0042-6989/92$5.00+ 0.00 Copyright 0 1992Pergamon Press plc
Vision Res. Vol. 32, No. 3, pp. 511-582, 1992 Printed in Great Britain. All rights reserved
Reduction of Fovea1 Desensitization with Blurred Backgrounds RUMEI Received
YUAN,*$
DEAN YAGER,*
ELIZABETH
17 January 1991; received for publication
THORPE
DAVIS?
31 July 1991
Photopic spatial desensitization and sensitization (Westheimer) functions were measured with sharp edged and blurred edged backgrounds. Data show that for 6-min backgrounds, desensitization is reduced for test flashes presented against blurred backgrounds compared to test flashes presented against backgrounds with sharp edges. We suggest that the transients from edges of retinal images, caused by high frequency fixational eye movements, contribute to the mechanisms of spatial desensitization and sensitization; the eflect of transients is reduced by blurring the backgrounds.
Westheimer functions Light adaptation Temporal transients Fovea Edge effects Blur Fixational eye movements
INTRODUCTION The classical explanation
of spatial desensitization and sensitization is based on the spatial center-surround antagonistic organization of a spatial receptive field (Westheimer, 1965, 1967). As the background size increases within the excitatory center of a receptive field, the adaptation level increases; thus, the threshold for the test stimulus increases (“desensitization”). When the background size exceeds the excitatory center, i.e. when it enters the inhibitory surround, the adapting level is reduced, and the test threshold decreases (“sensitization”). Some physiological studies on ganglion cells in monkey provide data that qualitatively correspond to this psychophysical finding. The luminance of a stimulus required to elicit a criterion response of a ganglion cell decreases as the stimulus size increases to fill the center mechanism with light and then, with the stimulus size further enlarged, increases as the antagonistic surround is reached (De Monasterio & Gouras, 1975). Moreover, a comparative study on measuring the “perceptive field” center psychophysically and the physiologically determined receptive field center, in monkey, showed that there is some correlation between the center sizes estimated by the two methods (Ransom-Hogg & Spillmann, 1980; Oehler, 1985; Spillmann, Ransom-Hogg & Oehler, 1987). However, the correspondence between Westheimer function dimensions and receptive field dimensions at the same retinal location is not good (Spillmann et al., *State College of Optometry, State University of New York, 100 E. 24 St, New York, NY 10010, U.S.A. tSchoo1 of Psychology, Georgia Institute of Technology, Atlanta, GA 30332-0170, U.S.A. $To whom all correspondence should be addressed.
Spatial vision
Theory of adaptation
1987). It is clear that a retinal ganglion cell spatial receptive field model does not provide a complete understanding of the mechanism of spatial desensitization and sensitization (see Verdon, 1988, for an excellent review). More studies have presented evidence to argue against the locus of sensitization being in a single ganglion cell. One of relevant findings is that dichoptic sensitization effects are found in the short-wavelength pathways (Verdon, 1988; Verdon, Haegerstrom-Portnoy & Adams, 1990). Furthermore, based on a multiple spatial-frequency channels theory, there would be several different spatial frequency channels at a given locus of the visual field, each corresponding to a different size of spatial receptive field at that locus of the visual field (e.g. Wilson & Bergen, 1979; Wilson & Gelb, 1984; also see a review by Olzak & Thomas, 1986). The retinal receptive field hypothesis of spatial sensitization has also been brought into question because borders may be effective regardless of the polarity of luminance relations in the visual field, at scotopic levels (e.g. Fry & Bartley, 1935; Crawford, 1940; Wyatt, 1972; Lennie & MacLeod, 1973). Although the processing for the scotopic Westheimer effect may be different from that for the photopic effect (see Hayhoe & Smith, 1989), the transients introduced by the stimulus borders may be important at photopic levels as well. Involuntary eye movements during fixation cause retinal image motion. The movement of edges across the spatial receptive field can cause temporal changes in illuminance at a given retinal spatial location. These transients may combine with the effect created by the steady illumination of the adapting field, resulting in a test threshold increase. Some sensitization persists when image motion is reduced by presenting the background for a very short 511
578
RIJMEI YUAN et al.
duration (Westheimer, 1967). However, studies on spatial sensitization conducted both with normal viewing and with stabilized-image conditions showed that at the fovea, the magnitude of sensitization is smaller with stabilized backgrounds {Tulunay-Keesey & Vassilev, 1974; Tulunay-Keesey & Jones, 1977; Hayhoe & Smith, 1989). For fovea1 cone vision, the threshold on small backgrounds was reduced by 0.38 log unit when the background was stabilized (Hayhoe & Smith, 1989). The mechanism proposed is that the transient stimulation of the edge is eliminated by stabilizing the retinal image; consequently, the threshold is not raised as much as in the condition with the transient effect present. The primary purpose of the present study was to re-examine the influences of the edge of the background on spatial densitization and sensitization. We reduced retinal image transients by blurring the edges of backgrounds and compared test thresholds with sharp-edged and blurred-edged backgrounds. We chose the fovea because the receptive fields are small and thus are more susceptible to the transient effects of an edge. Also, we wanted our data to be comparable with Hayhoe and Smith’s (1989) results from the fovea, and we wanted to replicate the original experiments on cone spatial desensitization and sensitization by Westheimer (1967) as closely as possible. METHODS
Apparatus Stimuli were presented on a Macintosh hi-resolution monochrome monitor; a Macintosh II computer contrailed the experiment and collected subjects’ responses via the keyboard. In order to obtain high spatial resolution, the subject viewed the monitor through a backward-positioned 6 x low-vision telescope, from a distance of 2 m, thus placing the monitor at an optical distance of 12 m from the subject. The pixel size of the monitor was 0.33125 mm, which at a distance of 12 m had an angular subtense of 5.69 set arc. The test stimulus was a circular spot 1.14 min arc in diameter, which was flashed for 60 msec with sharp temporal onsets and offsets. The stimulus was superimposed on a steady circular background which had the form of either a sharp-edged disk or a raised-cosine disk centered at the location of the corresponding sharpedged background, with a center plateau. All backgrounds, except in one control experiment, had a center luminance of 8.13 cd/m2; both the mean luminance and the total luminous flux for the stimulus pattern with a sharp edge were the same as those for the corresponding stimulus pattern with a blurred edge. The inflection points of the blurred backgrounds occurred at the same spatial locations as the edges of the corresponding sharp backgrounds. The diameters of the backgrounds ranged from 2.5 to 17 min arc, which were in the range used by Westheimer (1967). In a pilot study, we had found there was no further decrease of threshold for test flashes (i.e. spatial sensitization was not further increased) for
background diameters larger than 17 min arc. Therefore, we did not use backgrounds larger than 17 min in the study reported here. The stimuli and backgrounds were centered in a rectangular surround whose dimensions were 46 min vertical and 61 min horizontal; the surround had a luminance of 0.19 cd/m*. A profile of the 6-min backgrounds, with the stimulus and a portion of the surround, is shown in Fig. 1. Procedures Viewing was monocular with the right eye and with natural pupils, as in the Westheimer experiments. The subject was comfortably seated, placed his/her chin in a chin rest, and had the left eye occluded. The subject was instructed to maintain fixation on the center of the circular background. Before each block of trials, subjects adapted for 1 min after the background was turned on. Two different procedures were used, the Up-and-Down method and the method of constant stimuli. Up-and-Down method @on Bekesy), Test stimuli were flashed once per second. The duration of the test flash was 60 msec. The subject held down a button if the test flash was seen, and the luminance decreased in steps of O.lOlog unit. The subject held down the button until he/she did not see the test, and the button was released and the luminance increased by the same step size until the button was pressed. Thresholds were estimated by averaging ten to thirty reversal points for that condition. Method of constant stimuli. The method of constant stimuli, with a two-alternative forced-choice procedure, was used in the main experiment. Subjects initiated a trial by pressing a button, and two temporal intervals defined by tones, separated by 0.5 see, were presented. One of the intervals was blank, and the other contained a 60-msec stimulus test flash whose luminance was randomly chosen, without replacement, from a set of stimuli that spanned the range from nearly 100% correct to near 50% correct, as determined from pilot data. Four test luminance increments were used: 11.35, 22.79, 47.08 and 92.55 cd/m*. When the subject saw the stimutus in the first interval, he/she pressed “1” on the keyboard to respond; when the subject saw the stimulus in the second interval, he/she pressed ‘“2”. Feedback was
->1.14'
Vision Res. Vol. 32, No. 3, pp. 511-582, 1992 Printed in Great Britain. All rights reserved
Reduction of Fovea1 Desensitization with Blurred Backgrounds RUMEI Received
YUAN,*$
DEAN YAGER,*
ELIZABETH
17 January 1991; received for publication
THORPE
DAVIS?
31 July 1991
Photopic spatial desensitization and sensitization (Westheimer) functions were measured with sharp edged and blurred edged backgrounds. Data show that for 6-min backgrounds, desensitization is reduced for test flashes presented against blurred backgrounds compared to test flashes presented against backgrounds with sharp edges. We suggest that the transients from edges of retinal images, caused by high frequency fixational eye movements, contribute to the mechanisms of spatial desensitization and sensitization; the eflect of transients is reduced by blurring the backgrounds.
Westheimer functions Light adaptation Temporal transients Fovea Edge effects Blur Fixational eye movements
INTRODUCTION The classical explanation
of spatial desensitization and sensitization is based on the spatial center-surround antagonistic organization of a spatial receptive field (Westheimer, 1965, 1967). As the background size increases within the excitatory center of a receptive field, the adaptation level increases; thus, the threshold for the test stimulus increases (“desensitization”). When the background size exceeds the excitatory center, i.e. when it enters the inhibitory surround, the adapting level is reduced, and the test threshold decreases (“sensitization”). Some physiological studies on ganglion cells in monkey provide data that qualitatively correspond to this psychophysical finding. The luminance of a stimulus required to elicit a criterion response of a ganglion cell decreases as the stimulus size increases to fill the center mechanism with light and then, with the stimulus size further enlarged, increases as the antagonistic surround is reached (De Monasterio & Gouras, 1975). Moreover, a comparative study on measuring the “perceptive field” center psychophysically and the physiologically determined receptive field center, in monkey, showed that there is some correlation between the center sizes estimated by the two methods (Ransom-Hogg & Spillmann, 1980; Oehler, 1985; Spillmann, Ransom-Hogg & Oehler, 1987). However, the correspondence between Westheimer function dimensions and receptive field dimensions at the same retinal location is not good (Spillmann et al., *State College of Optometry, State University of New York, 100 E. 24 St, New York, NY 10010, U.S.A. tSchoo1 of Psychology, Georgia Institute of Technology, Atlanta, GA 30332-0170, U.S.A. $To whom all correspondence should be addressed.
Spatial vision
Theory of adaptation
1987). It is clear that a retinal ganglion cell spatial receptive field model does not provide a complete understanding of the mechanism of spatial desensitization and sensitization (see Verdon, 1988, for an excellent review). More studies have presented evidence to argue against the locus of sensitization being in a single ganglion cell. One of relevant findings is that dichoptic sensitization effects are found in the short-wavelength pathways (Verdon, 1988; Verdon, Haegerstrom-Portnoy & Adams, 1990). Furthermore, based on a multiple spatial-frequency channels theory, there would be several different spatial frequency channels at a given locus of the visual field, each corresponding to a different size of spatial receptive field at that locus of the visual field (e.g. Wilson & Bergen, 1979; Wilson & Gelb, 1984; also see a review by Olzak & Thomas, 1986). The retinal receptive field hypothesis of spatial sensitization has also been brought into question because borders may be effective regardless of the polarity of luminance relations in the visual field, at scotopic levels (e.g. Fry & Bartley, 1935; Crawford, 1940; Wyatt, 1972; Lennie & MacLeod, 1973). Although the processing for the scotopic Westheimer effect may be different from that for the photopic effect (see Hayhoe & Smith, 1989), the transients introduced by the stimulus borders may be important at photopic levels as well. Involuntary eye movements during fixation cause retinal image motion. The movement of edges across the spatial receptive field can cause temporal changes in illuminance at a given retinal spatial location. These transients may combine with the effect created by the steady illumination of the adapting field, resulting in a test threshold increase. Some sensitization persists when image motion is reduced by presenting the background for a very short 511
578
RIJMEI YUAN et al.
duration (Westheimer, 1967). However, studies on spatial sensitization conducted both with normal viewing and with stabilized-image conditions showed that at the fovea, the magnitude of sensitization is smaller with stabilized backgrounds {Tulunay-Keesey & Vassilev, 1974; Tulunay-Keesey & Jones, 1977; Hayhoe & Smith, 1989). For fovea1 cone vision, the threshold on small backgrounds was reduced by 0.38 log unit when the background was stabilized (Hayhoe & Smith, 1989). The mechanism proposed is that the transient stimulation of the edge is eliminated by stabilizing the retinal image; consequently, the threshold is not raised as much as in the condition with the transient effect present. The primary purpose of the present study was to re-examine the influences of the edge of the background on spatial densitization and sensitization. We reduced retinal image transients by blurring the edges of backgrounds and compared test thresholds with sharp-edged and blurred-edged backgrounds. We chose the fovea because the receptive fields are small and thus are more susceptible to the transient effects of an edge. Also, we wanted our data to be comparable with Hayhoe and Smith’s (1989) results from the fovea, and we wanted to replicate the original experiments on cone spatial desensitization and sensitization by Westheimer (1967) as closely as possible. METHODS
Apparatus Stimuli were presented on a Macintosh hi-resolution monochrome monitor; a Macintosh II computer contrailed the experiment and collected subjects’ responses via the keyboard. In order to obtain high spatial resolution, the subject viewed the monitor through a backward-positioned 6 x low-vision telescope, from a distance of 2 m, thus placing the monitor at an optical distance of 12 m from the subject. The pixel size of the monitor was 0.33125 mm, which at a distance of 12 m had an angular subtense of 5.69 set arc. The test stimulus was a circular spot 1.14 min arc in diameter, which was flashed for 60 msec with sharp temporal onsets and offsets. The stimulus was superimposed on a steady circular background which had the form of either a sharp-edged disk or a raised-cosine disk centered at the location of the corresponding sharpedged background, with a center plateau. All backgrounds, except in one control experiment, had a center luminance of 8.13 cd/m2; both the mean luminance and the total luminous flux for the stimulus pattern with a sharp edge were the same as those for the corresponding stimulus pattern with a blurred edge. The inflection points of the blurred backgrounds occurred at the same spatial locations as the edges of the corresponding sharp backgrounds. The diameters of the backgrounds ranged from 2.5 to 17 min arc, which were in the range used by Westheimer (1967). In a pilot study, we had found there was no further decrease of threshold for test flashes (i.e. spatial sensitization was not further increased) for
background diameters larger than 17 min arc. Therefore, we did not use backgrounds larger than 17 min in the study reported here. The stimuli and backgrounds were centered in a rectangular surround whose dimensions were 46 min vertical and 61 min horizontal; the surround had a luminance of 0.19 cd/m*. A profile of the 6-min backgrounds, with the stimulus and a portion of the surround, is shown in Fig. 1. Procedures Viewing was monocular with the right eye and with natural pupils, as in the Westheimer experiments. The subject was comfortably seated, placed his/her chin in a chin rest, and had the left eye occluded. The subject was instructed to maintain fixation on the center of the circular background. Before each block of trials, subjects adapted for 1 min after the background was turned on. Two different procedures were used, the Up-and-Down method and the method of constant stimuli. Up-and-Down method @on Bekesy), Test stimuli were flashed once per second. The duration of the test flash was 60 msec. The subject held down a button if the test flash was seen, and the luminance decreased in steps of O.lOlog unit. The subject held down the button until he/she did not see the test, and the button was released and the luminance increased by the same step size until the button was pressed. Thresholds were estimated by averaging ten to thirty reversal points for that condition. Method of constant stimuli. The method of constant stimuli, with a two-alternative forced-choice procedure, was used in the main experiment. Subjects initiated a trial by pressing a button, and two temporal intervals defined by tones, separated by 0.5 see, were presented. One of the intervals was blank, and the other contained a 60-msec stimulus test flash whose luminance was randomly chosen, without replacement, from a set of stimuli that spanned the range from nearly 100% correct to near 50% correct, as determined from pilot data. Four test luminance increments were used: 11.35, 22.79, 47.08 and 92.55 cd/m*. When the subject saw the stimutus in the first interval, he/she pressed “1” on the keyboard to respond; when the subject saw the stimulus in the second interval, he/she pressed ‘“2”. Feedback was
->1.14'
