofpeak brightness at intermediate exposure durations has beenreferred to as "temporal brightness enhancement" or the"Broca-Sulzer phenomenon." There is little empiricalagreement on whether chromaticity is a critical variable indetermining the form of the brightness-duration relation.9 -'3

Among previous studies there are a variety of methodologicaland stimulus differences (including luminance level, retinallocus of stimulation, and state of adaptation) that might ac-count for conflicting results. Further, chromatic test lightshave in some instances been equated for brightness ratherthan luminance.'12"3 Heterochromatic brightness matchingproduces highly variable results within and between observ-ers,3 and also causes luminance to vary among chromatic teststimuli. Thus, it is difficult to infer the separate contributionof stimulus luminance and chromaticity in determining du-ration effects on brightness.

In this paper we examine the relation between brightnessand duration for spectral lights that are equated for lumi-nance. Further, we introduce a new variable: chromaticstimuli were either presented as increments in luminanceabove an achromatic background field, or were matched inluminance to the background field to eliminate spatio-tem-poral luminance transients during stimulus presentation, acondition called hue substitution.' 4 A stimulus presented

in hue substitution represents chromaticity modulationwithout luminance modulation.

Recent studies of temporal and spatial processing of chro-matic stimuli'-' 9 suggest that hue substitution may selec-tively tap neural mechanisms sensitive to chromaticitymodulation (the chromatic opponent-color channels) withoutactivating mechanisms sensitive to changes in luminance (anachromatic nonopponent channel). The present researchwith the hue substitution method is intended to determinethe brightness-duration relation for chromatic channels inisolation, and to establish whether the relation betweenbrightness and duration for chromatic stimuli depends uponluminance transients.

METHOD

ObserversThe authors served as observers. Both have normal vision

with corrective lenses and normal color vision.

General psychophysical methodBrightness-duration relations have usually been studied

either with brightness-matching techniques8' 20 or with directscaling methods. 21' 22 We employed an alternative methoddeveloped by Bowen and Pokorny. 2 3 On each experimentaltrial, the observer viewed a sequence of two pulses of coloredlight of equivalent luminance and wavelength but differentduration, and judged which pulse (first or second) appearedbrighter. In the present study, the pulses differed in durationby 50 ms. From trial to trial, the order of pulse presentation(long versus short) and the duration of the shorter pulse werevaried. As data, we consider the percentage of trials on whichthe longer pulse was judged as brighter for a range of durationsfor the shorter pulse. For brief pulses, where brightness isincreasing with duration, the longer of two pulses should bejudged brighter on 100% of all trials. For pulses of long du-ration, where brightness does not vary with duration, thelonger of the two pulses should be judged brighter on 50% ofall trials (chance behavior). For intermediate pulse durations,

581 J. Opt. Soc. Am., Vol. 69, No. 4, April 1979 0030-3941/79/040581-04$00.50 � 1979 Optical Society of America 581

there are two possible outcomes. If temporal brightness en-hancement occurs, there should be a comparison for which theshorter pulse is consistently judged as brighter; i.e., the per-cent of trials on which the longer pulse is judged brightershould be below 50%1/. The value of the shorter pulse at whichthe data approach zero percent is an estimate of the pulseduration resulting in maximum brightness. If under certainstimulus conditions the Broca-Sulzer effect does not occur,the data should show a monotonic transition from 100% to 50%with no values below 50%.

We have noted elsewhere2 3 that this method does notevaluate absolute brightness level. Therefore we cannotevaluate the magnitude of obtained brightness change withduration or the magnitude of obtained brightness enhance-ment effects. But the method is sensitive to both the occur-rence of brightness enhancement and the variation ofbrightness with duration.

ApparatusThe apparatus and calibration procedures have been de-

scribed by Nissen and Pokorny.'8 We used a three-channeloptical projection system under the control of a PDP-15computer.

The stimulus array (Fig. 1) consisted of a 3.80 diam circularachromatic background field of uniform luminance. Chro-matic test pulses were presented at a 1.90 diam circular loca-tion centered within the background by replacing the whitelight present at that location with chromatic light of eitherequal or higher luminance. Observers fixated the center ofthe display monocularly. The surround was dark. Theachromatic background had chromaticity coordinates of x =0.3607, y = 0.3616, with a correlated color temperature of 4505K. Narrow-band interference filters were used to generatechromatic test stimuli of 504, 570, and 620 nm. A Chancebroad-band filter created a blue-light stimulus (dominantwavelength was 463 nm, excitation purity was 0.95).

Pulsed chromatic stimuli were presented under one of threepulse-to-background luminance conditions (Fig. 1): as in-

11.90

Z

cr3

7. E

1.0 LOG UNIT 0.3 LOG UNITINCREMENT INCREMENT

HUESUBSTITUTION

FIG. 1. The Stimulus array. Bottom: Relative luminance cross sectionsof the stimulus array during pulse presentation for the three luminanceconditions studied in this experiment. Stippling represents the amount ofcolored light present during pulse presentation. See text for discus-sion.

crements of 1.0 log unit (pulse luminance of 3.2 cd/M 2,background luminance of 0.32 cd/M 2), as increments of 0.3 logunit (pulse luminance of 3.2 cd/M2, background luminanceof 1.6 cd/M 2) or in hue substitution (pulse and backgroundluminance of 3.2 cd/M2). Pulse and background luminancewere equated for the hue substitution condition using het-erochromatic flicker photometry (See "Procedure" below).

Pulse durations for the shorter pulse ranged from 20 to 200ms at 20-ms intervals. The longer pulse had a duration equalto the shorter pulse plus 50 ms.

ProcedureA daily experimental run consisted of a block of 200 trials:

10 trials at each of 10 pulse durations for the shorter pulse(20-200 ms) for each of two wavelengths under a givenpulse-to-background luminance condition. On each experi-mental trial, a short pulse and long pulse of the same wave-length were presented. The two wavelengths to be testedwere each matched in luminance to the achromatic field (ata luminance of 3.2 cd/M2) using heterochromatic flickerphotometry at alternation rates of 8-11 Hz. For incrementconditions, 0.3 or 1.0 log unit of density was added to theachromatic field. On a given trial, the computer randomlyselected the wavelength to be tested (by rotation of a dualfilter box), the duration of the shorter pulse, and the order ofpulses (long versus short). The observer initiated a trial bydepressing a telegraph key. 1 s later the sequence of twopulses of light of the same wavelength was presented, with aninterval between pulses of 1.5 s. The observer then respondedby depressing one of two telegraph keys to indicate whichpulse, the first or the second, appeared brighter. The inter-trial interval was approximately 3 s. Selection and presen-tation of stimuli and tabulation of responses were completelyunder computer control.

For each pulse-to-background luminance condition withinan experimental session, two wavelengths were tested and agiven wavelength was tested with each of the other three testwavelengths in separate experimental sessions. Test wave-length pairs for each pulse-to-background luminance condi-tion were run in random order for two replications. Reporteddata were based on a total of 60 trials per duration/wave-length/pulse-to-background luminance condition collectedduring six daily experimental sessions.

RESULTS AND DISCUSSION

Figure 2 shows the percent of trials on which the longerpulse in a comparison was judged brighter as a function of theduration of the shorter pulse for the three luminance condi-tions at four test wavelengths. For the 1.0 log unit incrementcondition (top panels), the data are near 100% at short pulsedurations, fall below 50% for intermediate durations, and areasymptotic at 50% for long durations. This pattern of resultsindicates the occurrence of temporal brightness enhance-ment.

For increments of 0.3 log unit (middle panels) temporalbrightness enhancement is also evident. For both observers,pulse durations yielding peak brightness are shifted towardlonger values, an effect of pulse-to-background luminancecontrast.

582 J. Opt. Soc. Am., Vol. 69, No. 4, April 1979 Richard W. Bowen and Mary Jo Nissen 582

c

120 160 200 40 80DURATION OF SHORTER PULSE (macc)

FIG. 2. Functions relating the percent of trails on which the longer of twopulses was judged as brighter to the duration of the shorter pulse for thethree luminance conditions at each of four wavelengths for both observers.The longer pulse was equal to the duration of the shorter pulse plus 50Ms.

Under conditions of hue substitution, temporal brightnessenhancement does not occur. The curves for different testwavelengths are asymptotic at the 50% level.

There are three major observations concerning the resultsobtained in the increment and substitution conditions. First,maximum brightness occurs at shorter durations for incre-mental stimuli than for hue substitution stimuli. For subjectRWB, the average duration yielding maximum brightness was175 ms in the substitution condition, 165 ms in the 0.3 log unitincrement condition, and 100 ms in the 1.0 log unit incrementcondition. Measurements of the duration yielding maximumbrightness for RWB were subjected to a two-way analysis ofvariance (ANOVA), with factors including pulse-to-back-ground luminance condition (three levels) and wavelength(four levels), with experimental sessions (six levels) as theerror variable. The analysis indicated a significant effect ofpulse-to-background luminance condition [F(2,10) = 14.43,p <0.005]. For subject MJN, the average duration yieldingmaximum brightness was 160 ms in the substitution condition,135 ms in the 0.3 increment condition, and 110 ms in the 1.0increment condition. Although this is a sizeable effect in thepredicted direction, the ANOVA performed on these datashowed that the effect of condition did not reach statisticalsignificance [F(2,10) = 1.90, p <0.20], due to the large vari-ability in her data.

Assuming that stimuli in hue substitution selectively acti-vate only the color-opponent channels whereas incrementalstimuli activate both chromatic and achromatic channels,these results suggest that brightness grows over a longer rangeof pulse durations when responses are mediated by the chro-

HL ~ , - ., ,

I'- . , \ . .11;

583 J. Opt. Soc. Am., Vol. 69, No. 4, April 1979

matic channels alone. The growth of brightness contributedby chromatic mechanisms when luminance transients arepresent is unknown since both chromatic and achromaticmechanisms will respond to luminance transients.2 4' 25 Itmust be remembered that hue substitution achieves isolationof chromatic mechanisms through the elimination of lumi-nance transients at stimulus presentation.

Second, wavelength has no systematic effect on the durationat which maximum brightness occurs in increment or sub-stitution conditions. The analyses of variance showed nosignificant wavelength effect for either RWB [F(3,15) = 0.47]or MJN [F(3,15) = 1.37]. Finally, the interaction betweenwavelength and condition was not significant for either subject[F(6,30) = 0.37 for RWB and F(6,30) = 0.57 for MJN].

Based on experimental comparisonsl- 3 of spectral sensi-tivity for heterochromatic flicker photometry (a methodgenerating equal-luminance spectra) and heterochromaticbrightness matching (defining equal-brightness spectra), weknow that our equal-luminance test stimuli differ substan-tially in brightness. But for all conditions, inherent bright-ness differences have no systematic effect on the pulse dura-tion yielding maximum brightness.

Third, the Broca-Sulzer effect is absent under hue substi-tution conditions. Although for a given chromatic stimulusthe difference in brightness between pulse and backgroundis less for the hue substitution condition than for incrementalstimuli, this fact cannot account for the lack of temporalbrightness enhancement in hue substitution. A 463-nmstimulus in hue substitution differs in brightness from theachromatic background by a greater factor (the equivalent of0.5 log unit of density or more) than does 570 nm presentedas a 0.3 log unit increment", 2 -but the latter condition pro-duces brightness enhancement while the former does not.Brightness difference alone is therefore not a sufficient con-dition to produce temporal brightness enhancement.Brightness contributed by chromaticity differences and byluminance differences are dissociated in this paradigm, withluminance being the critical parameter for obtaining temporalbrightness enhancement. It may be that luminance tran-sients modify the response of the chromatic channels or pro-duce interactions between chromatic and achromatic channelresponses' 6 that are capable of generating temporal brightnessenhancement. Alternatively, it is also possible that the effectis mediated entirely by the achromatic channel.

Finally, we note that the absence of brightness enhance-ment for selective activation of chromatic channels by huesubstitution may be related to the lack of low-frequency in-hibition for sensitivity to pure hue modulation, while the oc-currence of the effect with luminance transients may reflectthe band-pass temporal characteristics obtained with lumi-nance modulation. 2 6

ACKNOWLEDGMENTS

We thank Joel Pokorny and Vivianne C. Smith for valuablecriticism during the course of the research and for a criticalreading of the manuscript. This work was supported in partby National Science Foundation Grant BNS 7817779 (R. W.Bowen), and by NIH, USPH, NEI Grant EY00901 (J. Pok-orny). Dr. Nissen was supported as a postdoctoral fellow onNIH, USPH, NEI Grant EY07010. She is now at the De-

Richard W. Bowen and Mary Jo Nissen 583

partment of Psychology, Florida State University, Tallahas-see, Florida 32306. Dr. Bowen is a Visiting Scholar at TheUniversity of Chicago. The major substance of this paper waspresented at the annual meeting of A.R.V.O., April, 1978.

'Luminance is defined by empirical measurements of either hetero-chromatic flicker photometry [G. Wagner, and R. M. Boynton,"Comparison of four methods of heterochromatic photometry,"J. Opt. Soc. Am. 62,1508-1515 (1972)] or by the minimally distinctborder criterion [R. M. Boynton, and P. K. Kaiser, "Vision: Theadditivity law made to work for heterochromatic photometry withbipartite fields," Science 161, 366-368 (1968)].

2 P. K. Kaiser and J. P. Comerford, "Flicker photometry of equallybright lights," Vision Res. 15, 1399-1402 (1975).

3J. P. Comerford and P. K. Kaiser, "Luminous efficiency functiondetermined by heterochromatic brightness matching," J. Opt. Soc.Am. 65, 466-468 (1975).

4R. M. Boynton, "Implications of the minimally distinct border," J.Opt. Soc. Am. 63, 1037-1043 (1973).

5S. L. Guth, N. J. Donley, and R. T. Marroco, "On luminance addi-tivity and related topics," Vision Res. 9, 537-576 (1969).

6S. L. Guth and H. Lodge, "Heterochromatic additivity, fovealspectral sensitivity, and a new color model," J. Opt. Soc. Am. 63,450-462 (1973).

7 D. Broca and A. Sulzer "La sensation lumineuse en fonction dutemps," J. Physiol. Pathol. Gen. 4, 632-640 (1902).

8 M. Katz, "Brief flash brightness," Vision Res. 4, 361-373 (1964).9 D. Broca and A. Sulzer, "Sensation lumineuse en fonction du temps

pour les lumieres colorees; Technique et resultats," C. R. Hebd.Seanc. Acad. Sci. (Paris) 137, 944-946 (1903).

10M. A. Bills, "The lag of visual sensation in its relation to wavelengthsand intensity of light," Psychol. Mongr. 28, No. 127 (1920).

"W. H. Stainton, "The phenomenon of Broca and Sulzer in fovealvision," J. Opt. Soc. Am. 16, 26-39, 1928.

12 R. J. Ball, "An investigation of chromatic brightness enhancementtendencies," Am. J. Optom. 41, 361-371 (1964).

3G. S. Wasserman, "Brightness enhancement and opponent-colorstheory," Vision Res. 6, 689-699 (1966).

14 F. Weingarten, "Wavelength effect on visual latency," Science 176,692-694 (1972).

15M. E. Breton, "Hue substitution: Wavelength latency effects," Vi-sion Res. 17, 435-443 (1977).

16 R. W. Bowen, J. Pokorny, and D. Cacciato, "Metacontrast maskingdepends on luminance transients," Vision Res. 17, 971-975(1977).

17R. W. Bowen, D. T. Lindsey and V. C. Smith, "Chromatic two-pulseresolution with and without luminance transients," J. Opt. Soc. Am.67, 1501-1507 (1977).

18M. J. Nissen and J. Pokorny, "Wavelength effects on simple reactiontime," Percept. Psychophys. 22, 457-462 (1977).

19J. Pokorny, R. W. Bowen, and D. T. Lindsey, "Duration thresholdsfor chromatic stimuli," J. Opt. Soc. Am. 67, 1308A (1977).

2 0T. S. Aiba and S. S. Stevens, "Relation of brightness to duration andluminance under light- and dark-adaptation," Vision Res. 4,391-401 (1964).

2 1D. Raab, "Magnitude estimation of the brightness of brief fovealstimuli," Science 135, 42-44 (1962).

22 M. F. Lewis, "Category judgments as functions of flash luminanceand duration," J. Opt. Soc. Am. 55, 1655-1660 (1965).

23 R. W. Bowen and J. Pokorny, "Target edge sharpness and temporalbrightness enhancement," Vision Res. 18, 1691-1695 (1978). Avariant of the brightness comparison method we describe may ac-tually be traced to N. MacDougal ["The variation of the intensityof visual sensation with the duration of the stimulus," Brit. J.Psychol. 1, 151-189 (1904)].

2 4R. L. DeValois, "Processing of intensity and wavelength informationby the visual system," Invest. Ophthalmol. 11, 417-426 (1972).

2 5C. Ingling and B. Drum "Retinal receptive fields: Correlations be-tween psychophysics and electrophysiology," Vision Res. 13,1151-1163 (1973).

26 D. H. Kelly and D. van Norren, "Two-band model of heterochro-matic flicker," J. Opt. Soc. Am. 67, 1081-1091 (1977).

White-light Fraunhofer diffractionRonald Bergsten

Department of Physics,Uniuersity of Wisconsin-Whitewater, Whitewater, Wisconsin 53190

Susan Huberty PopelkaAbbott Laboratories, North Chicago, Illinois 60064

(Received 17 July 1978)

Colorimetry and diffraction theory are combined to determine the color characteristics of white-light Fraunhofer diffraction patterns. The weighted-ordinate method of colorimetry is applied todetermine chromaticity information about single-slit, circular-aperture, and double-slit diffractionpatterns.

INTRODUCTION

Techniques for photographing diffraction patterns, a setof unique diffraction patterns illustrated in black and whiteand an excellent set of references are presented by Harris.'Photographs of diffraction patterns have recently been pub-lished in color.2'3

The luminance and chromaticity distribution of a white-light diffraction pattern are due to the intensity contributionsof each wavelength of which white light is composed.

White-light diffraction patterns must be analyzed by use ofcolorimetric techniques to determine the color characteristicsin terms of dominant wavelength (the wavelength of spectrallyhomogeneous light that, when mixed with white light, wouldproduce a color match), purity (measure of color saturation),and relative luminance. The science of colorimetry takes intoaccount the spectral distribution of the energy of the sourceof light, the spectral response characteristics of a standardobserver's eyes, and the intensity-versus-wavelength rela-tionship of the diffraction pattern. The tristimulus values

584 J. Opt. Soc. Am., Vol. 69, No. 4, April 1979 � 1979 Optical Society of America 5840030-3941/79/040584-06$00.50