Behavioural Processes, 7 (1982) 361-315 Elsevier Scientific Publishing Company,
VISUAL ACUITY SQUIRRELS
GERALD
AND SPATIAL
H. JACOBS,
Department
367 Amsterdam
DAVID
of Psychology,
Printed
CONTRAST
G. BIRCH*
University
-
SENSITIVITY
and BARBARA
of California,
in The Netherlands
BLAKESLEE
Santa Barbara,
*Present address: Berman-Gund Laboratory for the Study of Retinal Massachusetts Eye and Ear Infirmary, Boston, MA 02114 (U.S.A.) (Accepted
8 June
IN TREE
CA 93106
(U.S.A.)
Degenerations,
1982)
ABSTRACT Jacobs, G.H., sensitivity
Birch, D.G. and Blakeslee, B., 1982. Visual acuity in tree squirrels. Behav. Processes, 7 : 367-37 5.
and spatial
contrast
The visual sensitivity of tree squirrels from three different species (two western gray squirrels, Sciurus griseus; three fox squirrels, Sciurus niger; and an eastern gray squirrel, Sciurus carolinensis) was measured for spatial patterns defined by luminance differences. Spatial contrast sensitivity functions were determined in behavioral discrimination experiments in which the stimuli were sinusoidally-modulated gratings. At an average luminance level of 3.4 cd/m2 these squirrels were maximally sensitive to spatial frequencies of about 0.5 cycles/degree (c/d). These experiments, in conjunction with additional measurements involving discrimination of square-wave grating patterns, also indicate that the finest gratings these squirrels can discriminate average 2.2 c/d (SD = 0.42 c/d). There are no obvious differences in resolution acuity among the three species. Grating acuity also was measured at a much higher luminance level (340 cd/ml). The results of this experiment indicate that the spatial acuity of the tree squirrel increases with luminance level, reaching a maximum of 3.9 c/d at 340 cd/m2.
INTRODUCTION
Tree squirrels are diurnal, arboreal rodents. Their retinas are unusual among those of most mammals in that they contain more cones than rods; only about 40% of all tree-squirrel photoreceptors can be classified as rods (Cohen, 1964; West and Dowling, 1975; Yolton, 1975). It has been verified in both electrophysiological (Arden and Tansley, 1955; Gouras, 1964; Green and Dowling, 1975) and behavioral (Arden and Silver, 1962; Silver, 1966; Dippner and Armington, 1971; Jacobs, 1974) experiments that these two classes of photoreceptors provide the structural basis for functionally separable scotopic and photopic capacities. In addition to having a unique mixture of rods and cones, tree squirrel retinas are also notable for their lack of great regional variations. Thus, al-
03766357/82/0000-0000/$02.75
0 1982
Elsevier
Scientific
Publishing
Company
368
though these retinas show some central retinal specializations in the form of a visual streak and an area centralis, ganglion cell density is only modestly variant, with perhaps no more than a three- or four-fold decrease in density from the center of the retina to the periphery (Hughes, 1977). Although Walls (1942) believed that this so-called “universal macularity”, seen in the squirrels and other strongly diurnal species, should be associated with high spatial resolution, there is clearly nothing inevitable about the relationship. For instance, we recently found that ground squirrels, rodents also with cone-rich retinas and with modest regional variations in ganglion cell density, have relatively poor visual acuity. At the limit these animals could not resolve luminance grating targets having spatial frequencies of greater than 4 cycles/degree (c/d) (Jacobs et al., 1980). In general, spatial resolution depends not only on ganglion cell density and the retinal distribution of ganglion cells, but also on optical factors, the grain of the photoreceptor mosaic, and the resolving power of the neural visual system (Hughes, 1977). However, knowledge of these factors alone is not sufficient to permit a specification of the visual resolution capacities of the animal. For that, behavioral measurements are required. Accordingly, to determine what the spatial resolution capacities are for tree squirrels, we measured the abilities of representatives from three species to discriminate luminance gratings to obtain estimates of their visual acuity and contrast sensitivity. MATERIALS
AND METHODS
Measurements were made on individuals from three common species of North American tree squirrels: two (one male, one female) western gray squirrels (Sciurus griseus), three (one male, two females) fox squirrels (S. niger), and one female eastern gray squirrel (S. carolinensis). The western gray squirrels were trapped in Santa Barbara County, California, whereas the others were obtained from suppliers in central Texas. These animals were housed individually in a constant temperature colony room on 12 h dark: 12 h light days. Each was tested daily after approximately 22 h of food deprivation. The apparatus has been described in detail and illustrated elsewhere (Birch and Jacobs, 1979). Briefly, animals were tested in a three-choice discrimination apparatus. From a small test chamber the animals could look through three clear plexiglass panels toward three equidistant oscilloscope displays (Tektronix Model 360, P4 phosphors). At a testing distance of 30 cm each display subtended 11” of visual angle. A large (37’ X 37”) electroluminescent panel was mounted flush around each oscilloscope tube. These panels had the same space-averaged luminance as the faces of the oscilloscopes. They were illuminated throughout each test session and served to maintain a constant level of light adaptation. A response lever was mounted above each window. Below each was a food tray into which reinforcement was delivered (97 mg peanut-flavored Noyes pellets).
369
Luminance gratings having .a vertical orientation were generated (Jacobs et al., 1980) on any one of the three oscilloscopes, whereas the remaining two scopes were diffusely illuminated at the same average luminance level of 3.4 cd/m’. Contrast sensitivity functions were obtained with sinusoidallymodulated luminance gratings. For these measurements the contrast of the sinusoidal gratings was variable over a range of 70--0.1% through use of a series of fixed logarithmic attenuators. Resolution acuity was determined with square-wave luminance gratings at a constant contrast of 80%. In all instances, luminance contrast was measured by scanning the gratings with a narrow slit, while making continuous measurements of luminance with a Spectra UB brightness spot meter. The test apparatus was placed in an isolated, darkened room. The experiments were conducted with the aid of a laboratory computer. The computer selected the oscilloscope to receive the grating pattern, set the grating contrast level, delivered reinforcement as appropriate, and recorded the animals’ responses. The testing procedures have also been described previously (Jacobs et al., 1980). In brief, squirrels were trained to press the response lever corresponding to the oscilloscope containing the grating pattern. A test trial involved the presentation of a grating pattern on one of the three oscilloscopes. The trial was signalled to the subject by the occurrence of a cueing tone. The test trial terminated when the animal responded, or after 5 s without a response. During the 10 s intertrial interval, all of the oscilloscopes and the surrounding electroluminescent panels were illuminated uniformly at the same luminance level. A noncorrection procedure was employed; that is, the stimulus conditions on each trial were selected irrespective of whether the animal had responded correctly on the previous trial. Only a single spatial frequency, presented at multiple contrast levels, was tested in each session. These squirrels typically completed 150-250 test trials in each test session. Three different measurements of spatial vision in tree squirrels were made. First, to measure spatial contrast sensitivity, a number of different spatial frequencies, typically eight, between 0.11 and 2.9 c/d were tested, each at multiple contrast levels. At each spatial frequency a minimum of 100 test trials were accumulated at each contrast level across test sessions. From the resulting psychometric functions, threshold contast values (corresponding to 40% correct discrimination, P < 0.05) were determined for each frequency. Second, to measure acuity for square wave gratings, the animals were first trained to discriminate a low-frequency grating. Subsequently, the spatial frequency of the test grating was gradually increased, usually in steps of not more than 0.2 c/d until the animal’s performance declined to 40% correct. The spatial frequency required to produce this performance level was defined as a measure of squire-wave grating acuity. Third, in order to test resolution acuity at a much higher mean luminance level the oscilloscope displays were replaced by three light sources that were used to transilluminate photographic transparencies of square-wave grating patterns (contrast = 97%). These gratings had spatial frequencies ranging from 0.58 to 4.2 c/d.
370
The animal was initially trained to discriminate the coarsest grating from equiluminant, diffusely-illuminated panels. The spatial frequency of the test grating then was increased until performance dropped to threshold (40% correct). A minimum of 100 test trials were accumulated at each frequency tested. RESULTS
By the end of a lengthy preliminary training period all of the tree squirrels were producing reliable discrimination data. For instance, the average variation in discrimination performance for one western gray squirrel computed for 20 blocks (of 10 test trials each) was 7%. Other animals gave similar results. The psychometric functions generated from these data were qualitatively similar to those previously obtained from ground squirrels (Jacobs et al., 1980). Complete contrast sensitivity functions were measured for two fox squirrels and two western gray squirrels at an average luminance level of 3.4 cd/m’ and at a viewing distance of 30 cm. These functions are shown in Fig.1.
Spatial
Frequency
(c/d
1
Fig.1. Spatial contrast sensitivity functions for four tree squirrels. Left panel: data from two western gray squirrels; Right panel: data from two fox squirrels. The dashed portions of the curves represent freehand extrapolations of the functions. Viewing conditions: distance = 30 cm; target luminance = 3.4 cd/m2.
The contrast sensitivity values for the western gray squirrels (left panel, Fig.1) were very similar for all spatial frequencies. Peak sensitivity was at approximately 0.5 c/d with sensitivity gradually declining at the lower
frequencies, Extrapolation of the high frequency limb of this function yields cutoff values of 2.0 c/d for both subjects. The results for the two fox squirrels (right panel, Fig.1) show more individual variation with one animal having higher contrast sensitivity at all spatial frequencies. Despite the individual variation in contrast sensitivity for the fox squirrels, the salient features of the functions are similar to those obtained from the western gray squirrels. Specifically, peak sensitivity was again about 0.5 c/d with a clear decline in sensitivity to the lowest spatial frequencies tested. The extrapolated high frequency cutoff locations for the two fox squirrels were 1.8 and 2.9 c/d respectively. Seven measurements (from five animals, two of whom were tested at each of two target distances) of the ability of tree squirrels to resolve squarewave grating patterns having a mean luminance of 3.4 cd/m2 were obtained with methods described earlier. These results are summarized in Table I. Two aspects of these results bear emphasis. First, we found no evidence of any clear differences in spatial resolution among the three species of tree squirrels tested. Second, two fox squirrels were tested at each of two viewing distances, 30 and 50 cm, and their performances for these two conditions were not systematically different: an average high frequency threshold of 2.18 c/d at 30 cm and an average of 2.25 c/d at a target distance of 50 cm. Taken across all of these measurements, when the target had an average luminance of 3.4 cd/m2, the finest grating that the tree squirrels could resolve had an average spatial frequency of 2.2 c/d (corresponding to a bar separation of 13.6’). All of the measurements reported thus far were obtained at a target luminance of only 3.4 cd/m’, a limitation imposed by the oscilloscopes used to generate the test stimuli. Although this value is at least two log units higher
TABLE
I
Acuity measurements on three different species of tree squirrels. The results given are the finest square-wave grating pattern (luminance = 3.4 cd/m2) that could be resolved. Each threshold value represents an independent determination. Both the fox squirrels were tested at two target distances Subjects
High frequency Distance:
~_ Fox
squirrels
Western
gray squirrels
Eastern
gray squirrel
Mean resolution
30 cm
threshold
(c/d)
Distance:
2.5 1.8 2.0 1.8 2.8
1.9 2.6
2.18
2.25
50 cm
372
than the luminance required to reach cone threshold in tree squirrels (Jacobs and Birch, 1975) and may, therefore, be assumed to reflect the operation of photopic mechanisms, it is still low relative to the light levels these animals might be expected to encounter in their normal environments. To determine if spatial resolution changes drastically at higher luminance levels, the eastern gray squirrel was tested to determine its ability to discriminate high contrast square-wave gratings having average luminances of 3.4 and 340 cd/m*. As illustrated in Fig.2, at 3.4 cd/m’ this squirrel dropped to threshold performance at a spatial frequency of 2.8 c/d. When the target luminance was increased by two log units, to 340 cd/m*, spatial resolukon also increased, with the maximum resolution now at 3.9 c/d. c e--a
0.5
1.5 Spatial
2.5 Frequency
340cd/m2
3.5
_
4.5
(c/d]
Fig.2. Results from an experiment in which an eastern gray squirrel was tested to determine its ability to discriminate high contrast, square wave gratings at two different luminance levels. Each data point represents the mean for 100 test trials. The dashed line indicates the performance level taken as representing threshold performance.
DISCUSSION
The results of several measurements of visual acuity in tree squirrels reported here can be economically summarized as follows: (1) Tree squirrels are maximally sensitive to sinusoidal gratings (luminance = 3.4 cd/m*) with a spatial frequency of 0.5 c/d. (2) There is no evidence for systematic differences in visual acuity among three species of tree squirrels. (3) Eleven measurements of acuity in tree squirrels, obtained from tests involving the use of square-wave stimuli (n = 7) and from estimates of highfrequency cutoff locations in spatial contrast sensitivity curves at a target luminance of 3.4 cd/m* (n = 4), yield an average maximal resolution value of 2. c/d (SD = 0.42 c/d). (4) A comparison of the visual acuity of one squirrel tested at 3.4 and
313
340 cd/m2 suggests that tree squirrel visual acuity may improve somewhat at higher light levels, in our measurements reaching a maximum of 3.9 c/d. Many features of the visual system may serve to set limitations on spatial resolution. Hughes (1977) has suggested that the anatomical feature of the vertebrate eye that comes closest to predicting the maximal possible spatial resolution is ganglion cell density. He showed that a computation using Shannon’s sampling theorem based on the density of retinal ganglion cells predicts a resolution limit for the human visual system close to that actually measured in behavioral experiments. We have made a similar computation for the tree squirrel. Hughes (1977) lists a retinal magnification factor of 0.11 mm/d and a maximum gnaglion cell density of 25 OOO/mm’ for the “grey squirrel”. Using these estimates, the Shannon computation predicts a resolution limit for the ganglion cells of the tree squirrel to be 8.7 c/d. This value is more than one octave higher than any of our animals were able to achieve under any testing condition. Despite the close match between the results predicted by ganglion cell density and behavioral measurements of acuity in humans, it appears that the resolution that various nonhuman species can achieve is consistently lower than the ganglion cell prediction. For example, for the cat the prediction is 11 c/d (Hughes, 1977) and the behavioral acuity limit is about 6 c/d (Blake et al., 1974); for the pigmented rat a prediction of 2.3 c/d (Hughes, 1977) and a measured value of 1.2 c/d (Birch and Jacobs, 1979); for the galago a prediction of 8 c/d (DeBruyn et al., 1980) and a measured value of 5 c/d (Langston and Fox, 1981). Why each of these species, unlike the buman, fails to achieve as high a resolution value as that predicted by ganglion cell density cannot yet be specified, and likely may vary among species. One strong possibility is that these retinas contain several functionally independent populations of ganglion cells. If so, then use of overall ganglion cell density in the computation will lead to an overestimation of the limit of spatial resolution of that visual system (Hughes, 1981). Measurements of spatial resolution are heavily influenced by details of the testing situation, thus comparisons of our results on tree squirrels with results from measurements on other species should only be made with caution. However, three different rodent groups (rats, ground squirrels, and tree squirrels) have been tested in our laboratory under essentially identical test conditions and these results should be reasonably comparable. Table II summarizes the results of these studies and suggests three points. First, at a luminance level of 3.4 cd/m*, the acuities of these species are ordered in the same way as are the ratios of cones to rods in each animal. Second, the rodent species with primarily one receptor type (rats and ground squirrels) show no increase in visual acuity with increases in target luminance over a range from 3.4 to 340 cd/m2. Third, at high luminance levels arboreal and terrestrial squirrels have about the same abilities to resolve stationary luminance grating patterns. Finally, it should be emphasized that although these measurements set
374 TABLE
II
Results of measurements of visual acuity in four types of rodents tested at two luminance levels. The sources of these results were: rats (Birch and Jacobs, 1979); ground squirrels (Jacobs et al., 1980); tree squirrels (this study) Rodent
Albino
Maximum
rats
Pigmented rats Ground squirrels Tree squirrels
spatial
resolution
3.4 cd/m2
340 cd/m2
0.4 c/d
-
1.2 c/d 3.9 c/d 2.2 c/d
1.2 c/d 4.0 c/d 3.8 c/d
limits for some aspects of spatial resolution in tree squirrels, they were obtained under highly restricted viewing conditions, thus reflecting only one aspect of the abilities of these animals to make spatial discriminations in their natural environments. In our earlier experiments with ground squirrels we found that although their abilities to resolve stationary, luminancedefined patterns were limited, they were relatively quite good at discriminating among stimuli containing temporal changes (Jacobs et al., 1980). One might expect this to also be true for tree squirrels. We note in this regard that the gross electrical response recorded from the eye of the gray squirrel (S. griseus) has been shown to be able to resolve flickering lights up to at least 70 Hz (Yolton, 1975), and directionally-selective ganglion cells in the retina of S. carolinensis reportedly respond to moving grating patterns having spatial frequencies of up to 5 c/d (Cooper and Robson, 1966). ACKNOWLEDGMENTS
This research (EY 00105).
was supported
by a grant from the National
Eye Institute
REFERENCES Arden, G.B. and Silver, P.H., 1962. Visual thresholds and spectral sensitivity of the grey squirrel (Sciurus carolinensis leucotis). J. Physiol. 163: 540-557. Arden, G.B. and Tansley, K., 1955. The spectral sensitivity of the pure-cone retina of the grey squirrel (Sciurus carolinensis Zeucotis) J. Physiol., 127: 592-02. Birch, D., and Jacobs, G.H., 1979. Spatial contrast sensitivity in albino and pigmented rats. Vision Res., 19: 933-937. Blake, R., Cool, S.J. and Crawford, M.L.J., 1974. Visual resolution in the cat. Vision Res., 14: 1211-1217. Cohen, A.I., 1964. Some observations on the fine structure of the retinal receptors of the American grey squirrel. Invest. Ophthalmol., 3: 198-216. Cooper, G.F. and Robson, J.G., 1966. Directionally selective units in the retina of the grey squirrel. J. Physiol., 186: 116-117P.
375 Debruyn, E.J., Wise, V.L., and Casagrande, V.A., 1980. The size and topographic arrangement of retinal ganglion cells in the galago. Vision Res., 20 : 315-327. Dippner, R. and Armington, J., 1971. A behavioral measure of dark adaptation in the American red squirrel. Psychonom. Sci., 24: 43-45. Gouras, P., 1964. Duplex function in the grey squirrel’s electroretinogram. Nature (London), 203: 761-768. Green, D.G. and Dowling, J.E., 1975. Electrophysiological evidence for rod-like receptors in the grey squirrel, ground squirrel and prairie dog retinas. J. Comp. Neurol., 159: 461-472. Hughes, A., 1977. The topography of vision in mammals of contrasting life style: comparative optics and retinal organisation. In: F. Crescitelli (Editor). Handbook of Sensory Physiology, Vol. VII/5. Springer, Berlin, pp. 613-656. Hughes, A., 1981. Cat retina and the sampling theorem: the relation of transient and sustained brisk-unit cut-off frequency to 01 and P-mode cell density. Exp. Brain Res., 42: 196-202. Jacobs, G.H., 1974. Scotopic and photopic visual capacities of an arboreal squirrel (Sciurus niger). Brain Behav. Evol., 10: 307-321. Jacobs, G.H. and Birch, D., 1975. Increment-threshold functions for different rodent species. Vision Res., 15: 375-373. Jacobs, G.H., Blakeslee, B., McCourt, M.E. and Tootell, R.H., 1980. Visual sensitivity of ground squirrels to spatial and temporal luminance variations. J. Comp. Physiol., 136: 291-299. Langston, A. and Fox, R., 1981. Spatial contrast sensitivity of the galago. Invest. Ophthalmol. Vis. Sci. (Supplement), 20: 49. Silver, P.H., 1966. A Purkinje shift in the spectral sensitivity of grey squirrels. J. Physiol., 186: 439-450. Walls, G.L., 1942. The vertebrate eye. Cranbrook Institute of Science, Bloomfield Hills, We2,Ik.W. and Dowling, J.E., 1975. Anatomical evidence for cone and rod-like receptors in the grey squirrel, ground squirrel, and prairie dog. J. Comp. Neurol., 159: 439-460. Yolton, R.L., 1975. The visual system of the western gray squirrel: anatomical, electroretinographic and behavioral studies. Doctoral Dissertation, University of Texas, Austin.
VISUAL ACUITY SQUIRRELS
GERALD
AND SPATIAL
H. JACOBS,
Department
367 Amsterdam
DAVID
of Psychology,
Printed
CONTRAST
G. BIRCH*
University
-
SENSITIVITY
and BARBARA
of California,
in The Netherlands
BLAKESLEE
Santa Barbara,
*Present address: Berman-Gund Laboratory for the Study of Retinal Massachusetts Eye and Ear Infirmary, Boston, MA 02114 (U.S.A.) (Accepted
8 June
IN TREE
CA 93106
(U.S.A.)
Degenerations,
1982)
ABSTRACT Jacobs, G.H., sensitivity
Birch, D.G. and Blakeslee, B., 1982. Visual acuity in tree squirrels. Behav. Processes, 7 : 367-37 5.
and spatial
contrast
The visual sensitivity of tree squirrels from three different species (two western gray squirrels, Sciurus griseus; three fox squirrels, Sciurus niger; and an eastern gray squirrel, Sciurus carolinensis) was measured for spatial patterns defined by luminance differences. Spatial contrast sensitivity functions were determined in behavioral discrimination experiments in which the stimuli were sinusoidally-modulated gratings. At an average luminance level of 3.4 cd/m2 these squirrels were maximally sensitive to spatial frequencies of about 0.5 cycles/degree (c/d). These experiments, in conjunction with additional measurements involving discrimination of square-wave grating patterns, also indicate that the finest gratings these squirrels can discriminate average 2.2 c/d (SD = 0.42 c/d). There are no obvious differences in resolution acuity among the three species. Grating acuity also was measured at a much higher luminance level (340 cd/ml). The results of this experiment indicate that the spatial acuity of the tree squirrel increases with luminance level, reaching a maximum of 3.9 c/d at 340 cd/m2.
INTRODUCTION
Tree squirrels are diurnal, arboreal rodents. Their retinas are unusual among those of most mammals in that they contain more cones than rods; only about 40% of all tree-squirrel photoreceptors can be classified as rods (Cohen, 1964; West and Dowling, 1975; Yolton, 1975). It has been verified in both electrophysiological (Arden and Tansley, 1955; Gouras, 1964; Green and Dowling, 1975) and behavioral (Arden and Silver, 1962; Silver, 1966; Dippner and Armington, 1971; Jacobs, 1974) experiments that these two classes of photoreceptors provide the structural basis for functionally separable scotopic and photopic capacities. In addition to having a unique mixture of rods and cones, tree squirrel retinas are also notable for their lack of great regional variations. Thus, al-
03766357/82/0000-0000/$02.75
0 1982
Elsevier
Scientific
Publishing
Company
368
though these retinas show some central retinal specializations in the form of a visual streak and an area centralis, ganglion cell density is only modestly variant, with perhaps no more than a three- or four-fold decrease in density from the center of the retina to the periphery (Hughes, 1977). Although Walls (1942) believed that this so-called “universal macularity”, seen in the squirrels and other strongly diurnal species, should be associated with high spatial resolution, there is clearly nothing inevitable about the relationship. For instance, we recently found that ground squirrels, rodents also with cone-rich retinas and with modest regional variations in ganglion cell density, have relatively poor visual acuity. At the limit these animals could not resolve luminance grating targets having spatial frequencies of greater than 4 cycles/degree (c/d) (Jacobs et al., 1980). In general, spatial resolution depends not only on ganglion cell density and the retinal distribution of ganglion cells, but also on optical factors, the grain of the photoreceptor mosaic, and the resolving power of the neural visual system (Hughes, 1977). However, knowledge of these factors alone is not sufficient to permit a specification of the visual resolution capacities of the animal. For that, behavioral measurements are required. Accordingly, to determine what the spatial resolution capacities are for tree squirrels, we measured the abilities of representatives from three species to discriminate luminance gratings to obtain estimates of their visual acuity and contrast sensitivity. MATERIALS
AND METHODS
Measurements were made on individuals from three common species of North American tree squirrels: two (one male, one female) western gray squirrels (Sciurus griseus), three (one male, two females) fox squirrels (S. niger), and one female eastern gray squirrel (S. carolinensis). The western gray squirrels were trapped in Santa Barbara County, California, whereas the others were obtained from suppliers in central Texas. These animals were housed individually in a constant temperature colony room on 12 h dark: 12 h light days. Each was tested daily after approximately 22 h of food deprivation. The apparatus has been described in detail and illustrated elsewhere (Birch and Jacobs, 1979). Briefly, animals were tested in a three-choice discrimination apparatus. From a small test chamber the animals could look through three clear plexiglass panels toward three equidistant oscilloscope displays (Tektronix Model 360, P4 phosphors). At a testing distance of 30 cm each display subtended 11” of visual angle. A large (37’ X 37”) electroluminescent panel was mounted flush around each oscilloscope tube. These panels had the same space-averaged luminance as the faces of the oscilloscopes. They were illuminated throughout each test session and served to maintain a constant level of light adaptation. A response lever was mounted above each window. Below each was a food tray into which reinforcement was delivered (97 mg peanut-flavored Noyes pellets).
369
Luminance gratings having .a vertical orientation were generated (Jacobs et al., 1980) on any one of the three oscilloscopes, whereas the remaining two scopes were diffusely illuminated at the same average luminance level of 3.4 cd/m’. Contrast sensitivity functions were obtained with sinusoidallymodulated luminance gratings. For these measurements the contrast of the sinusoidal gratings was variable over a range of 70--0.1% through use of a series of fixed logarithmic attenuators. Resolution acuity was determined with square-wave luminance gratings at a constant contrast of 80%. In all instances, luminance contrast was measured by scanning the gratings with a narrow slit, while making continuous measurements of luminance with a Spectra UB brightness spot meter. The test apparatus was placed in an isolated, darkened room. The experiments were conducted with the aid of a laboratory computer. The computer selected the oscilloscope to receive the grating pattern, set the grating contrast level, delivered reinforcement as appropriate, and recorded the animals’ responses. The testing procedures have also been described previously (Jacobs et al., 1980). In brief, squirrels were trained to press the response lever corresponding to the oscilloscope containing the grating pattern. A test trial involved the presentation of a grating pattern on one of the three oscilloscopes. The trial was signalled to the subject by the occurrence of a cueing tone. The test trial terminated when the animal responded, or after 5 s without a response. During the 10 s intertrial interval, all of the oscilloscopes and the surrounding electroluminescent panels were illuminated uniformly at the same luminance level. A noncorrection procedure was employed; that is, the stimulus conditions on each trial were selected irrespective of whether the animal had responded correctly on the previous trial. Only a single spatial frequency, presented at multiple contrast levels, was tested in each session. These squirrels typically completed 150-250 test trials in each test session. Three different measurements of spatial vision in tree squirrels were made. First, to measure spatial contrast sensitivity, a number of different spatial frequencies, typically eight, between 0.11 and 2.9 c/d were tested, each at multiple contrast levels. At each spatial frequency a minimum of 100 test trials were accumulated at each contrast level across test sessions. From the resulting psychometric functions, threshold contast values (corresponding to 40% correct discrimination, P < 0.05) were determined for each frequency. Second, to measure acuity for square wave gratings, the animals were first trained to discriminate a low-frequency grating. Subsequently, the spatial frequency of the test grating was gradually increased, usually in steps of not more than 0.2 c/d until the animal’s performance declined to 40% correct. The spatial frequency required to produce this performance level was defined as a measure of squire-wave grating acuity. Third, in order to test resolution acuity at a much higher mean luminance level the oscilloscope displays were replaced by three light sources that were used to transilluminate photographic transparencies of square-wave grating patterns (contrast = 97%). These gratings had spatial frequencies ranging from 0.58 to 4.2 c/d.
370
The animal was initially trained to discriminate the coarsest grating from equiluminant, diffusely-illuminated panels. The spatial frequency of the test grating then was increased until performance dropped to threshold (40% correct). A minimum of 100 test trials were accumulated at each frequency tested. RESULTS
By the end of a lengthy preliminary training period all of the tree squirrels were producing reliable discrimination data. For instance, the average variation in discrimination performance for one western gray squirrel computed for 20 blocks (of 10 test trials each) was 7%. Other animals gave similar results. The psychometric functions generated from these data were qualitatively similar to those previously obtained from ground squirrels (Jacobs et al., 1980). Complete contrast sensitivity functions were measured for two fox squirrels and two western gray squirrels at an average luminance level of 3.4 cd/m’ and at a viewing distance of 30 cm. These functions are shown in Fig.1.
Spatial
Frequency
(c/d
1
Fig.1. Spatial contrast sensitivity functions for four tree squirrels. Left panel: data from two western gray squirrels; Right panel: data from two fox squirrels. The dashed portions of the curves represent freehand extrapolations of the functions. Viewing conditions: distance = 30 cm; target luminance = 3.4 cd/m2.
The contrast sensitivity values for the western gray squirrels (left panel, Fig.1) were very similar for all spatial frequencies. Peak sensitivity was at approximately 0.5 c/d with sensitivity gradually declining at the lower
frequencies, Extrapolation of the high frequency limb of this function yields cutoff values of 2.0 c/d for both subjects. The results for the two fox squirrels (right panel, Fig.1) show more individual variation with one animal having higher contrast sensitivity at all spatial frequencies. Despite the individual variation in contrast sensitivity for the fox squirrels, the salient features of the functions are similar to those obtained from the western gray squirrels. Specifically, peak sensitivity was again about 0.5 c/d with a clear decline in sensitivity to the lowest spatial frequencies tested. The extrapolated high frequency cutoff locations for the two fox squirrels were 1.8 and 2.9 c/d respectively. Seven measurements (from five animals, two of whom were tested at each of two target distances) of the ability of tree squirrels to resolve squarewave grating patterns having a mean luminance of 3.4 cd/m2 were obtained with methods described earlier. These results are summarized in Table I. Two aspects of these results bear emphasis. First, we found no evidence of any clear differences in spatial resolution among the three species of tree squirrels tested. Second, two fox squirrels were tested at each of two viewing distances, 30 and 50 cm, and their performances for these two conditions were not systematically different: an average high frequency threshold of 2.18 c/d at 30 cm and an average of 2.25 c/d at a target distance of 50 cm. Taken across all of these measurements, when the target had an average luminance of 3.4 cd/m2, the finest grating that the tree squirrels could resolve had an average spatial frequency of 2.2 c/d (corresponding to a bar separation of 13.6’). All of the measurements reported thus far were obtained at a target luminance of only 3.4 cd/m’, a limitation imposed by the oscilloscopes used to generate the test stimuli. Although this value is at least two log units higher
TABLE
I
Acuity measurements on three different species of tree squirrels. The results given are the finest square-wave grating pattern (luminance = 3.4 cd/m2) that could be resolved. Each threshold value represents an independent determination. Both the fox squirrels were tested at two target distances Subjects
High frequency Distance:
~_ Fox
squirrels
Western
gray squirrels
Eastern
gray squirrel
Mean resolution
30 cm
threshold
(c/d)
Distance:
2.5 1.8 2.0 1.8 2.8
1.9 2.6
2.18
2.25
50 cm
372
than the luminance required to reach cone threshold in tree squirrels (Jacobs and Birch, 1975) and may, therefore, be assumed to reflect the operation of photopic mechanisms, it is still low relative to the light levels these animals might be expected to encounter in their normal environments. To determine if spatial resolution changes drastically at higher luminance levels, the eastern gray squirrel was tested to determine its ability to discriminate high contrast square-wave gratings having average luminances of 3.4 and 340 cd/m*. As illustrated in Fig.2, at 3.4 cd/m’ this squirrel dropped to threshold performance at a spatial frequency of 2.8 c/d. When the target luminance was increased by two log units, to 340 cd/m*, spatial resolukon also increased, with the maximum resolution now at 3.9 c/d. c e--a
0.5
1.5 Spatial
2.5 Frequency
340cd/m2
3.5
_
4.5
(c/d]
Fig.2. Results from an experiment in which an eastern gray squirrel was tested to determine its ability to discriminate high contrast, square wave gratings at two different luminance levels. Each data point represents the mean for 100 test trials. The dashed line indicates the performance level taken as representing threshold performance.
DISCUSSION
The results of several measurements of visual acuity in tree squirrels reported here can be economically summarized as follows: (1) Tree squirrels are maximally sensitive to sinusoidal gratings (luminance = 3.4 cd/m*) with a spatial frequency of 0.5 c/d. (2) There is no evidence for systematic differences in visual acuity among three species of tree squirrels. (3) Eleven measurements of acuity in tree squirrels, obtained from tests involving the use of square-wave stimuli (n = 7) and from estimates of highfrequency cutoff locations in spatial contrast sensitivity curves at a target luminance of 3.4 cd/m* (n = 4), yield an average maximal resolution value of 2. c/d (SD = 0.42 c/d). (4) A comparison of the visual acuity of one squirrel tested at 3.4 and
313
340 cd/m2 suggests that tree squirrel visual acuity may improve somewhat at higher light levels, in our measurements reaching a maximum of 3.9 c/d. Many features of the visual system may serve to set limitations on spatial resolution. Hughes (1977) has suggested that the anatomical feature of the vertebrate eye that comes closest to predicting the maximal possible spatial resolution is ganglion cell density. He showed that a computation using Shannon’s sampling theorem based on the density of retinal ganglion cells predicts a resolution limit for the human visual system close to that actually measured in behavioral experiments. We have made a similar computation for the tree squirrel. Hughes (1977) lists a retinal magnification factor of 0.11 mm/d and a maximum gnaglion cell density of 25 OOO/mm’ for the “grey squirrel”. Using these estimates, the Shannon computation predicts a resolution limit for the ganglion cells of the tree squirrel to be 8.7 c/d. This value is more than one octave higher than any of our animals were able to achieve under any testing condition. Despite the close match between the results predicted by ganglion cell density and behavioral measurements of acuity in humans, it appears that the resolution that various nonhuman species can achieve is consistently lower than the ganglion cell prediction. For example, for the cat the prediction is 11 c/d (Hughes, 1977) and the behavioral acuity limit is about 6 c/d (Blake et al., 1974); for the pigmented rat a prediction of 2.3 c/d (Hughes, 1977) and a measured value of 1.2 c/d (Birch and Jacobs, 1979); for the galago a prediction of 8 c/d (DeBruyn et al., 1980) and a measured value of 5 c/d (Langston and Fox, 1981). Why each of these species, unlike the buman, fails to achieve as high a resolution value as that predicted by ganglion cell density cannot yet be specified, and likely may vary among species. One strong possibility is that these retinas contain several functionally independent populations of ganglion cells. If so, then use of overall ganglion cell density in the computation will lead to an overestimation of the limit of spatial resolution of that visual system (Hughes, 1981). Measurements of spatial resolution are heavily influenced by details of the testing situation, thus comparisons of our results on tree squirrels with results from measurements on other species should only be made with caution. However, three different rodent groups (rats, ground squirrels, and tree squirrels) have been tested in our laboratory under essentially identical test conditions and these results should be reasonably comparable. Table II summarizes the results of these studies and suggests three points. First, at a luminance level of 3.4 cd/m*, the acuities of these species are ordered in the same way as are the ratios of cones to rods in each animal. Second, the rodent species with primarily one receptor type (rats and ground squirrels) show no increase in visual acuity with increases in target luminance over a range from 3.4 to 340 cd/m2. Third, at high luminance levels arboreal and terrestrial squirrels have about the same abilities to resolve stationary luminance grating patterns. Finally, it should be emphasized that although these measurements set
374 TABLE
II
Results of measurements of visual acuity in four types of rodents tested at two luminance levels. The sources of these results were: rats (Birch and Jacobs, 1979); ground squirrels (Jacobs et al., 1980); tree squirrels (this study) Rodent
Albino
Maximum
rats
Pigmented rats Ground squirrels Tree squirrels
spatial
resolution
3.4 cd/m2
340 cd/m2
0.4 c/d
-
1.2 c/d 3.9 c/d 2.2 c/d
1.2 c/d 4.0 c/d 3.8 c/d
limits for some aspects of spatial resolution in tree squirrels, they were obtained under highly restricted viewing conditions, thus reflecting only one aspect of the abilities of these animals to make spatial discriminations in their natural environments. In our earlier experiments with ground squirrels we found that although their abilities to resolve stationary, luminancedefined patterns were limited, they were relatively quite good at discriminating among stimuli containing temporal changes (Jacobs et al., 1980). One might expect this to also be true for tree squirrels. We note in this regard that the gross electrical response recorded from the eye of the gray squirrel (S. griseus) has been shown to be able to resolve flickering lights up to at least 70 Hz (Yolton, 1975), and directionally-selective ganglion cells in the retina of S. carolinensis reportedly respond to moving grating patterns having spatial frequencies of up to 5 c/d (Cooper and Robson, 1966). ACKNOWLEDGMENTS
This research (EY 00105).
was supported
by a grant from the National
Eye Institute
REFERENCES Arden, G.B. and Silver, P.H., 1962. Visual thresholds and spectral sensitivity of the grey squirrel (Sciurus carolinensis leucotis). J. Physiol. 163: 540-557. Arden, G.B. and Tansley, K., 1955. The spectral sensitivity of the pure-cone retina of the grey squirrel (Sciurus carolinensis Zeucotis) J. Physiol., 127: 592-02. Birch, D., and Jacobs, G.H., 1979. Spatial contrast sensitivity in albino and pigmented rats. Vision Res., 19: 933-937. Blake, R., Cool, S.J. and Crawford, M.L.J., 1974. Visual resolution in the cat. Vision Res., 14: 1211-1217. Cohen, A.I., 1964. Some observations on the fine structure of the retinal receptors of the American grey squirrel. Invest. Ophthalmol., 3: 198-216. Cooper, G.F. and Robson, J.G., 1966. Directionally selective units in the retina of the grey squirrel. J. Physiol., 186: 116-117P.
375 Debruyn, E.J., Wise, V.L., and Casagrande, V.A., 1980. The size and topographic arrangement of retinal ganglion cells in the galago. Vision Res., 20 : 315-327. Dippner, R. and Armington, J., 1971. A behavioral measure of dark adaptation in the American red squirrel. Psychonom. Sci., 24: 43-45. Gouras, P., 1964. Duplex function in the grey squirrel’s electroretinogram. Nature (London), 203: 761-768. Green, D.G. and Dowling, J.E., 1975. Electrophysiological evidence for rod-like receptors in the grey squirrel, ground squirrel and prairie dog retinas. J. Comp. Neurol., 159: 461-472. Hughes, A., 1977. The topography of vision in mammals of contrasting life style: comparative optics and retinal organisation. In: F. Crescitelli (Editor). Handbook of Sensory Physiology, Vol. VII/5. Springer, Berlin, pp. 613-656. Hughes, A., 1981. Cat retina and the sampling theorem: the relation of transient and sustained brisk-unit cut-off frequency to 01 and P-mode cell density. Exp. Brain Res., 42: 196-202. Jacobs, G.H., 1974. Scotopic and photopic visual capacities of an arboreal squirrel (Sciurus niger). Brain Behav. Evol., 10: 307-321. Jacobs, G.H. and Birch, D., 1975. Increment-threshold functions for different rodent species. Vision Res., 15: 375-373. Jacobs, G.H., Blakeslee, B., McCourt, M.E. and Tootell, R.H., 1980. Visual sensitivity of ground squirrels to spatial and temporal luminance variations. J. Comp. Physiol., 136: 291-299. Langston, A. and Fox, R., 1981. Spatial contrast sensitivity of the galago. Invest. Ophthalmol. Vis. Sci. (Supplement), 20: 49. Silver, P.H., 1966. A Purkinje shift in the spectral sensitivity of grey squirrels. J. Physiol., 186: 439-450. Walls, G.L., 1942. The vertebrate eye. Cranbrook Institute of Science, Bloomfield Hills, We2,Ik.W. and Dowling, J.E., 1975. Anatomical evidence for cone and rod-like receptors in the grey squirrel, ground squirrel, and prairie dog. J. Comp. Neurol., 159: 439-460. Yolton, R.L., 1975. The visual system of the western gray squirrel: anatomical, electroretinographic and behavioral studies. Doctoral Dissertation, University of Texas, Austin.
