- Email: [email protected]
Border-contrast: Corner brightening effects
Virmn Rex.
Vol. 9. pp. 1309-1313.
Pergamoa
Press 1969.
LETTER
BORDER-CONTRAST: (Received
Printed in Great
TO THE EDITORS
CORNER
8 January
Britain.
BRIGHTENING
EFFECTS1
1969; in revised form 26 March 1969)
WHEN a sharp black and white border was viewed several degrees from the line of direct vision, subjective brightening of the white background near the border was found (HARMS and AULHORN,1955; PAYNE and WHITE, 1967). This effect was not present with fovea1 viewing. Conversely, a uniform gradual change in luminance between dark and light fields gave rise to a subjectively light band at the edge of the high luminance field and a dark band at the edge of the low luminance field. This “Mach band” effect is only seen with fovea1 viewing (RATLIFF, 1965 ; FIORENTINI,1957). It has been suggested by TSHERMAK(1930) that edge effects should be differentiated from area contrast effects. The Hermann grid was investigated by PAYNEand ANDERSON (1969) to see if the grid effect, which was only seen with indirect vision, could possibly be attributed to edge rather than area contrast effects. Some of the results of this study indicated subjective brightening (lateral excitation) of the white background in the vicinity of the black right angle comers. The purpose of this study was to investigate the effect of a black comer angle on lateral excitation of the surrounding white field. Several quantitative methods of measuring border effects have been used. Viewing a point near a border with one eye and simultaneously making a brightness match with a test spot projected into the other eye was used (FIORENTINI,1957). A small slit of variable luminance was superimposed on the test location at the border and a brightness match was made (WATRGIENICZ, 1963). A small test spot was moved over the border display and difference thresholds above the background were measured (HARMSand AULHORN, 1955). Average evoked potential measurements taken from an observer’s occipital cortex to a test flash on the border display were also used (PAYNEand WHITE, 1967). All of these methods reflect subjective brightness at specified points on the border although some of the measurements were indirect (difference thresholds and evoked potentials). A small suprathreshold spot on a bright background appeared dim as compared to the same intensity spot viewed on a dark background. Simple reaction time (RT) was used as a measure of flash brightness (PAYNE, 1967). This flash brightness measure was accurate only within certain ranges of test flash brightness and background intensities. If the flash were too dim (bright) relative to the background, only long (short) RTs were obtained and did not accurately reflect flash brightness. If a constant intensity test spot (flash) were projected at various positions on a border display and differences in RTs found, it was inferred that long (short) RTs were caused by subjectively bright (dim) backgrounds. 1This resea’rch was supported in part by funds provided for biological and medical research by the State of Washington Measure No. 171.
Lrrt~x TO THE EDITORS
1310
FIG. I. Dark ili,oh] the Hermann grtd IS vtetved tndirectly. These spots disappear uith direct (fove:tl) viwrnp. Flaring in the margins can be observed as well. The intensity of the effect curt be enhanced by vary ina the viewing distance from the figure.
Change in the amount of black on the white field should not affect the state of adaptation of the eye since the “sensitivity of, the retina is only determined by focul stutes ~~.~~~~r~t~~~~of single retinal spots” (H.-\R.Msand ACLHORN, 195% METHODS
AND
XPP.ARATUS
.A movabls 100 per cent rag vellum paper screen was stretched on a frame using a trampohne-type suspension. Black paper figures (horizontal dimension 30-j cm. vertical dimension 40.5 cm) were attached to the screen using small magnets. A 6 mm (1 min visual angle) test spot was projected flush on the back oithe viewing screen via a tube covered with 0403 in. matte acetate. The duration ofthe test tiash was 40 msec. A rmali neon fixation light located directly above the test spot placed the test flash 1.3’ on the upper retina. The test spot always remained in the same retinal location (within fixation errors) while the background. containing the black figure. moved. Thus compensations for variations in retinal sensitivity need not be made. The luminance of the viewing screen was I4 mL. The test spot luminous intensity is best specified for purposes of experimental replication as matching in brightness a luminance of 60.5 mL. The background moved in a vertical direction relative to the test spot; thus the test spot xvas positioned along the Y-coordinate ofa I cm unit Cartesian coordinate system tvhose origin is indicated by a 0 on the inserts of Fig. 2. Thr lower right-hand corner of the 90 recrangle was lowed at (- 1. ii. The bottom edge of the test spot was successively positioned at ail 1 cm points on the Y-axis beween -- IO and IO cm on rhs line X = 0. The relative position of this line is indicated by the numbers LO.0. - 10 on the inserts of Fig. 2. Each cm on the screen = 145 min visual angle = 7~ on the retina using Gullstrand‘s No. 2 simplified eye with nodal point at 16.5 mm ( EMSLEY, 1936).
Procedure The observer was seated in an opbthalrnoio~j~~l chair. equipped bvith a head rest. 23.3 mm i’rom the Lieu;@ screen. Loud *hire noise masked out auditory cuss to the Tess Hash. Observers LVP (normal vision) and D&4 (corrected to normal vision) reacted to the test flash by initiating a 3 set tixed fore-period and lifting ;t st~iusas soon as the test Hash appeared. The RT was recorded both on paper rapt and punched card. Data analysts was done by computer. Forty RTs \vere made for each of four counter-balanced orders of 21 :est spot locations by the t\\o observers for tive displays. Protiles of both observer’s data uere similar enough. although D-\‘s urre consistently longer, to warrant averaging.
Border-Contrast:
Comer Brightening Effects
1311
RESULTS
Mean reaction time (RT) as a function of vertical test spot location along the line X = 0 (X-Y Cartesian coordinates) for display figures with various corner angles is shown in Fig. 2. The inserts show the shape of the display figures and indicate the location of the coordinate system relative to these figures. On the white screen with no border present, RTs were shortest (test flash brightest-background inferred to be the dimmest). Reaction times along the long black border, O’, indicated a small, but detectable, increased change in subjective background brightness attributable to the border. Introduction of a 30” angle in the black display figure caused a noticeable increase in RTs just below the comer. The 60 and 90’ comers caused a progressive increase in RTs spreading up the Y-axis (the corner was at (- 1, 1) and slightly below the comer). Progressive brightening of the white background inferred from dim test spot flashes about the corner was identified as the comer angle increases. More points below Y = - 10 would have been needed for the 30, 60, and 90” curves to coalesce with the no border curve.
DISCUSSION
For a judiciously chosen background luminance and test spot brightness, subjective dimming of a test flash was easily perceived and nearly became invisible near the 90’ corner. Reaction time did provide a quantitative measure of this effect. Inferences concerning subjective background brightness changes were indirect. However, these inferences were likely correct in view of the similarity of these results to those obtained in similar border studies using direct brightness matching. Several models of neural inhibition have been formulated (reviewed by RATLIFF,1965). Although none was addressed directly to the effects of corners, one fact was clear: constants for most models were chosen so that the inhibitory influence must decrease to zero at about 20 min visual angle. Results of the experiment agreed well with this: at the point (0. - 10) which corresponds to 16 min visual angle from the comer, only slight elevation of RT over the no-border condition was found. When vertical stripes with horizontal sinusoidal variation in luminance was varied in wavelength (distance between the middle of successive dark or light bands), changes in subjective brightness of both the light and dark stripes were observed. BRYNGDAHL (1966) found that maximum subjective brightness occurred at wavelengths equal to 15 min visual angle while viewing the stripes at 2.5’ eccentricity. Thus the distance between the maximum luminance of the light and minimum of the dark stripes was 7.5 min visual angle. Contrast effects were found out to wavelengths equal to 110 min visual angle. Although the border effects in this study did not return to the “no border” condition within 14.5 min visual angle, extrapolations made from the data points indicate it is very possible that the curves would coalesce much before 55 min visual angle. Hence, it is unclear that spatial contrast effects studies by Bryngdahl are identical to the effects studied here. Perhaps overlapping border effects from different black figures interact in such a manner as to produce the grid effects seen in Fig. 1. The visual angles between figures for which the effect was seen varied from several minutes to about 4’. A four degree visual angle between black figures discredited this hypothesis if only border data obtained from the parafovea were consulted. It was thought possible that lateral excitation extends further from a black border as a function of increasing eccentricity on the retina. Reaction time methods were not suited to measure border contrast effects on the peripheral retina: RTs
LETTERTO THE EDITORS
1312
4
306 ,o
298 -
l
-10
o -10.
A 90.
;::: (1
IO 0
0
305
-10
;!\ ; i: i :
No lordor
-8
-6
-4 -2 0 2 4 Test Spot Position (‘f Coordino~e)
6
8
IO
FIG. 2.Mean reaction time (RT) based on 320 observations for each of 21 points to a flash of light (6 mm dia.) plotted against ordinate values (X-Y coordinate, cm) on the line X=0. The lower right-hand corner of the 90” rectangle (see inserts) was located at (- I, 1) using a I cm X-Y Cartesian coordinate system whose origin was indicated by a 0 on the inserts. The bottom edge of the test spot was successively positioned at all 1 cm points on the Y-axis between - 10 and IO cm on the line X=0 by moving the background. The relative position of this line was indicated by the numbers IO. 0, - IO on the inserts. Each cm on the screen = I.45 min visual angIe = 7~ on the retina. Long (short) RTs implied subjectire brightening (darkening) of the background.
became very long and did not reflect subjective changes in brightness. Perhaps a threshold signal-detection procedure would be appropriate. BRYNGDAHL (1966) did find that the perceived brightness of only the white stripes increased with larger wavelengths as his display was viewed at greater eccentricities (up to 10 ). There are no rods at the fovea, only cones, while the peripheral retina contains both rods and cdnes but a preponderance of rods. Rods have many more neural interconnections than do cones. Lateral neural inhibition within rod neural networks as an explanation for edge effects is appealing, but in this experiment the test flash was photopic. B&&Y (1960) favors neural inhibition as an explanation of lateral inhibition and discredits a chemical mechanism on the basis that similar effects occur with other senses (most importantly the skin) where chemical responsibility is unlikely. Because of the suprathreshold appearance of this border effect, the peripheral cones were likely responsible. But there is the question as to whether in vision duplicity held
Border-Contrast:
Comer Brightening Effects
1313
rigorously or whether rods continued to transmit information at high luminance levels (WEALE, 1951). Although no evidence obtained from this study could be considered a direct answer to this question, the behavioral study of border effects as a function of retinal location would help to provide an answer. H. PAYNE2 D. E. ANDERSON
w.
Washington State University, Pullman, Washington 99163, U.S.A.
REFERENCES BUSY, G. V. (1960). Neural inhibitory units of the eye and skin. Quantitative description of contrast phenomena. J. opt. Sec. Am. SO, 1060-1070. BWNGDAHL, 0. (1966). Perceived contrast variation with eccentricity of spatial sine-wave stimuli. Vision Res. 6, 553-565.
EMSLEY,H. (1936). Visual Optics, Tinling, Liverpool. FIORENTINI,A. (1957). Fovea1 and extrafoveal contrast threshold at a point of a n&-uniform
field. Atti. Fd.
Giorgio Ronchi 12, 180-186.
HARMS,H. and AULHORN,E. (1955). Studien uber den Grenzkontrast. I. Mitteilung. Ein neues Grenzphanomen. Graefes Arch. Ophrhal. 157, 3-23. PAYNE,W. H. (1967). Reaction times on a circle about the fovea. Science. N. Y. 155, 481. PAYNE,W. H. and WHITE,C. T. (1967). Extrafoveal visibility at a border. J. opt. Sot. Am. 57, 276-277. PAYNE,W. H. and ANDERSON, D. E. (1969). Border-visibility investigation of the Hermann Grid. J. opr. Sot. Am. 59, 229-23
1.
RATLIFF,F. (1965). Mach Bands: Quanritative Studies on Neural Networks in th\eRetina, Holden-Day, San Francisco. TSHERMAK,A. T. (1960). Hanbuch der Mormalen und Pathologischen Physiologie (edited by A. GETHE, G. BERGMANN,G. EMBDEN and A. ELLINGER). pp. 478-500, Springer, Berlin. WATRASIEWICZ, B. M. (1963). Measurement of the Mach effect in microscopy. Optica Acta 10, 209-216. WEALE,R. (1951). The duplicity theory of vision. Ann. R. Coil. Surg. 16, 16-35.
* From the Departments
of Computer Science, Psychology, and the Computing Center.
Vol. 9. pp. 1309-1313.
Pergamoa
Press 1969.
LETTER
BORDER-CONTRAST: (Received
Printed in Great
TO THE EDITORS
CORNER
8 January
Britain.
BRIGHTENING
EFFECTS1
1969; in revised form 26 March 1969)
WHEN a sharp black and white border was viewed several degrees from the line of direct vision, subjective brightening of the white background near the border was found (HARMS and AULHORN,1955; PAYNE and WHITE, 1967). This effect was not present with fovea1 viewing. Conversely, a uniform gradual change in luminance between dark and light fields gave rise to a subjectively light band at the edge of the high luminance field and a dark band at the edge of the low luminance field. This “Mach band” effect is only seen with fovea1 viewing (RATLIFF, 1965 ; FIORENTINI,1957). It has been suggested by TSHERMAK(1930) that edge effects should be differentiated from area contrast effects. The Hermann grid was investigated by PAYNEand ANDERSON (1969) to see if the grid effect, which was only seen with indirect vision, could possibly be attributed to edge rather than area contrast effects. Some of the results of this study indicated subjective brightening (lateral excitation) of the white background in the vicinity of the black right angle comers. The purpose of this study was to investigate the effect of a black comer angle on lateral excitation of the surrounding white field. Several quantitative methods of measuring border effects have been used. Viewing a point near a border with one eye and simultaneously making a brightness match with a test spot projected into the other eye was used (FIORENTINI,1957). A small slit of variable luminance was superimposed on the test location at the border and a brightness match was made (WATRGIENICZ, 1963). A small test spot was moved over the border display and difference thresholds above the background were measured (HARMSand AULHORN, 1955). Average evoked potential measurements taken from an observer’s occipital cortex to a test flash on the border display were also used (PAYNEand WHITE, 1967). All of these methods reflect subjective brightness at specified points on the border although some of the measurements were indirect (difference thresholds and evoked potentials). A small suprathreshold spot on a bright background appeared dim as compared to the same intensity spot viewed on a dark background. Simple reaction time (RT) was used as a measure of flash brightness (PAYNE, 1967). This flash brightness measure was accurate only within certain ranges of test flash brightness and background intensities. If the flash were too dim (bright) relative to the background, only long (short) RTs were obtained and did not accurately reflect flash brightness. If a constant intensity test spot (flash) were projected at various positions on a border display and differences in RTs found, it was inferred that long (short) RTs were caused by subjectively bright (dim) backgrounds. 1This resea’rch was supported in part by funds provided for biological and medical research by the State of Washington Measure No. 171.
Lrrt~x TO THE EDITORS
1310
FIG. I. Dark ili,oh] the Hermann grtd IS vtetved tndirectly. These spots disappear uith direct (fove:tl) viwrnp. Flaring in the margins can be observed as well. The intensity of the effect curt be enhanced by vary ina the viewing distance from the figure.
Change in the amount of black on the white field should not affect the state of adaptation of the eye since the “sensitivity of, the retina is only determined by focul stutes ~~.~~~~r~t~~~~of single retinal spots” (H.-\R.Msand ACLHORN, 195% METHODS
AND
XPP.ARATUS
.A movabls 100 per cent rag vellum paper screen was stretched on a frame using a trampohne-type suspension. Black paper figures (horizontal dimension 30-j cm. vertical dimension 40.5 cm) were attached to the screen using small magnets. A 6 mm (1 min visual angle) test spot was projected flush on the back oithe viewing screen via a tube covered with 0403 in. matte acetate. The duration ofthe test tiash was 40 msec. A rmali neon fixation light located directly above the test spot placed the test flash 1.3’ on the upper retina. The test spot always remained in the same retinal location (within fixation errors) while the background. containing the black figure. moved. Thus compensations for variations in retinal sensitivity need not be made. The luminance of the viewing screen was I4 mL. The test spot luminous intensity is best specified for purposes of experimental replication as matching in brightness a luminance of 60.5 mL. The background moved in a vertical direction relative to the test spot; thus the test spot xvas positioned along the Y-coordinate ofa I cm unit Cartesian coordinate system tvhose origin is indicated by a 0 on the inserts of Fig. 2. Thr lower right-hand corner of the 90 recrangle was lowed at (- 1. ii. The bottom edge of the test spot was successively positioned at ail 1 cm points on the Y-axis beween -- IO and IO cm on rhs line X = 0. The relative position of this line is indicated by the numbers LO.0. - 10 on the inserts of Fig. 2. Each cm on the screen = 145 min visual angle = 7~ on the retina using Gullstrand‘s No. 2 simplified eye with nodal point at 16.5 mm ( EMSLEY, 1936).
Procedure The observer was seated in an opbthalrnoio~j~~l chair. equipped bvith a head rest. 23.3 mm i’rom the Lieu;@ screen. Loud *hire noise masked out auditory cuss to the Tess Hash. Observers LVP (normal vision) and D&4 (corrected to normal vision) reacted to the test flash by initiating a 3 set tixed fore-period and lifting ;t st~iusas soon as the test Hash appeared. The RT was recorded both on paper rapt and punched card. Data analysts was done by computer. Forty RTs \vere made for each of four counter-balanced orders of 21 :est spot locations by the t\\o observers for tive displays. Protiles of both observer’s data uere similar enough. although D-\‘s urre consistently longer, to warrant averaging.
Border-Contrast:
Comer Brightening Effects
1311
RESULTS
Mean reaction time (RT) as a function of vertical test spot location along the line X = 0 (X-Y Cartesian coordinates) for display figures with various corner angles is shown in Fig. 2. The inserts show the shape of the display figures and indicate the location of the coordinate system relative to these figures. On the white screen with no border present, RTs were shortest (test flash brightest-background inferred to be the dimmest). Reaction times along the long black border, O’, indicated a small, but detectable, increased change in subjective background brightness attributable to the border. Introduction of a 30” angle in the black display figure caused a noticeable increase in RTs just below the comer. The 60 and 90’ comers caused a progressive increase in RTs spreading up the Y-axis (the corner was at (- 1, 1) and slightly below the comer). Progressive brightening of the white background inferred from dim test spot flashes about the corner was identified as the comer angle increases. More points below Y = - 10 would have been needed for the 30, 60, and 90” curves to coalesce with the no border curve.
DISCUSSION
For a judiciously chosen background luminance and test spot brightness, subjective dimming of a test flash was easily perceived and nearly became invisible near the 90’ corner. Reaction time did provide a quantitative measure of this effect. Inferences concerning subjective background brightness changes were indirect. However, these inferences were likely correct in view of the similarity of these results to those obtained in similar border studies using direct brightness matching. Several models of neural inhibition have been formulated (reviewed by RATLIFF,1965). Although none was addressed directly to the effects of corners, one fact was clear: constants for most models were chosen so that the inhibitory influence must decrease to zero at about 20 min visual angle. Results of the experiment agreed well with this: at the point (0. - 10) which corresponds to 16 min visual angle from the comer, only slight elevation of RT over the no-border condition was found. When vertical stripes with horizontal sinusoidal variation in luminance was varied in wavelength (distance between the middle of successive dark or light bands), changes in subjective brightness of both the light and dark stripes were observed. BRYNGDAHL (1966) found that maximum subjective brightness occurred at wavelengths equal to 15 min visual angle while viewing the stripes at 2.5’ eccentricity. Thus the distance between the maximum luminance of the light and minimum of the dark stripes was 7.5 min visual angle. Contrast effects were found out to wavelengths equal to 110 min visual angle. Although the border effects in this study did not return to the “no border” condition within 14.5 min visual angle, extrapolations made from the data points indicate it is very possible that the curves would coalesce much before 55 min visual angle. Hence, it is unclear that spatial contrast effects studies by Bryngdahl are identical to the effects studied here. Perhaps overlapping border effects from different black figures interact in such a manner as to produce the grid effects seen in Fig. 1. The visual angles between figures for which the effect was seen varied from several minutes to about 4’. A four degree visual angle between black figures discredited this hypothesis if only border data obtained from the parafovea were consulted. It was thought possible that lateral excitation extends further from a black border as a function of increasing eccentricity on the retina. Reaction time methods were not suited to measure border contrast effects on the peripheral retina: RTs
LETTERTO THE EDITORS
1312
4
306 ,o
298 -
l
-10
o -10.
A 90.
;::: (1
IO 0
0
305
-10
;!\ ; i: i :
No lordor
-8
-6
-4 -2 0 2 4 Test Spot Position (‘f Coordino~e)
6
8
IO
FIG. 2.Mean reaction time (RT) based on 320 observations for each of 21 points to a flash of light (6 mm dia.) plotted against ordinate values (X-Y coordinate, cm) on the line X=0. The lower right-hand corner of the 90” rectangle (see inserts) was located at (- I, 1) using a I cm X-Y Cartesian coordinate system whose origin was indicated by a 0 on the inserts. The bottom edge of the test spot was successively positioned at all 1 cm points on the Y-axis between - 10 and IO cm on the line X=0 by moving the background. The relative position of this line was indicated by the numbers IO. 0, - IO on the inserts. Each cm on the screen = I.45 min visual angIe = 7~ on the retina. Long (short) RTs implied subjectire brightening (darkening) of the background.
became very long and did not reflect subjective changes in brightness. Perhaps a threshold signal-detection procedure would be appropriate. BRYNGDAHL (1966) did find that the perceived brightness of only the white stripes increased with larger wavelengths as his display was viewed at greater eccentricities (up to 10 ). There are no rods at the fovea, only cones, while the peripheral retina contains both rods and cdnes but a preponderance of rods. Rods have many more neural interconnections than do cones. Lateral neural inhibition within rod neural networks as an explanation for edge effects is appealing, but in this experiment the test flash was photopic. B&&Y (1960) favors neural inhibition as an explanation of lateral inhibition and discredits a chemical mechanism on the basis that similar effects occur with other senses (most importantly the skin) where chemical responsibility is unlikely. Because of the suprathreshold appearance of this border effect, the peripheral cones were likely responsible. But there is the question as to whether in vision duplicity held
Border-Contrast:
Comer Brightening Effects
1313
rigorously or whether rods continued to transmit information at high luminance levels (WEALE, 1951). Although no evidence obtained from this study could be considered a direct answer to this question, the behavioral study of border effects as a function of retinal location would help to provide an answer. H. PAYNE2 D. E. ANDERSON
w.
Washington State University, Pullman, Washington 99163, U.S.A.
REFERENCES BUSY, G. V. (1960). Neural inhibitory units of the eye and skin. Quantitative description of contrast phenomena. J. opt. Sec. Am. SO, 1060-1070. BWNGDAHL, 0. (1966). Perceived contrast variation with eccentricity of spatial sine-wave stimuli. Vision Res. 6, 553-565.
EMSLEY,H. (1936). Visual Optics, Tinling, Liverpool. FIORENTINI,A. (1957). Fovea1 and extrafoveal contrast threshold at a point of a n&-uniform
field. Atti. Fd.
Giorgio Ronchi 12, 180-186.
HARMS,H. and AULHORN,E. (1955). Studien uber den Grenzkontrast. I. Mitteilung. Ein neues Grenzphanomen. Graefes Arch. Ophrhal. 157, 3-23. PAYNE,W. H. (1967). Reaction times on a circle about the fovea. Science. N. Y. 155, 481. PAYNE,W. H. and WHITE,C. T. (1967). Extrafoveal visibility at a border. J. opt. Sot. Am. 57, 276-277. PAYNE,W. H. and ANDERSON, D. E. (1969). Border-visibility investigation of the Hermann Grid. J. opr. Sot. Am. 59, 229-23
1.
RATLIFF,F. (1965). Mach Bands: Quanritative Studies on Neural Networks in th\eRetina, Holden-Day, San Francisco. TSHERMAK,A. T. (1960). Hanbuch der Mormalen und Pathologischen Physiologie (edited by A. GETHE, G. BERGMANN,G. EMBDEN and A. ELLINGER). pp. 478-500, Springer, Berlin. WATRASIEWICZ, B. M. (1963). Measurement of the Mach effect in microscopy. Optica Acta 10, 209-216. WEALE,R. (1951). The duplicity theory of vision. Ann. R. Coil. Surg. 16, 16-35.
* From the Departments
of Computer Science, Psychology, and the Computing Center.