Featured
Research
Featured
ResearchColor Vision of BirdsFrancisco J. Varela, Adrian G. Palacios, and Timothy H. Goldsmith
from "Vision, brain, and behavior in birds". Eds. Zeigler and Bischof. 1993. Cambridge, MA: MIT Press.
The role of color vision in an animal's perception, behavior, and ecological setting, and its underlying retina and neuronal mechanisms vary enormously in different groups of animals. Amidst this diversity, birds have arguably the most elaborate and interesting color vision. The evidence for this assertion comes from various sources. (1) Foremost are existing knowledge about different types of cells and pigments in the retina and (2) behavioral experiments demonstrating various chromatic abilities. Less extensive evidence can be drawn from (3) physiological analyses of neural mechanisms and (4) occasional ecological observations. A potential sources of evidence, only now becoming available for birds, is (5) identification and analysis of genes for the proteins (opsins) of visual pigments (Okano et al., 1992). Information from all these levels needs to be brought together for a detailed understanding of the color vision in any animal group. This chapter summarizes for birds the existing knowledge in the first four of these categories of evidence. Supplementary information of a broader comparative nature can be found elsewhere (Goldsmith, 1990; Neumeyer, 1991; Thompson et al., 1992).
RETINAL BASES
The cones of birds (see Nalbach et al., chapter 2) present more morphological complexity and diversity than those of mammals. First, the inner segment characteristically contains a colored oil droplet adjacent to the base of the Outer segment, forming a filter through which light must pass before reaching the visual pigment. The oil droplets are of several colors, due to the presence of different carotenoids. Second, in addition to single cones, there are also double cones, consisting of two closely contiguous cells known as the principal and accessory members of the pair.
Cone Oil Droplets
Knowledge of cone spectral sensitivity is critical to understanding how any color vision functions. The spectral sensitivity of a bird cone, however, is determined jointly by the product of the spectral transmittance of the oil droplet and the spectral absorptance (i.e., l-transmittance) of the visual pigment. The most direct method of determining the transmittance spectrum of individual oil droplets would appear to be microspectrophotometry, but this measurement involves a serious technical problem. The concentrations of carotenoids, particularly in the deeper-colored droplets, are very high, with peak absorbances (i.e., optical densities) greater than 10 or even 20. In other words, only a minute fraction of the incident light is transmitted by the droplet. The signal from the detector used to measure transmittance therefore becomes dominated by stray light, and an accurate measurement of the droplet's absorption spectrum is not possible. The seriousness of this problem is evident in published records of oil droplet spectra in which the apparent peak absorbance of the dense red and yellow droplets is only 0.7 (Bowmaker and Martin, 1985; Jane and Bowmaker, 1988). In some early work this problem was not recognized; in more recent work on birds it has been dealt with in two different ways.
The simplest approach is to assume that the droplets function as long-pass cut-off filters and to characterize them by the wavelength at 50% transmission and the slope of the transmittance curve in the cutoff region, which the experimenter tries to determine at wavelengths where the transmittance is still high enough to be measured accurately. From this information it is possible to calculate the effect of the droplet on the absorptance of the underlying visual pigment. This approach gives no information on the chemical identity of the absorber, a point of interest because differences in the spectral region of cut off can be due either to different carotenoids or to differences in concentration of the same carotenoid.
A more elaborate approach, first introduced by Liebman and Granda (1975) in a study of turtle oil droplets, is to expand individual cone oil droplets by fusing them with larger droplets of mineral oil. The resulting droplet is not only larger, it has a lower concentration of carotenoid, permitting a sufficiently accurate measurement of spectral absorbance to be useful in chemical identification (Goldsmith et al., 1984). In principle, no information is lost, because with knowledge of the diameters of the original and final droplets it is possible to calculate the transmittance as it was in vivo. An important finding from this approach is that one class of droplets commonly present in many orders of birds contains a mixture of two carotenoids whose proportions frequently vary in different cells in the same retina (table 5.1, P droplets). What appears superficially to be a homogeneous class of droplets is therefore actually heterogeneous. The implications for the spectral properties of afferent chromatic channels are not yet known.
Unfortunately no common agreement has yet emerged on a terminology to characterize avian oil droplets, due mainly to variation (for example, in the aforementioned P droplets), to uncertainties in relating results obtained by the two methods in different laboratories, and to uncertainties whether new measurements from sparsely studied species fit an existing scheme of classification. Table 5.1 tries to relate the measurements and classification of Goldsmith et al. (1984), who reduced the concentration of carotenoids by expanding individual droplets (first column), to the results of Bowmaker (1977), Bowmaker and Knowles (1977), and Partridge (1989), who measured the cut-off properties of undiluted droplets (last columns). As suggested by the interrogation marks (?) in the last two columns, there is considerable uncertainty in bringing these systems into agreement. Confusion abounds, for recently Jane and Bowmaker (1988) have also loosely employed the terminology in column 1, with the result that in early papers this research group used "C" to refer to bright yellow droplets containing high concentrations of C40 xanthophyll, and later to the virtually colorless, galloxanthin-containing droplets.There are a number of additional issues that require clarification. First, orange droplets are not yet adequately defined; some may contain carotenoids not listed in table 5.1; others may simply hold a higher concentration of zeaxanthin (or lutein) or a lower concentration of astaxanthin. Second, microspectrophotometry in the near UV suggests subclasses of C and P droplets containing an apocarotenoid with at 375 nm (and thus with fewer double bonds than galloxanthin). As has now been noted by several workers, some sauropsid oil droplets fluoresce weakly (e.g., Kolb and Jones, 1987). In bird retinas the fluorescence is associated with droplets containing the apocarotenoid galloxanthin or the unknown carotenoid absorbing maximally about 375 nm (Goldsmith et al., 1984), but the source of the fluorescence has not been identified.
The functions of oil droplets remain a subject for speculation. Suggested functions, would are not mutually exclusive, include narrowing of the spectral sensitivity functions of the cones, with implications for avian color space (Lythgoe, 1979; Barlow, 1982; Govardovskii, 1983; Goldsmith, 1990), protecting the outer segment from short wavelength photo damage (Kirschfeld, 1982), improvement of contrast or relief from chromatic aberration (Muntz, 1972; Miller, 1979), and concentration of light on the outer segment (Baylor and Fettiplace, 1975; Ives et al., 1983; Young and Martin, 1984).
The distributions and proportions of the different colors of oil droplets show variations between species in ways that correlate with diet or other aspects of the ecology (Peiponen, 1964; Partridge, 1989). The dorsal retina tends to have more of the deeply colored red and yellow droplets, particularly in birds that look through an air-water interface (Muntz, 1972). (In pigeons the red and orange droplets are associated with a quadrant of the retina directed forward and down [the "red field"], but this is not characteristic of most other orders of birds.) Of the droplets containing galloxanthin (droplets P and C in the first column of table 5.1), those supplemented with an additional carotenoid (P) are frequently more abundant in the ventral retina, the ones with less color (C), the dorsal retina (Goldsmith et al., 1984). In general, birds with nocturnal or crepuscular habits have many fewer deeply pigmented droplets. See also Nalbach et al., this volume.
Visual Pigments of Cones
Light is trapped in the cones of the eye by pigments consisting of a protein (opsin) conjugated to a fat-soluble molecule called retinal, which is the aldehyde of vitamin A. Cone cells sensitive in different regions of the spectrum owe their different spectral absorption to the presence of different opsins. The retinas of birds characteristically express the genes for three or more cone opsins, each located in a distinct subset of cones. A major source of information about the spectral properties of these pigments has been the microspectrophotometric measurements of Bowmaker and his colleagues. Figure 5.1 summarizes what is known from direct measurements of pigments (filled symbols); in a number of instances the properties of the associated oil droplets have also been determined, and figure 5.1 also indicates the amount of spectral shift that these filters can impose on the effective absorption by the accompanying cone pigments (curved arrows and open symbols).
A wide selection of birds have at least three or four cone pigments, and as one pigment may be associated with more than one type of oil droplet (in different cells), the possible number of chromatic channels could, in principle, be larger than the number of cone pigments. In every species examined so far, the pigment absorbing at longest wavelengths (near 570 nm, roughly the spectral region that is occupied by the "red"- and "green" -sensitive pigments in the primate retina) is by far the most abundant. This is iodopsin, first extracted from chickens by Wald et al. (1955), and it appears to be the visual pigment that dominates the photopic spectral sensitivity of birds (Wright, 1979; Wortel et al., 1984; Remy and Emmerton, 1989). The measurements on penguins suggest that in this species the pigment has responded to natural selection for vision in a blue aquatic environment by shifting its absorption spectrum to shorter wavelengths, to an max at 543 nm (Bowmaker and Martin, 1985). Birds characteristically have more than one cone pigment absorbing at short wavelengths, but in general those pigments absorbing in the blue and violet are known from small samples of cones, in some cases as few as a single cell (but see also Fager and Fager, 1981).Physiological Measurements of Spectral Sensitivity in the Retina
By recording light-elicited, transretinal voltages (electroretinogram) in opened eyecups of a variety of birds, mostly passerines, Chen and Goldsmith (1986) were able to measure spectral sensitivity functions under different states of adaptation to colored lights. The conditions were photopic, and the presence of aspartate (used to block the synapses between receptors and retinal neurons) indicated that the spectral responses reflected the properties of receptors. As the light was incident from the vitreal side, the receptors were filtered by oil droplets as they would in vivo. In all species the dark-adapted retinas were maximally sensitive over a broad wavelength band around 580 nm, indicating the presence of the 570-nm cone pigment referred to above. When the retinas were adapted with increasing intensities of orange and yellow lights, the spectral sensitivity functions shifted to shorter wavelengths, finally leaving a distinct maximum in the near UV at 370 nm. Of the 15 species studied (ruby-throated hummingbird [Archilochus colubris], pigeon [Columba livia], and 13 passerines), all had sensitivity maxima at 370 and 570-580 nm. In the chickadee (Parus atricapillus), house sparrow (Passer domesticus), house finch (Carpodacus mexicanus), northern cardinal (Cardinalis cardinalis), and song sparrow (Melospiza melodia) (upper line) it was possible to isolate additional maxima at 450 and 480 nm by using adapting lights generated by mixing near UV with orange or yellow light. Other species (lower line) (American robin [Turdus migratorius], brown thrasher [Hylocichla mustelina], and barn swallow [Hirundo rustica] had a cone with max at 510 nm. In still other species there was only a single bird with which to work and not all the sensitivity maxima shown in Figure 5.2 were detected before the experiment ended. For the pattern shown in the lower row, there is probably a fourth cone operating in the blue-violet region of the spectrum.
BEHAVIORAL EVIDENCE
Although physiological data may indicate that an animal has color vision, the final judgment about its presence and use must be based on behavioral experiments and ecological observations. A particularly important question that can be settled only by behavioral experiments is whether the animal is able to distinguish between wavelength (color) and intensity (brightness) cues. Only if this question can be answered in the affirmative is it possible to conclude that the animal has color vision. Behavioral experiments can also be used to quantify the capacity for color vision. For example, in which spectral regions is the capacity to discriminate between monochromatic lights of similar wavelengths the keenest? What mixtures of colors are indistinguishable to an animal? Knowledge of these and similar behavioral capacities is critical in understanding how information about color is being handled by the nervous system.
Wavelength DiscriminationKnowing how the animal's ability to discriminate one wavelength from another varies in different spectral regions can be very informative. The results of such measurements are plotted as AX (the minimal separation between two wavelengths that can just be detected as different) vs. 1. The animal's ability to discriminate is measured by increasing AX from I until a criterion of (say) 70% correct responses is attained. Since for each wavelength one can obtain two values of AX (one to either side of 1), the final curve is usually plotted as the mean of the two. It is essential that wavelengths to be compared be adjusted for the animal's subjective brightness, which means that the experimenter must obtain the animal's spectral sensitivity function. As a consequence, these experiments are very time consuming, and only a handful of results are available.
Table 5.2 shows some results from the pigeon (Columba livia) (Hamilton and Coleman, 1933; Wright and Cumming, 1971; Blough, 1961; Riggs et al., 1972; Schneider, 1972; Wright, 1972; Jitsumori, 1978; Nuboer and Wortel, 1987; Palacios et al., 1990a). Each value of wavelength in table 5.2 corresponds to a minimum in the curve of AX vs. X and identifies a local spectral region of good wavelength discrimination. Figure 5.3a shows two wavelength discrimination functions (Emmerton and Delius, 1980; Palacios et al., 1990a), one of which extends into the near UV. Best discrimination was observed at 370, 460, 530, and 595 nm. Whether the appearance of minima at 500 nm (and at shorter wavelengths) is due to differences in experimental design (caused, for example, by the birds imaging the target on different parts of the retina) or differences between individual birds remains to be determined. Results for the hummingbird (Archilochus alexandri) do not reveal a deterioration in wavelength discrimination at short wavelengths, indicating the need to base general conclusions on more than one species (Goldsmith et al., 1981) (figure 5.3b).
Wavelength discrimination curves are interesting because they are an indication, however partial and incomplete, of the different chromatic processes that are actually operating in color vision. Antagonistic interactions between neural signals (opponent mechanisms) are responsible for minima in the wavelength discrimination curve. The inputs to these neural mechanisms, however, need not correspond to single types of cone. There are more minima in the wavelength discrimination functions of pigeons than of Old World primates, suggesting that avian color vision may be more elaborate than the trichromacy of humans.Color Mixing
In the color vision of Old World primates, three independent chromatic stimuli must be available in color mixtures to produce a visual match with any arbitrarily selected colored light. Our color vision is thus said to be trichromatic, and this is related to the fact that we have three different spectral classes of cone cells. Determining the dimensionality in a vector space (Wyszecki and Stiles, 1981). An example of such a plot appears in figure 5.5, and further explanation is provided below. Until recently, color mixing data for birds were conspicuously absent. An early successful color match was reported in the qualitative observations of Delius et al. (1972). These authors tested young herring and lesser black-backed gulls (Larus argentatus, L. fuscus), which indicate a desire to be fed by pecking spontaneously at a bright orange spot on the parent's beak. When the experimenter replaced this target with two pure spectral lights (536 and 620 nm) or a mixture of both, the young bird pecked preferentially at the mixture.
A quantitative study on pigeons has recently been reported (Palacios et al., 1990b; Palacios and Varela, 1992). The principal method was a modified autoshaping procedure, which takes advantage of the animal's spontaneous tendency to peck. After a short period of training, the birds pecked continuously whenever a positive stimulus was present (Palacios et al., 1990a). When the animal compared a monochromatic reference stimulus (S-) with an appropriate mixture of two spectral sources, a decrease from 90% correct responses to 50% (chance) indicated that the animal was unable to distinguish the mixture from the reference wavelength. These experiments establish that pigeons can make a variety of color matches throughout their visible spectrum. Examples of dichromatic matches made by four pigeons at the long wavelength end of the spectrum are shown in figure 5.4, where the abscissa shows the proportion of 640 and 580 nm light required for confusion with S+ = 600 nm.
The fundamental principle that is involved in color matches is that two physically different lights (i.e., 600 nm monochromatic light [orange] and an appropriate mixture of 640 and 580 nm lights [red and yellow] in the example of figure 5.4) are able to excite the population of retinal cones in an identical manner. It therefore becomes appropriate to compare the results with predictions derived from knowledge of the spectral properties of those cones that are likely to have been activated during the color matching experiments (Palacios et al., 1990b; Palacios and Varela, 1992). These predictions can be made quantitative by calculating the fraction of the incident light that is trapped by each spectral type of cone, where the properties of each spectral type are determined by the absorption spectrum of the visual pigment and the effect of the overlying oil droplet.As in our own eye, the color vision of the pigeon is dichromatic in a spectral region limited to long wavelengths, and the experimental data on pigeons are consistent with the presence of cone systems in the red field having maxima and 575 and 619 nm. For the middle-wavelength region, primary mechanisms with maxima at 485-nm for the red field and 525 nm for the yellow field appear to be present, along with two additional mechanisms in the violet at 415 nm and in the near UV at 350 or 360 nm. These experiments thus provide strong evidence for a pentachromatic nature to the pigeon's color space, but three and fourway color mixtures are still needed to establish this point more fully.
The Maxwell color triangle is one of the graphic conventions that can be used to convey information about color matches. In this convention, the proportions in which the three receptors in a trichromatic system are activated are plotted on a triangular coordinate system in which points at the three vertices of the triangle correspond to exclusive activation of each of three cone systems (Rushton, 1972). Figure 5.5 shows how this graphic convention can be expanded to a color tetrahedron to accommodate a tetrachromatic color space.
Color Constancy and Color ContrastIn addition to several independent cone systems, neural interactions are also important participants in color vision. The well-known and related phenomena of color constancy (the relative invariance of perceived hue with changes in illumination) and color contrast (the appearance of complementary colors adjacent to or following the presentation of a colored stimulus) are manifestations of lateral neural interactions and play important roles in color vision theory (Hurvich, 1981; Jameson and Hurvich, 1989; Hurlbert, 1986). In fact, the integrative processes underlying these psychophysical phenomena are universal participants in all color vision systems.
Color constancy has been shown quantitatively in the bee and qualitatively in the goldfish, and some preliminary evidence is available on color induction in pigeons (Budnik, 1985). In these experiments pigeons were trained to respond to a positive stimulus generated using broad band filters (Kodak Wratten) so that the animal effectively responded to a chromatic class. After the training phase, the animal was tested by replacing the positive stimulus with a neutral field surrounded by a ring of the complementary color. In many cases the animal responded to the test field as though it were colored like the original field. This effect was dependent on the area of the annular surround and the distance between surround and the spectrally neutral center, as is to be expected from a lateral integrative effect in which the center took on a hue complementary to the surround. In fact, these geometrical parameters seem quite comparable in humans, the pigeon and the honeybee, as shown in figure 5.6. Needless to say, quantitative data on color contrast and color constancy are sorely needed for birds.
NEUROPHYSIOLOGYSurprisingly, a survey of the literature reveals very few experimental studies on the neural substrate of avian color vision. Birds and mammals use homologous structures in the brain in importantly different ways. For example, in birds the optic tectum rather than the lateral geniculate nucleus of the thalamus, is the major relay center for ascending visual information. It is therefore not possible to transfer knowledge about areas of the mammalian brain that are involved in color vision to an understanding of the color vision of birds, and, conversely, other regions of the avian brain are involved in color vision than those that play a role in mammals. In view of the different evolutionary histories of avian and mammalian brains, comparative data should be very rewarding.
Optic Nerve
Donner (1953) was the first to study the responses of optic nerve fibers in the pigeon. More recently Marin (1983; see Varela et al., 1983) recorded from more than 200 fibers in the optic tract of the quail. Of these, 18% were tonic fibers exhibiting sustained responses that were readily modulated by changing the wavelength (adjusted for equal brightness) of the stimulus. The spectral preferences of these ganglion cells were clearly at short wavelengths; 42% responded best between 410 and 450 nm. (No UV stimuli were used in this study.) All of these cells showed an "off " response to lights of complementary color, which, if presented in color mixtures, could also inhibit the tonic response. The receptive fields of both chromatic actions were spatially coextensive; these cells are therefore color-opponent units without a center/surround organization.
Optic Tectum
A careful study on pigeons using substitution of chromatic stimuli of the same brightness demonstrated a small number of tectal units that responded when one color was changed to another (Jassik-Gerschenfeld et al., 1977). Instead of using static stimuli, Letelier (1983) studied the responses of 82 tectal cells to a moving bar of monochromatic light projected on a dark background. Using such stimuli, 30% of the cells responded; the spectral sensitivity functions were narrow, and most cells were maximally sensitive toward short wavelengths. When a monochromatic slit or edge was moved against its complementary wavelength of equal luminosity, the activity of the cell did not change, suggesting that these units do not have antagonist chromatic inputs. This inference may be wrong, however, since for many tectal cells the responses to leading and trailing edges varied differently with wavelength, even though the total spike count would have led one to classify these cells as broad-band luminosity units. When studied on a finer time scale, a moving bar against a chromatic background does modulate these responses, since the ratio of the size of the leading edge/trailing edge response varies with hue. We have tentatively named this behavior temporal-opponency (Varela et al., 1983). Whether this is a significant and novel feature of the neural basis of avian color vision remains to be confirmed by future experiments. On the other hand, it supports the view that the tectum is an important locus of information processing in avian color vision.
Diencephalon
In the avian homologue of the lateral geniculate nucleus of mammals, the complex designated as the nucleus opticus principalis thalami, color responses have not been detected yet (Maxwell and Granda, 1979). A few units sensitive to color have been reported in the nucleus rotundus, the thalamic relay nucleus between the optic tectum and the ectostriatum of the cortex (tectofugal pathway) (Yazulla and Granda, 1973).
Most thalamic color-sensitive units have been found in the ventral lateral geniculate nucleus (GLv). This structure is virtually nonexistent in mammals; in birds it projects reciprocally and topographically to the optic tectum as well as receiving ascending input from the retina and descending input from the Wulst, the homologue of the mammalian visual cortex (Crossland and Uchwat, 1979; Guiloff et al., 1987). Many neurons (48% of 156 units studied) in the GLv are color opponent (Maturana and Varela, 1982). Like units in the optic tract, these cells respond tonically, their antagonistic receptive fields are similar, and sensitivity to blue and violet predominates. These cells are insensitive to luminance changes over 2 log units. The anatomical location of the ventral thalamic nucleus (bridging the tecto and thalamofugal pathways) and its small size make it hard to interpret its role in color vision. In a recent study pigeons were trained to make color discriminations before and after bilateral chemical lesions of the GLv (Palacios et al., 1991). After a transient phase, the discrimination thresholds were not altered, so at least for this task, the GLv does not seem to be essential. In contrast, even a small amount of damage to the nucleus rotundus immediately above the GLv induces a deficit in discrimination (Hodos, 1969; Palacios et al., 1991).
ECOLOGICAL CONSIDERATIONS
Although we still lack extensive knowledge of the ecological role of color vision in most animal species, the available evidence demonstrates that speculations about color vision should not be driven by simplified, top-down computational models that are based on the evolutionarily unique features of Old World primates, specifically humans. Instead, as the following examples will illustrate, color vision must be understood within the context of the behavioral repertoires of diverse animals.
The colors and distributions of retinal oil droplets of birds vary considerably among species (Martin, 1977, 1986; Jane and Bowmaker, 1988; Martin and Lett, 1985; Budnik et al., 1984). For example, the common tern, a fish-eating predator, has a large number of red and yellow droplets in the dorsal retina, while the barn swallow, which catches insects, has few deeply colored droplets (Peiponen, 1964; Goldsmith et al., 1984). Partridge (1989) has shown by means of cluster analysis that the ecological niche (herbivore, fishing, etc.) is more important in predicting the kinds and distribution of oil droplets than phylogenetic relationships. The evidence suggests that oil droplets respond to natural selection faster than opsins.
The presence of ultraviolet pigments in birds may also provide an example of variation with ecological niche, but too little is yet known of the distribution of UV receptors to draw firm conclusions. For example, the presence of UV visual pigments has been linked to bird-fruit coevolution, including the dissemination of seeds (Snow, 1971; Burkhardt, 1982), and to ethological factors involving animal recognition (Weedon, 1963; Durrer, 1986), but these hypotheses are propelled by the tradition in comparative physiology of seeking particular adaptive explanations for all phenomena. The reflectances of bird plumages have been shown to have shorter wavelength content than reflectances of natural objects of interest to humans and other primates (Burkhardt, 1989; Hudon and Brush, 1989; Brush, 1990), and this suggests the possibility of a perceptual color space of higher dimensionality than three (Barlow, 1982; Bonnardel and Varela, 1991). Evolutionary arguments supported by both comparative physiology and molecular genetics indicate the trichromacy of primate color vision is recently derived, and it is only in avian vision that it is possible to find the quintessence of diurnal capability, evolutionarily uncompromised by a long history of nocturnality as encountered in the mammalian line (Goldsmith, 1990). Seen this way, the question is not "Why do many birds see UV?" but rather "Why is it that most mammals do not?"
The question of how UV sensitivity is utilized by birds is not resolved. UV sensitivity is exploited, for if it were not, natural selection would have abandoned it, as almost certainly occurred in most of the evolutionary line that led to mammals. But in discussing birds, we remain in the realm of speculation. For example, Nuboer's (1986, pp. 370-371) words call attention to the possible use of UV sensitivity in aerial navigation as well as also illustrating the traditional appeal to special adaptation: "the excellent spectral discrimination within this range . . . represents an adaptation to the coloration of an unclouded sky. This property enables the pigeon to evaluate short-wave gradients in the sky, ranging from white at the sun's locus to highly saturated (ultra) violet at angles of 90o to the axis between observer and sun." Furthermore, since pigeon navigation is based on orientation with respect to the sun's azimuth, "the perception of colour gradients in the sky may control navigation indirectly when the sun is hidden by clouds." All of this may be true, but one could also generate adaptationist hypotheses having to do with feather or fruit reflectance, no one of which is likely to have sufficient generality to account for the wide distribution of UV sensitivity in birds.
A different but perhaps more important perspective on the ecological role of color vision is that the process yields a set of perceptual categories that have functional significance, even "cognitive significance" (Jacobs, 1981, pp. 170-171) for animals that must deal with a variety of behavioral interactions and ecological circumstances. A color category can guide behavior in various ways depending on the things that exemplify it: in the case of fruits, it guides feeding; in the case of animal coloration, it may guide various social interactions, such as mating; and these roles are not necessarily mutually exclusive. Pigeons have been shown to group spectral stimuli into categories of hue, and the brightly colored feathers of birds, particularly those exhibiting sexual dimorphism, must have cognitive significance for behavior, especially behavior involving sexual recognition. Finally, although discrimination of objects is obviously important for these kinds of behaviors, the cognitive significance of color may have an affective dimension (perhaps related to the overall hormonal/motivational level of the animal) that cannot be explained simply in terms of discrimination of objects.
CONCLUSIONS
Much research remains to be done on the relations among color vision, perceptual color categories, and animal behavior (Hailman, 1977; Burtt, 1979). Although color as a perceptual category with cognitive significance obviously plays a great role in human life, there is, with the exception of hymenopterous insects (Menzel and Backhaus, 1991), still little evidence about the roles of color perception in nonhuman animals, especially nonprimates. In the case of birds, however, it seems safe to conclude that the experience of color does exist. Moreover, the evidence that we have presented serves to demonstrate our point that the operation of color vision must ultimately be understood at all the levels of analysis to which we referred in the introduction, from molecular events in cells to the behavioral repertoires of animals seen in the ecological conditions in which their kind evolved.
In no species of bird is there a clear understanding of the number and properties of the pigment-oil droplet combinations and their distributions in the retina, providing a detailed description of the receptor basis for that animal's color vision. Multiple foveae and other specialized regions that are suggested by nonuniform distributions of oil droplets (such as the red quadrant of the pigeon's retina) indicate a degree of complexity that is not present in the human eye and for which our own sensory experience provides little intuitive understanding. Birds may have a generalized system of color vision, but individual species may also have features of their eyes adapted to specific visual tasks or conditions, and attention to this ecological dimension in formulating hypotheses about visual function is likely to be critical.
As in primates, pigments absorbing at relatively long wavelengths are the major contributors to photopic sensitivity, but at least some birds probably discriminate short wavelengths in the violet region of the spectrum rather better than primates. The presence of a receptor process in the near UV, long thought to be the exclusive province of insects and foreign to our own visual experience, now seems to be commonly present in many nonmammalian vertebrates (e.g., Kreithen and Eisner, 1978; Goldsmith, 1980; Avery et al., 1983; Harosi and Hashimoto, 1983; Arnold and Neumeyer, 1987) and has recently also been found in rodents (Jacobs et al., 1991). Birds can detect near UV, and it may have a chromatic quality of its own; the possible role of UV in avian color vision needs to be explored.
The information that is available on the receptor substate and on color abilities such as discrimination and color matches shows that at least four chromatic channels are likely to be very commonly present in the retinas of birds. This makes birds true tetrachromats, and perhaps the only pentachromats in the animal kingdom.
The evolutionary history and radiation of vertebrate color vision present us with enormous diversity, and the mammalian or even primate perspective is a narrow pedestal from which to view and understand this evolutionary scene. The tetra- or pentachromatic color space of birds appears to be the most complex in nature and is likely involved in virtually all areas of the animals' lives, from the discrimination and recognition of objects to more complex behavioral tasks such as navigation, the classification of objects, and social and sexual behavior. The next years should see considerable progress in understanding these various evolutionary, physiological, behavioral, and ecological factors and in solving the mysteries of this appealing facet of natural history.
ACKNOWLEDGMENTS
F.V. acknowledges with gratitude the support of the Prince Trust Fund, and C.N.R.S. (Unite Associee 1199); A.P., the Simone & Cino del Duca and Philippe Foundations; and T.H.G., National Eye Institute (NIH) Grants EY-00222 and EY-00785.
REFERENCES
Arnold, K., and Neumeyer, C. Wavelength discrimination in the turtle. Pseudemys scripta elegans. Vision Res. 27 (1987),1501-1511.
Avery, J. A., Bowmaker, J. K., Djamgoz, M. B. A., and Downing, J. E. G. Ultraviolet sensitive receptors in a freshwater fish. 1. Physiol. 334 (1983), 23-24.
Barlow, H. B. What causes trichromacy? A theoretical analysis using comb-filtered spectra. Vision Res. 22 (1982), 635-643.
Baylor, D. A., and Fettiplace, R. Light path and photon capture in turtle photoreceptors. J. Physiol. 248 (1975), 433-464.
Blough, D. S. The shape of some wavelength generalization gradients. Percept. Psychophys. 4 (1961), 31-40.
Bonnardel, V., and Varela, J. F. A frequency view of colour: Measuring the human sensitivity to square-wave spectral power distributions. Proc. Roy. Soc. Ser. B 245 (1991), 165-171.
Bowmaker, J. K. The visual pigments oil droplets and spectral sensitivity of the pigeon. Vision Res. 17 (1977), 1129-1138.
Bowmaker, J. K., and Knowles, A. The visual pigments and oil droplets of the chicken retina. Vision Res. 17 (1977), 755-764.
Bowmaker, J. K., and Martin, G. R. Visual pigments and colour vision in a nocturnal bird, strix aluco (tawn owl). Vision Res. 18 (1978), 1125-1130.
Bowmaker, J. K., and Martin, G. R. Visual pigments and oil droplets in the penguin, Spheniscus humbolti. J. Comp. Physiol. A 156 (1985), 71-77.
Brush, A. H. Metabolism of carotenoid pigments in birds. FASEB 4 (1990), 2969-2977.
Budnik, V. "Un estudio comparativo de la induccion cromatica." Tesis de Licenciatura, Facultad de Ciencias, Universidad de Chile, Santiago, Chile, 1985.
Budnik, V., Mpodozis, J., Varela, J., and Maturana, H. R. Regional specialization of the quail retina: Ganglion cell density and oil droplet distribution. Neurosci. Lett. 51 (1984), 145-150.
Burkhardt, D. Birds, berries and UV: A note on some consequences of UV vision in birds. Naturwissenschaften 69 (1982), 153-157.
Burkhardt, D. UV vision: a bird's eye view of feathers. J. Comp. Physiol. A 164 (1989), 787-796.
Burtt, E. H. The Behavioral Significance of Color. New York, Garland STPM Press, 1979.
Chen, D. M., and Goldsmith, T. H. Four spectral classes of cone in the retinas of birds. Comp. Physiol. A 159 (1986), 473-479.
Crossland, W. J., and Uchwat, C. J. Topographic projections of the retina and optic tectum upon the ventral lateral geniculate nucleus in the chick. J. Comp. Neurol. 185 (1979), 87-106.
Delius, J. D., Thompson, G., Allen, K. L., and Emmerton, J. Colour mixing and preferences in neonate gulls. Experientia 28 (1972), 1244-1246.
Donner, K. 0. The spectral sensitivity of the pigeon's retinal elements. J. Physiol. 122 (1953), 524-537.
Durrer, H. Colouration. In J. Bereiter-Hahn, A. G. Matoltsy, and K. S. Richards (Eds.), Biology of the Integument, the Skin of Birds. Springer, Berlin, 1986, Vol. 2, pp. 239-247.
Emmerton, J., and Delius, J. D. Wavelength discrimination in the 'visible' and ultraviolet spectrum by pigeons. J. Comp. Physiol. A 141 (1980), 47-52.
Fager, L. Y., and Fager, R. S. Chicken blue and chicken violet, short wavelength sensitive visual pigments. Vision Res. 21 (1981), 581-586.
Goldsmith, T. H. Hummingbirds near ultraviolet light. Science 207 (1980), 786-788.
Goldsmith, T. H. Optimization, constraint, and history in the evolution of eyes. Q. Rev. Biol. 65 (1990), 281-322.
Goldsmith, T. H., Collins, J. S., and Perlman, D. L. A wavelength discrimination function for the hummingbird Archilochus alexandri, J. Comp Physiol. A 143 (1981), 103-110.
Goldsmith, T. H., Collins, J. S., and Licht, S. The cone oil droplets of avian retinas. Vision Res. (1984), 1661-1671.
Govardovskii, V. 1. On the role of oil drops in colour vision. Vision Res. 23 (1983), 17391740.
Govardovskii, V. I., and Zueva, L. V. Visual pigments of chicken and pigeon. Vision Res. 17 (1977), 537-543.
Guiloff, G. D., Maturana, R. H., and Varela, F. J. Cytoarchitecture of the avian ventral lateral geniculate nucleus. J. Comp Neurol. 264 (1987), 509-526.
Hailman, J. P. Optical Signals: Animal Communication and Light. Bloomington and London, Indiana University Press, 1977.
Hamilton, W. F., and Coleman, T. B. Trichromatic vision in the pigeon as illustrated by the spectral hue discrimination curve. J. Comp. Psychol. 15 (1933), 183-191.
Harosi, F. I., and Hashimoto, Y. Ultraviolet visual pigment in a vertebrate: A tetrachromatic cone system in the dace. Science 223, (1983) 1021-1023.
Hodos, W. Color discrimination deficits after lesions of the nucleus rotundus in pigeons. Brain Behav. Evol. 2 (1969), 185-200.
Hudon, J., and Brush, A. H. Probable dietary basis of a color variant of the cedar waxing. J. Field Ornithol. 60 (1989), 361-368.
Hurlbert, A. Formal connections between lightness algorithms. J. Opt. Soc. Am. A 3 (1986), 1684-1693.
Hurvich, L. M. Color Vision. Sinauer Associates, Sunderland, MA, 1981.
Ives, J. T., Normann, R. A., and Barber, P. W. Light intensification by cone oil droplets: electromagnetic considerations. J. Opt. Soc. Am. A 73 (1983), 1725-1731.
Jacobs, G. H. Comparative Color Vision. Academic Press, New York, 1981.
Jacobs, G. H., Neitz, J., and Deegan, J. F. Retinal receptors in rodents maximally sensitive to ultraviolet light. Nature (London) 253 (1991), 655-656.
Jameson, D., and Hurvich, L. Essay concerning color constancy. Annu. Rev. Psychol. 40 (1989), 1-22.
Jane, S. D., and Bowmaker, J. K. Tetrachromatic colour vision in the duck (Alias platyrhynchos L.): Microspectrophotometry of visual pigments and oil droplets. 1. Comp. Physiol. A 62 (1988), 225-235.
Jassik-Gerschenfeld, D., Lange, R. L., and Ropert, N. Response of movement detecting cells in the optic tectum of pigeons to change of wavelength. Vision Res. 17 (1977), 11391146.
Jitsumori, M. Wavelength discrimination function derived from post-discrimination gradients in the pigeon. Jpn. Psychol. Res. 20 (1978), 18-28.
Kirschfeld, K. Carotenoid pigments: their possible role in protecting against photooxidation in eyes and photoreceptor cells. Proc, Roy. Soc. London B 216 (1982), 71-85.
Kolb, H., and Jones, J. The distinction by light and electron microscopy of two types of cone containing colorless oil droplets in the retina of the turtle. Vision Res. 27 (1987), 1445-1458.
Kreithen, M., and Eisner, T. Ultraviolet light detection by the homing pigeon. Nature (London) 272 (1978), 347-348.
Letelier, J. "Respuestas cromaticas en el tectum optico de la paloma." Tesis de Licenciatura, Facultad de Ciencias, Universidad de Chile, Santiago, Chile, 1983.
Liebman, P. A., and Granda, A. M. Super dense carotenoid spectra resolved in single oil droplets. Nature (London) 253 (1975), 370-372.
Lythgoe, J. N. The Ecology of Vision. Clarendon Press, Oxford, 1979.
Marin, G. "Respuestas cromaticas en el tracto optica de la codorniz." Tesis de Licenciatura, Facultad de Ciencias, Universidad de Chile, Santiago, Chile, 1983:
Martin, G. R. Absolute visual threshold and scotopic spectral sensitivity in the tawny owl, Strix aluc. Nature (London) 268 (1977), 636-638.
Martin, G. R. The eye of a passiform bird, the european starling (Sturnus vulgaris): Eye movement amplitude, visual fields and schematic optics. J. Comp. Physiol. A 159 (1986), 545-557.
Martin, G. R., and Lett, B. T. Formation of associations of coloured and flavoured food with induced sickness in five avian species. Behav. Neural. Biol. 43 (1985), 223-237.
Maturana, H. R., and Varela, F. J. Color-opponent responses in the avian lateral geniculate: a study in the quail (Coturnix coturnix japonica). Brain Res. 247 (1982), 227-241.
Maxwell, J. H., and Granda, A. M. Receptive fields of movement-sensitive cells in the pigeon thalamus. In A. M. Granda and J. H. Maxwell (Eds.), Neural Mechanism of Behavior in the Pigeon. Plenum Press, New York, 1979.
Menzel, R., and Backhaus, W. Colour vision in insects. In J. Cronly-Dillon (Ed.), Vision and Visual Dysfunction. Macmillan Press, London, 1991, Vol. 6, P. Gouras (Ed.), pp. 262-293.
Miller, W. H. Ocular optical filtering. In H. Autrum (Ed.), Handbook of Sensory Physiology, VII, Part 6A, Vision in Invertebrates. Springer-Verlag, Berlin, 1979, pp. 69-143.
Muntz, W. R. Inert absorbing and reflecting pigments. In H. J. A. Dartnall (Ed.), Handbook of Sensory Physiology, VII/I: Photochemistry of Vision. Springer-Verlag, Berlin, 1972, pp. 530-565.
Neumeyer C. Simultaneous color contrast in the honeybee. J. Comp. Physiol. A 139 (1980), 163-176.
Neumeyer, C. Evolution of colour vision. In J. Cronly-Dillon (Ed.), Vision and Visual Dysfunction. Macmillan, London, 1991, Vol. 2, J. R. Cronly-Dillon and R. L. Gregory (Eds.), pp. 284-305.
Nuboer, J. F. A comparative view on color vision. Netherlands J. Zool. 36 (1986), 344-380.
Nuboer, K., and Wortel, J. F. Colour vision via the pigeon's red and yellow retinal fields. In Kulikowski, C. M. Dickinson, and I. J. Murray (Eds.), Seeing Contour and Colour. Pergamon Press, Oxford, 1987.
Okano, T., Kojima, D., Fukada, Y., Shichicla, Y., and Yoshizawa, T. Primary structures of chicken cone visual pigments: vertebrate rhodopsins have evolved out of cone visual pigments. Proc. Natl. Acad. Sci. U. S. A. 89 (1992), 5932-5936.
Palacios, A. G., and Varela, F. J. Color mixing in the pigeon (columbia livia) 11. A psychophysical determination in the middle, short and near-UV wavelength range. Vision Res. 32 (1992), 1947-1953.
Palacios, A., Bonnardel, V., and Varela, F. An autoshaping method for wavelength discrimination on birds. C. R. Acad. Sci. Paris 311(111) (1990a), 213-218.
Palacios, A., Martinoya, C., Bloch, S., and Varela, F. Color mixing in the pigeon. A psychophysical determination in the longwave spectral range. Vision Res. 30 (1990b), 587-596.
Palacios, A., Gioanni, H., and Varela, F. Chromatic discrimination in pigeons after thalamic lesions of nuclei Rotundus (Rt) and Geniculatus Lateralis ventralis (GLv): A psychophysical study. C. R. Acad. Sci. Paris 312(111) (1991), 113-116.
Partridge, J. C. The visual ecology of avian cone oil droplets. J. Comp. Physiol. A 165 (1989), 415-426.
Peiponen, V. A. Zur Bedeutung der Olkugeln im Farbensehen der Sauropsiden. Ann. Zool. Fenn. 1 (1964), 281-302.
Remy, M., and Emmerton, J. Behavioral spectral sensitivities of different retinal areas ill pigeons. Behav. Neurosci. 103 (1989), 170-177.
Riggs, L. A., Blough, P. M., and Schafer, K. L. Electrical responses of the pigeon eye to changes in wavelength of the stimulating light. Vision Res. 12 (1972), 981-991.
Rushton, W. A. H. Pigments and signals in colour vision. J. Physiol. 220 (1972), 1-31.
Schneider, B. Multidimensional scaling of colour differences in the pigeon. Percept. Psychophys. 12 (1972), 373-378.
Snow, D. W. Evolutionary aspects of fruit eating by birds. Ibis 113 (1971), 194-202.
Thompson, E., Palacios, A. G., and Varela, F. J. Ways of coloring: Comparative color vision as a case study for cognitive science. Behav. Brain Sci. 15 (1992), 1-75.
Varela, F. J., Letelier, J. C., Marin, G., and Maturana, H. R. The neuphysiology of avian color vision. Arch. Biol. Med. Exp. 16 (1983), 291-303.
Wald, G., Brown, P. K., and Smith, P. H. Idopsin J. Gen. Physiol. 38 (1955), 623-681.
All material is copyrighted by the respective
authors or publishers. Contact for reprint permission should be made
directly to the publisher or author. In citing these works or extracts
thereof, citation should be made to the original publication, and not the
website URL.
INDEX
Copyright 1999 by Patrick Thrush