Pupillary responses to stimulus structure^ colour and movement J. L. Barbur, A. J. Harlow and A. Sahraie
Applied Vision Research Centre, The City University, Northampton Square, London FCl V OHB, UK (Received 7 November 1991) Pupillary responses to stimuli which favour the preferential stimulation of neural mechanisms involved in the detection of visual attributes such as colour, spatial structure, movement and light flux changes on the retina have been measured and compared. Pupil responses to a decrement in stimulus luminance (i.e., a flash of darkness), suggest that at least three components are involved in this response, their relative contribution being determined largely by stimulus size, contrast and presentation time. A comparison of pupil responses to gratings of equal and lower space-averaged luminance shows that the amplitude of pupillary constriction at grating onset for the equal luminance condition is about twice that measured with similar gratings in the lower luminance condition. Pupillary responses to chromatic isoluminant gratings are in general of longer latency when compared to responses of similar amplitude elicited by achromatic gratings. Small pupillary constrictions elicited by the onset of coherent movement in dynamic, random dot patterns are also demonstrated under stimulus conditions which eliminate pupillary responses to sudden light flux changes on the retina. The results support an earlier hypothesis which suggests that the onset of sudden changes in neural activity in the visual cortex when a visual stimulus is presented to the eye causes an overall perturbation which weakens transiently the regulatory inhibitory input to the pupillomotor nucleus. This, in turn, results in a transient increase in the efferent parasympathetic innervation of the iris sphincter muscle and hence the observed constriction of the pupil. The characteristics of the pupillary response reflect the properties of the mechanisms and the number of neurones which participate in the detection of each simulus attribute.
The techniques developed for the assessment of image reproduction in optical systems' have been used successfully to describe image quality in the human eye^'^, and have formed the basis of many psychophysical and electrophysiological studies of spatial vision*'^. The majority of psychophysical investigations involve the measurement of the minimum contrast required to detect a grating pattern as a function of spatial frequency. Psychophysical studies involving the use of suprathreshold gratings have been less numerous*, although suprathreshold gratings have been used frequently in electrophysiological and non-invasive studies^. Since the pupil response is associated classically with changes in light flux level on the retina, the use of sinusoidal gratings to study the function of the pupil response has been very limited. Of greater interest has been the effect of pupil size on measurements of contrast sensitivity, which has received careful examination^. Barbur and Forsyth reported the existence of a small pupillary constriction in response to a sinusoidal grating', and the absence of such responses in cases of hemianopia caused by damage to central visual pathways'"'^^ Pupillary responses to checker board pattern reversal have been observed and reported previously^^-'^. Such findings have often been explained in terms of stimulus induced accommodation changes.
The generation of the stimuli and the measurements of the pupil response were carried out on the P-SCAN 100 system which has been described previously'®''"'. The apparatus allows the simultaneous, binocular measurement of pupil size and the corresponding 2-D movements of the eyes. The statistical methods employed in extracting the parameters of interest are equivalent to fitting the best circle to the pupil, and yield a resolution
© 1992 Butterworth-Heinemann for British College of Optometrists 0275-5408/92/020137-05
Ophthal. Physiol. Opt., 1992, Vol. 12, April
or a light reflex response driven by spatially localized increments in light flux level on the retina. Systematic pupillary constrictions to foveal exchanges of coloured lights have also been reported'"*", together with paradoxical contractions of the pupil in response to flashes of darkness'*. In this paper we extend the results of previous studies and report new findings on pupillary responses to increments and decrements in stimulus luminance as a function of stimulus duration, to gratings of equal and lower space-averaged luminance, to isoluminant chromatic gratings and to movement onset in dynamic random dot patterns. The results suggest that several pathways may be involved in the control of the pupil response, and help to explain some earlier findings.
Methods
137
Pupiilary responses to stimulus structure, colour and movement: J. L. Barbur et al. for the measurement of pupil diameter and eyemovements better than O.Oi mm and 4 min arc, respectively. T"he subject viewed binocularly a well defined fixation stimulus which was generated in the middle of a uniform background field of angular subtense 20 x 15 degrees. A high resolution, 60 Hz, colour monitor (Hewlett Packard Model DI 187A) Lichtmess technik was used to generate the visual stimuli. The chromaticity co-ordinates of each phosphor were measured using a Gamma Scientific (Model DR-2) Lichtmess technik telespectroradiometer and the phosphor luminance versus applied gun voltage relationship was measured using a luminance meter (Model 1003) Lichtmess technik for each possible gun voltage value. The experimental programs made full use of the calibration data file by computing the nearest red, green and blue phosphor luminances required to reproduce a specified luminance, chromaticity triplet. The colorimetric transformations used have been described previously and can be found in Wyszecki and Stiles^*. The use of interpolation techniques was therefore avoided. The motion stimuli were generated on a 70 Hz, colour monitor (Hewlett Packard Model DI 182B) which provided a uniform background of angular subtense of 17 X 13 degrees. Visual stimuli Light flux changes. The visual stimulus was a square of side subtending 6 degrees and was presented in the centre of a uniform background field of luminance 34cdm"^ and x,}'-chromaticity coordinates 0.292, 0.326. The luminance of the test target was either 102.6 or 11.26cdm"^, and was presented to the subject as a square pulse lasting for either 0.55 or 1.1s. The contrast of the stimulus defined as the logarithm of the ratio of the luminance of the target to that of the background field was ±0.48. Achromatic gratings. The sinusoidal grating pattern was circular, of 80% contrast and subtended a visual angle of 5 degrees. Each grating was presented as a flash of duration 1.1 s. In the equal luminance condition, the space-averaged luminance of the grating was equal to that of the uniform background field. The measured luminance was usually within 0.3% of that measured over the same display area for the uniform background field. In this condition the onset of the grating results in grating bars which represent localized increments and decrements in light flux by comparison with the uniform background. In the lower luminance condition, the grating bars represent only" localized decrements in luminance by comparison with the uniform background field. The peak luminance in the grating remains the same as that of the uniform background. For a grating of 80% contrast, the averaged luminance of the grating was reduced to 19 cd m"^. Isoluminant chromatic gratings. Isoluminant, red/green, chromatic gratings of spatial frequency 1.21 c/deg were generated by modulation of the red and green phosphor luminances. The red phosphor luminance output was modulated sinusoidally so as to yield a specified luminance contrast for the red component. Isoiuminance was then set by adjusting the luminance contrast of the green component around the CIE isoiuminance point so as to yield minimum apparent motion during phase reversal of the red/green grating.
138
Ophthal. Physiol. Opt., 1992, Vol. 12, April
Coherent movement in dynamic, random dot patterns. The patterns employed to study the pupil motion response were similar to those described by Baker and Braddick''. The display background subtended a visual angle of 17 x 13 degrees and had a luminance of 13cdm~^; 80 discrete elements, each subtending a visual angle of 0.22 x 0.26 degrees were used to generate either random movement or coherent movement. The luminance of the dots was 76cd m~^ and the coherent movement interval was always preceded and followed by 3.2 and 2 s of dynamic noise, respectively. The same displacement step was used during the random and the coherent motion stages of the stimulus, although the step size changed appropriately for difl"erent speeds of movement. The test stimulus consisted of two rectangular areas of 7 x 5.2 degrees positioned symmetrically with respect to the centre of the uniform background and separated by 1.8 degrees. The fixation stimulus was placed in the centre of the background and the direction of coherent movement was always either towards or away from the fixation stimulus. Unless otherwise stated, each pupil response trace represents the average of 64 single measurements.
Results Pupil light refiex Figure 1 shows pupil responses to the presentation of a square stimulus which is either of positive or negative contrast with respect to the uniform background field. 3.2 h
2.6
Time (s) Figure I Comparison of pupil responses to incremental and decremental changes in light flux level on the retina. The light stimulus was a square target of size 6 degrees and was presented in the centre of a uniform background field of luminance Lj,, equal to 34 cd m~^. Two stimulus presentation times were used: O, 0.55 s, +ve contrast; • , 0.55s, —ve contrast; D , I l s , +ve contrast; • , 1.1s, —ve contrast. The stimulus trace, shown along the abscissa, indicates the 1.1s condition. The luminance of the test stimulus, L,, was either 102.6 or 11.26 cd m " ^. These luminance values yield corresponding contrasts, C, of +0.48, where C = log( L,/Lb). The .x,>'-chromaticity coordinates of the background and the stimulus were 0.292, 0.326. The data obtained with a stimulus of negative contrast were shifted up by 0.1 mm for clarity of presentation. Each trace represents the average of 32 measurements
Pupillary responses to stimulus structure, colour and movement: J. L. Barbur et al. The stimulus of positive contrast produces the classic pupiilogram with a latency of 220 ms and a large response amplitude of some 0.3 mm. The pupil response amplitude is measured as a change in pupil diameter, since this measure is affected less by the initial size of the pupil^^. The pupil responses to a flash of negative contrast show an initial constriction at flash onset, followed very rapidly by a large dilation triggered, presumably, by the decrease in light flux level on the retina. The dilation amplitude varies with the size of the stimulus relative to that of the background field and is not always measurable'°. The stimulus ofl"set triggers a clear constriction of the pupil which is likely to reflect the light flux increment at stimulus olTset. Since this response is generally of large amplitude, it can make the preceding pupillary changes undetectable for short stimulus durations and this can result in what appears to be a clear constriction of the pupil to a flash of darkness'°'''^. Pupil responses to grating patterns Figure 2 shows a comparison of pupillary responses to a chromatic and an achromatic grating of 1.21 c/deg presented as a 1.1 s flash. This grating frequency was selected for presentation since the two responses are of appoximately equal amplitude and therefore the difference in response latency cannot be attributed simply to a difference in afferent signal amplitude. The response amplitude is about 0.08 mm for both gratings, the latency of the achromatic grating is 280 ms and that measured for the chromatic grating is 360 ms. Another feature of interest is the absence of the small pupil constriction at grating offset in the case of the chromatic grating stimulus. When achromatic gratings are used, the small constriction at grating offset is also absent but only in
the high frequency range (i.e., for grating spatial frequencies above some 12 c/deg). Figure 3 shows a comparison of pupil responses to achromatic gratings of equal and lower space-averaged luminance, when compared with that ol' the uniform background field. In the lower luminance condition, the onset of the grating causes no increment in light flux ievel at any point in the image. The gratings were of 80% contrast and the spatial frequency was 4.8 c/deg. The pupil response amplitude for the lower luminance condition was about half that elicited with the equal luminance grating. The increase in light flux level at stimulus offset in the case of the lower luminance grating triggers a pupil light reflex response similar to that observed in Figure 1. Pupil responses to coherent movement Figure 4 shows pupil responses triggered by the onset of coherent movement in a dynamic random dot pattern. The traces presented refer to a range of speeds from 1.27 to 7.6 degrees s~'. The pupil response amplitudes are all
Applied Vision Research Centre, The City University, Northampton Square, London FCl V OHB, UK (Received 7 November 1991) Pupillary responses to stimuli which favour the preferential stimulation of neural mechanisms involved in the detection of visual attributes such as colour, spatial structure, movement and light flux changes on the retina have been measured and compared. Pupil responses to a decrement in stimulus luminance (i.e., a flash of darkness), suggest that at least three components are involved in this response, their relative contribution being determined largely by stimulus size, contrast and presentation time. A comparison of pupil responses to gratings of equal and lower space-averaged luminance shows that the amplitude of pupillary constriction at grating onset for the equal luminance condition is about twice that measured with similar gratings in the lower luminance condition. Pupillary responses to chromatic isoluminant gratings are in general of longer latency when compared to responses of similar amplitude elicited by achromatic gratings. Small pupillary constrictions elicited by the onset of coherent movement in dynamic, random dot patterns are also demonstrated under stimulus conditions which eliminate pupillary responses to sudden light flux changes on the retina. The results support an earlier hypothesis which suggests that the onset of sudden changes in neural activity in the visual cortex when a visual stimulus is presented to the eye causes an overall perturbation which weakens transiently the regulatory inhibitory input to the pupillomotor nucleus. This, in turn, results in a transient increase in the efferent parasympathetic innervation of the iris sphincter muscle and hence the observed constriction of the pupil. The characteristics of the pupillary response reflect the properties of the mechanisms and the number of neurones which participate in the detection of each simulus attribute.
The techniques developed for the assessment of image reproduction in optical systems' have been used successfully to describe image quality in the human eye^'^, and have formed the basis of many psychophysical and electrophysiological studies of spatial vision*'^. The majority of psychophysical investigations involve the measurement of the minimum contrast required to detect a grating pattern as a function of spatial frequency. Psychophysical studies involving the use of suprathreshold gratings have been less numerous*, although suprathreshold gratings have been used frequently in electrophysiological and non-invasive studies^. Since the pupil response is associated classically with changes in light flux level on the retina, the use of sinusoidal gratings to study the function of the pupil response has been very limited. Of greater interest has been the effect of pupil size on measurements of contrast sensitivity, which has received careful examination^. Barbur and Forsyth reported the existence of a small pupillary constriction in response to a sinusoidal grating', and the absence of such responses in cases of hemianopia caused by damage to central visual pathways'"'^^ Pupillary responses to checker board pattern reversal have been observed and reported previously^^-'^. Such findings have often been explained in terms of stimulus induced accommodation changes.
The generation of the stimuli and the measurements of the pupil response were carried out on the P-SCAN 100 system which has been described previously'®''"'. The apparatus allows the simultaneous, binocular measurement of pupil size and the corresponding 2-D movements of the eyes. The statistical methods employed in extracting the parameters of interest are equivalent to fitting the best circle to the pupil, and yield a resolution
© 1992 Butterworth-Heinemann for British College of Optometrists 0275-5408/92/020137-05
Ophthal. Physiol. Opt., 1992, Vol. 12, April
or a light reflex response driven by spatially localized increments in light flux level on the retina. Systematic pupillary constrictions to foveal exchanges of coloured lights have also been reported'"*", together with paradoxical contractions of the pupil in response to flashes of darkness'*. In this paper we extend the results of previous studies and report new findings on pupillary responses to increments and decrements in stimulus luminance as a function of stimulus duration, to gratings of equal and lower space-averaged luminance, to isoluminant chromatic gratings and to movement onset in dynamic random dot patterns. The results suggest that several pathways may be involved in the control of the pupil response, and help to explain some earlier findings.
Methods
137
Pupiilary responses to stimulus structure, colour and movement: J. L. Barbur et al. for the measurement of pupil diameter and eyemovements better than O.Oi mm and 4 min arc, respectively. T"he subject viewed binocularly a well defined fixation stimulus which was generated in the middle of a uniform background field of angular subtense 20 x 15 degrees. A high resolution, 60 Hz, colour monitor (Hewlett Packard Model DI 187A) Lichtmess technik was used to generate the visual stimuli. The chromaticity co-ordinates of each phosphor were measured using a Gamma Scientific (Model DR-2) Lichtmess technik telespectroradiometer and the phosphor luminance versus applied gun voltage relationship was measured using a luminance meter (Model 1003) Lichtmess technik for each possible gun voltage value. The experimental programs made full use of the calibration data file by computing the nearest red, green and blue phosphor luminances required to reproduce a specified luminance, chromaticity triplet. The colorimetric transformations used have been described previously and can be found in Wyszecki and Stiles^*. The use of interpolation techniques was therefore avoided. The motion stimuli were generated on a 70 Hz, colour monitor (Hewlett Packard Model DI 182B) which provided a uniform background of angular subtense of 17 X 13 degrees. Visual stimuli Light flux changes. The visual stimulus was a square of side subtending 6 degrees and was presented in the centre of a uniform background field of luminance 34cdm"^ and x,}'-chromaticity coordinates 0.292, 0.326. The luminance of the test target was either 102.6 or 11.26cdm"^, and was presented to the subject as a square pulse lasting for either 0.55 or 1.1s. The contrast of the stimulus defined as the logarithm of the ratio of the luminance of the target to that of the background field was ±0.48. Achromatic gratings. The sinusoidal grating pattern was circular, of 80% contrast and subtended a visual angle of 5 degrees. Each grating was presented as a flash of duration 1.1 s. In the equal luminance condition, the space-averaged luminance of the grating was equal to that of the uniform background field. The measured luminance was usually within 0.3% of that measured over the same display area for the uniform background field. In this condition the onset of the grating results in grating bars which represent localized increments and decrements in light flux by comparison with the uniform background. In the lower luminance condition, the grating bars represent only" localized decrements in luminance by comparison with the uniform background field. The peak luminance in the grating remains the same as that of the uniform background. For a grating of 80% contrast, the averaged luminance of the grating was reduced to 19 cd m"^. Isoluminant chromatic gratings. Isoluminant, red/green, chromatic gratings of spatial frequency 1.21 c/deg were generated by modulation of the red and green phosphor luminances. The red phosphor luminance output was modulated sinusoidally so as to yield a specified luminance contrast for the red component. Isoiuminance was then set by adjusting the luminance contrast of the green component around the CIE isoiuminance point so as to yield minimum apparent motion during phase reversal of the red/green grating.
138
Ophthal. Physiol. Opt., 1992, Vol. 12, April
Coherent movement in dynamic, random dot patterns. The patterns employed to study the pupil motion response were similar to those described by Baker and Braddick''. The display background subtended a visual angle of 17 x 13 degrees and had a luminance of 13cdm~^; 80 discrete elements, each subtending a visual angle of 0.22 x 0.26 degrees were used to generate either random movement or coherent movement. The luminance of the dots was 76cd m~^ and the coherent movement interval was always preceded and followed by 3.2 and 2 s of dynamic noise, respectively. The same displacement step was used during the random and the coherent motion stages of the stimulus, although the step size changed appropriately for difl"erent speeds of movement. The test stimulus consisted of two rectangular areas of 7 x 5.2 degrees positioned symmetrically with respect to the centre of the uniform background and separated by 1.8 degrees. The fixation stimulus was placed in the centre of the background and the direction of coherent movement was always either towards or away from the fixation stimulus. Unless otherwise stated, each pupil response trace represents the average of 64 single measurements.
Results Pupil light refiex Figure 1 shows pupil responses to the presentation of a square stimulus which is either of positive or negative contrast with respect to the uniform background field. 3.2 h
2.6
Time (s) Figure I Comparison of pupil responses to incremental and decremental changes in light flux level on the retina. The light stimulus was a square target of size 6 degrees and was presented in the centre of a uniform background field of luminance Lj,, equal to 34 cd m~^. Two stimulus presentation times were used: O, 0.55 s, +ve contrast; • , 0.55s, —ve contrast; D , I l s , +ve contrast; • , 1.1s, —ve contrast. The stimulus trace, shown along the abscissa, indicates the 1.1s condition. The luminance of the test stimulus, L,, was either 102.6 or 11.26 cd m " ^. These luminance values yield corresponding contrasts, C, of +0.48, where C = log( L,/Lb). The .x,>'-chromaticity coordinates of the background and the stimulus were 0.292, 0.326. The data obtained with a stimulus of negative contrast were shifted up by 0.1 mm for clarity of presentation. Each trace represents the average of 32 measurements
Pupillary responses to stimulus structure, colour and movement: J. L. Barbur et al. The stimulus of positive contrast produces the classic pupiilogram with a latency of 220 ms and a large response amplitude of some 0.3 mm. The pupil response amplitude is measured as a change in pupil diameter, since this measure is affected less by the initial size of the pupil^^. The pupil responses to a flash of negative contrast show an initial constriction at flash onset, followed very rapidly by a large dilation triggered, presumably, by the decrease in light flux level on the retina. The dilation amplitude varies with the size of the stimulus relative to that of the background field and is not always measurable'°. The stimulus ofl"set triggers a clear constriction of the pupil which is likely to reflect the light flux increment at stimulus olTset. Since this response is generally of large amplitude, it can make the preceding pupillary changes undetectable for short stimulus durations and this can result in what appears to be a clear constriction of the pupil to a flash of darkness'°'''^. Pupil responses to grating patterns Figure 2 shows a comparison of pupillary responses to a chromatic and an achromatic grating of 1.21 c/deg presented as a 1.1 s flash. This grating frequency was selected for presentation since the two responses are of appoximately equal amplitude and therefore the difference in response latency cannot be attributed simply to a difference in afferent signal amplitude. The response amplitude is about 0.08 mm for both gratings, the latency of the achromatic grating is 280 ms and that measured for the chromatic grating is 360 ms. Another feature of interest is the absence of the small pupil constriction at grating offset in the case of the chromatic grating stimulus. When achromatic gratings are used, the small constriction at grating offset is also absent but only in
the high frequency range (i.e., for grating spatial frequencies above some 12 c/deg). Figure 3 shows a comparison of pupil responses to achromatic gratings of equal and lower space-averaged luminance, when compared with that ol' the uniform background field. In the lower luminance condition, the onset of the grating causes no increment in light flux ievel at any point in the image. The gratings were of 80% contrast and the spatial frequency was 4.8 c/deg. The pupil response amplitude for the lower luminance condition was about half that elicited with the equal luminance grating. The increase in light flux level at stimulus offset in the case of the lower luminance grating triggers a pupil light reflex response similar to that observed in Figure 1. Pupil responses to coherent movement Figure 4 shows pupil responses triggered by the onset of coherent movement in a dynamic random dot pattern. The traces presented refer to a range of speeds from 1.27 to 7.6 degrees s~'. The pupil response amplitudes are all
