I walk slowly toward the yard without saying a word. A dog peers out. I am spotted. A cautious bark alerts the other dogs, and now the yard is filled with 22 eyes staring straight at me. I continue to approach, still silent. The dogs sniff the air, crane their necks, stare, and begin to pace and bark nervously. It is apparent they’re not sure who I am. How close must I get before they recognize me, their breeder, the person who shares their daily life and puts the food bowls in front of them? As it turns out, very close. I finally speak, and the relief is palpable as they swamp me with enthusiastic greetings. The scary part is, these are salukis---sighthounds!
Sighthounds are the group of greyhound-like dogs that have been selectively bred for thousands of years for their ability to chase swift game by sight. They are one of several groups of dogs that rely extensively on vision in order to perform the tasks for which they were bred. Retrievers depend upon visually tracking falling birds to mark their place; herders depend upon detecting slight movements of the stock as well as their master; police and military dogs make extensive use of their visual sense in carrying out their duties. Perhaps most important of all, guide dogs must act as the eyes for their visually impaired handlers. Dogs may be famous for their sense of smell, but for at least some dogs, good vision is every bit as essential.
Selective breeding has achieved the unsurpassed diversity in physical and behavioral characteristics that is the hallmark of the domestic dog. If selection can achieve such remarkable results, can it also act on sensory abilities? Do sighthounds and other breeds in which vision is so vital to function have superior vision? And if so, can we further select for visual excellence to produce, for example, guide dogs with the best eyes possible? Before these questions can be addressed, the visual capabilities of dogs in general must be investigated.
The visual sense, like the eye itself, is made up of a number of related components. The most basic function is the ability to perceive light, but this ability has been fine-tuned in many ways, so that slight differences between intensities or wavelengths of light can be perceived. No eye can do it all, and in every species evolution has acted to fine tune the abilities essential to that animal’s lifestyle. The dog is no exception. The human and canine eye are built upon the same basic design, but each has modifications that enable it to perform optimally according to that species’ lifestyle. Humans evolved as a diurnal species, from omnivorous (but often fruit-eating) tree-dwelling primate ancestors. Dogs evolved as a nocturnal or crepuscular (active at dawn and dusk) species, from omnivorous (but usually meat-eating) cursorial (running) ancestors. We would expect the different evolutionary pressures to create visual systems with different specialties. Human eyes do not see well in the dark, but have great acuity, color perception and depth perception. The capabilities of canine eyes are less well documented, but they are clearly very different from those of humans.
The eye is often compared to a camera, and in many ways this is an apt description. Like the camera, it has an aperture (pupil in the eye), lens (cornea and lens in the eye), and receptive surface or film (retina in the eye). Like the camera, these features can be adjusted or modified to cope with different lighting conditions. Like the camera, the eye continually makes compromises between sensitivity at low levels of light and sensitivity to fine detail.
Anybody who has walked at night with a dog can attest to the dog’s apparent well-developed night vision. Could my dogs’ equally apparent inability to recognize me be the price they pay for an increased ability to see in the dark? Just how well can dogs see fine details? Many factors affect an animal’s acuity, including pupil size, optics of the eye, and retinal design.
Light enters the pupil of the eye, which is the aperture controlled by the iris. The wider the pupil or aperture, the more light can enter, which is essential in dim settings. Large pupils are characteristic of animals that tend to be active in dim light. Notice how much larger your dog’s pupils are than your own. But there is a trade-off: with a larger aperture, the depth of field (or distance over which objects can be put into clear focus) decreases. Thus, in order to achieve focus over a large range the pupil must be constricted, which means the environment must be bright. After passing through the pupil, the light passes through the lens. Camera lenses are rated for their light gathering ability; more expensive lenses gather more light and can be used with smaller apertures, thus combating the depth of field loss otherwise inherent in dim light. The same is true for eyes. Larger lenses have greater light gathering ability and are usually found in animals active in dim light. Dogs have lenses that are much larger than human lenses. Actually, unlike the camera, eyes have two lenses, because the outer clear surface of the eye, the cornea, acts like a strong lens as well. An animal with a large pupil must have a concomitantly large cornea, and larger corneas are usually found in animals requiring good night vision. Notice how much larger your dog’s corneas are than yours.
Besides gathering light, a lens bends light rays as they pass through it. This ability to bend, or “refract”, light is an essential feature of a lens. The more a lens can bend light, the more powerful the lens is said to be. In the ideal eye or camera, the power of the lens would be such that the entire scene would be perfectly focussed upon the light sensitive surface (either the retina or film). This ideal state can only be achieved with a pinhole aperture, however, so that the lens must be fine-tuned in order to bring objects at different distances into focus. In the camera this fine-tuning is achieved by moving the lens back and forth. In the eye, fine-tuning is achieved by changing the curvature of the lens, a process known as accommodation.
In humans, this accommodative ability decreases with age because the lens gradually hardens and the muscles that control the lens shape gradually weaken. The result is increasing difficulty in focussing on objects at close range. In a sense, dogs don’t have this aging problem---but only because they may be essentially born with the accommodative ability of a person 40 to 60 years of age!
When optimally focussed, light forms a sharp image exactly at the plane of the light sensitive receptors. In the camera, this would be the film surface, In the eye, it is the retina, located at the rear of the globe. If the refractive powers of the cornea and lens are too powerful for the distance to the retina, the light will come to a focus before it ever reaches the retina, and will go on to become unfocussed by the time it falls on the retina. This condition, called nearsightedness or myopia, results in difficulty bringing distant objects into clear focus. If the cornea and lens have too little refractive power for the distance to the retina, the light will still be unfocussed when it falls on the retina. This condition, called farsightedess or hyperopia, results in difficulty bringing close objects into clear focus. Only when the eye’s refractive power is in perfect accordance with its retinal distance will light rays be brought into sharp focus on the retina. This is the desirable refractive state known as emmetropia.
It’s not difficult to estimate the refractive state in a cooperative dog by use of a retinoscope, an instrument for observing the focal point of a beam of light that has passed through all of the refracting surfaces of the eye. In the hands of a skilled practitioner, retinoscopy yields results very close to those achieved by more extensive testing used in optical exams for people (the familiar “A vs. B” choices). The results of early research seemed to suggest that most dogs should be wearing glasses. As far back as 1901, researchers reported that domestic dogs were myopic, but that wild-caught canids (wolves, jackals and dingos) were emmetropic or slightly hyperopic1. Deviations from emmetropia are measured in diopters (D), the power of a lens necessary to bring the image into focus. Powers of plus or minus 1 D would indicate mild degrees of hyperopia or myopia, respectively; powers of plus or minus 3 D would indicate that strong glasses would be necessary! These early researchers reported an average of -3 D, indicating strong myopia 2. In the 1920s, a more extensive study of over 100 dogs found great variability among dogs, ranging from –4.5 to +2.0 D 3. Yet modern studies of refractive states in dogs have suggested that most dogs are within 0.5 D of emmetropia4,5.
In a recent study of 240 dogs, most dogs were found to be nearly emmetropic6. These researchers compared the results from breeds in which they examined five or more representatives (German Shepherd Dog, Chesapeake Bay Retriever, Cocker Spaniel, Golden Retriever, Labrador Retriever, Poodle, Rottweiler, Miniature Schnauzer, Chinese Shar Pei, Springer Spaniel, Terriers as a group, and mixed-breeds as a group). Over half of the German Shepherds, Rottweilers, and Schnauzers were myopic, significantly more than found in the other groups. Rottweilers were the most severely affected, with myopic Rottweilers averaging almost -3 D. Interestingly, in all of these breeds the myopia tended to occur within the same families. A group of German Shepherd guide dogs had significantly lower prevalence of myopia (34%) compared to non-guide German Shepherds. The retriever breeds tended to be more hyperopic than the other breeds in the study, but the degree of hyperopia was not great (averages from +0.4 to +0.8 D depending on breed). These results, especially when combined with an earlier (but unsubstantiated) report that greyhounds were usually from +0.5 to +1.5 D hyperopic7, are consistent with the idea that refractive state may have a hereditary component.
Besides providing some tantalizing evidence that breeds differences may exist in refractive states, this (and previous) studies found a greater tendency toward myopia with increasing age in dogs. The myopia was especially apparent in much older dogs. In the only reported longitudinal study of canine refractive states, six-month-old Beagles averaged +0.4D, and two years later the same dogs averaged –0.5D 8. Understanding the refractive states of dogs, and especially older dogs, has led to some practical implications. This is because many dogs develop cataracts or other lens problems that necessitate removing the lens so that vision can be saved or restored. Removing the lens obviously removes a major optical component of the eye. When this operation was first performed on dogs, it was customary to simply remove the lens without trying to restore the eye’s pre-operative refractive state. Yet without a lens, dogs are terribly hyperopic, averaging about +14 D. With the advent of intraocular prosthetic lenses, it was hoped that dogs could be restored to near emmetropic refractive states. Initial attempts with implants produced dogs that were still severely hyperopic, however, because the lenses were calculated at the strength necessary to achieve emmetropia had they been in front of the eye, as spectacle lenses are. Placement of the lens is an integral part of determining the refractive state of any optical system (in this case, the eye), and an intraocular lens would need a very different power than a contact lens or spectacle lens. Optical models of the canine eye suggested that a prosthetic lens would have to be much stronger than the lenses initially tried; stronger, in fact, than the intraocular prosthetic lenses used for humans 9,10 . This reflects the larger lens of the dog, as well as its more rearward placement in the dog’s eye, compared to the human. In fact, empiric trials found the best average implant lens strength was +41.5D 11. Although the refractive states weren’t always perfect for every dog, they were closer than previous results; interestingly, no effect of breed or body size was found concerning how well the approximation worked. Now dogs can not only profit from having cataracts removed, but can expect sharp vision following lens replacement with the appropriate intraocular prosthesis.
In the emmetropic dog, the optics of the eye bring the image into perfect focus onto the retina. Now the anatomy of the retina imposes the next limiting factor in perceiving fine details. Returning to our camera analogy, the eye’s retina is like the camera’s film. Any photographer knows that film comes in different sizes and speeds. Anyone who has tried to enlarge a photo from tiny 110 camera film knows how poor the end result is. The film is simply too small to record fine details. For best results, a large area of film needs to be covered so that there is plenty of room for details to be recorded. This is why professional photographers use expensive large-format cameras. The same is true for eyes. The light sensitive receptor cells are about the same size in all mammals, whether they are in elephants or mice. Obviously, any more can be packed onto a larger retina, and the size of the image on the retina can be greater if that retina is big.
Over 100 years ago, "Leukart’s Law" postulated that swifter species have larger eyes. This assertion appears to be true generally for mammals, with swifter species occupying more open environments where both speed and good distance acuity can be utilized. In dogs it is easy to confuse actual eye, or globe, size with the size of the opening between the lids (called the palpebral fissure). As long as this fissure is large enough that the pupil is not obscured when fully dilated, the size and the shape of the opening, as opposed to the globe, is primarily of concern for aesthetic and health reasons, rather than visual acuity. In fact, the size of the globe varies between breeds far less than would outwardly appear; despite efforts of many toy dog breeders to produce smaller, more proportional eyes, there is apparently some physiological limit beyond which it is difficult to reduce eye size in dogs. For reasons of acuity, this is probably a good thing.
Besides absolute size, the retina and film both depend upon another feature to ensure good acuity. Film captures images because it is coated with an emulsion containing silver grains that undergo a chemical reaction when exposed to light. In very dim light, the chances of a silver grain being hit by sufficient light to cause a reaction can be increased simply by making the silver grain larger. The result is film that is very sensitive in low light levels, but that creates a "grainy" image lacking fine detail. In bright light, it’s better to select a film coated with tiny grains of silver, which can create an image of exquisite detail.
Animals can’t select different film or retinal speeds according to lighting conditions, but they have evolved several ways of coping. Like film, receptors contain chemicals that react when exposed to light. One way is to use both large and small "grains", or receptor types, in the same retina. The two types of specialized receptors are the rods and cones. The rods are analogous to the large grains; not only is each rod very sensitive to light, but the responses from groups of rods are pooled and analyzed by higher level processing cells. This response pooling increases the area over which light is caught (in essence, creating a larger "grain"), thus increasing sensitivity at the expense of acuity. In contrast, the cones are like the small silver grains; they won’t detect very dim light but if the light is sufficiently bright, their fine mosaic can result in the ability to discriminate fine details.
Rods predominate in the retinas of nocturnal animals, and cones predominate in the retinas of diurnal animals. Most mammals have both. If the cones were evenly distributed among the rods, the increased acuity due to their fine mosaic would be somewhat negated. To combat this, the rods and cones are unevenly distributed, with more rods toward the periphery and more cones toward the center of the retina. In some animals in which fine vision is especially critical (such as humans) a small pure cone area, called the fovea, is placed directly in the line of vision. In fact, the fovea is the part of your eye you are using to read this text. It covers only a very small area of your vision, however; try reading with your finger blocking your central vision and notice how difficult it is. This can give you some idea of how it must be like for an animal without a fovea to make out fine details.
Do dogs have a fovea? In 1902 a researcher claimed to have found a fovea in sighthounds, but not in other breeds of dog 12. A more extensive investigation using 50 Greyhounds in the 1950s found no such area, however 13. It is now generally agreed that no dogs have an area of pure-cones, although they do have an area of increased cone density toward the center of the retina. Some recent evidence has revived the possibility that breed differences may exist in cone distribution. This evidence comes from examinations not of cones, but of retinal ganglion cells.
Signals from the rods and cones reach the brain by means of intermediate processing cells known as ganglion cells. The important thing to know about these cells is that their density roughly mirrors both the number of cones and the resulting acuity in different parts of the retina, and they are easier to count than either rods or cones. Most species have an oval area in the middle of the retina in which ganglion cell density is greatest. Other species have a horizontal streak across the retina in which ganglion cell density is greatest. The dog has both: a central oval area is superimposed upon a horizontal streak, although there is great variation in the extent to which either predominates. Species with more highly developed streaks tend to be fast animals that live on the open plains, such as gazelles and horses. The streak is believed to aid animals in scanning the horizon for predators. Of the carnivores so far investigated, the cheetah and the Greyhound have been reported to have the most highly developed streaks, with less developed streaks in the dingo, fox, and other breeds of dog 14.
A 1992 study compared ganglion cell distribution in wolves, German Shepherd Dogs, and Beagles, with some interesting results 15. Perhaps not unexpectedly, wolves had higher overall ganglion cell densities than domestic dogs. They also had more pronounced streaks than German Shepherds and some Beagles. But unexpectedly, dogs from one family of Beagles had more pronounced streaks and higher ganglion cell densities than those from another family of Beagles, suggesting the possibility of a genetic component. If such large differences in number and distribution exists between these dogs, it would be interesting to examine other breeds that have been selected for good vision. Although this evidence suggests that individual differences might also exist in visual acuity among dogs, surprisingly little is known about visual acuity in dogs in general
The evidence so far indicates that even if dogs are emmetropic, the lower cone and ganglion cell density compared to humans suggest rather poor acuity. A more precise estimate of acuity can be calculated by considering the optics of the eye and the resulting size of an image falling on the retina, in conjunction with ganglion cell density. This theoretical resolving power, or Nyquist limit, is usually reported as the finest separation of a series of evenly spaced parallel lines that an animal should be able to differentiate as lines rather than a uniform gray field. Each black/white line pair is called a cycle, and the acuity thus calculated is reported in cycles per degree. Degrees are a way of describing acuity that is independent of viewing distance; briefly, a degree describes how large an area on the retina an image covers. For example, the full moon subtends an area of about 4 degrees on your retina. The Nyquist limit thus calculated for the dog results in values of 4.5 to 6.5 cycles per degree. A threshold value of 5 cycles per degree would indicate that a dog could just discern 5 black-white stripes on a one degree spot, or 20 black-white stripes fit on an area roughly the size of the full moon. Humans would have no difficulty seeing this number of stripes in such a pattern. This is the theoretical visual acuity limit for the dog; it’s a little more difficult to find out if the dog can really make out this detail.
One way to see if an animal can discern fine detail is to train it to respond to large stripes, but not to a uniform gray field. As the stripes are made progressively finer, at some point the dog won’t be able to tell the difference between the fine stripes and the uniform gray field. Early studies provided a dismal view of canine acuity. A single Bull Terrier was unable to discriminate between two striped patterns of about 1 versus 35 cycles per degree—a blatant difference to human eyes (and for that matter, chickens and monkeys, which were tested in the same study) 16. In fact, people can discern stripes of up to about 30 to 50 cycles per degree, depending upon other factors; beyond that, the stripes appear to blend together into a homogeneous gray field. More recently, a single Poodle was able to discriminate between striped patterns as fine as approximately 6 cycles per degree 17, in much closer agreement with the theoretical Nyquist limit. It is also roughly the same as the threshold obtained in cats (the cat’s threshold is actually a little better, but that may be because the cat has been the subject of extensive sophisticated research in this area). Thus, even in a dog with emmetropic vision, its retinal composition limits its acuity to a much lower level than that humans enjoy.
For practical purposes the more useful question may not be the absolute limits of acuity, but rather the ability of dogs to use that acuity in discriminating forms. Most of the early form discrimination information available for dogs were byproducts of attempts to study experimental neurosis during Pavlovian conditioning 18; as such, stimulus parameters were seldom described explicitly. Dogs were able to discriminate a letter "T" from other figures, a cross from a square, a circle from a square, a circle from another circle twice its size, a circle from an ellipse having an axis ratio of 8:9, clockwise from counterclockwise movement, and horizontal from vertical movement 19.
One of the most imaginative early reports of dog visual recognition was by a 19th century scientist who trained his own dog to bring him different cards according to what the dog wanted. For example, on the cards were printed such phrases as "eat" or "go out" 20. Though undeniably cute, the procedure suffers from too many methodological problems to be particularly informative. In the 1930s, several groups of researchers used operant conditioning techniques to train dogs to make visual discriminations. One dog was trained to discriminate inverted from upright triangles (with sides of from 2" to 9"); the dog could still make the distinction even when only the triangles’ bases or base corners were shown 21. Another dog easily discriminated triangles from other figures 22. These abilities should come as no surprise to dog owners, many of whom have taught their dog hand signals. In fact, an interesting example of form and movement discrimination is the recent report of a bilaterally deaf Dalmatian taught to respond to over 30 signs of American Sign Language 23.
There is no question, then, that dogs have a much poorer ability to discriminate detail than people do---but is it so much poorer that it explains why dogs might even have difficulty recognizing their own human family members by sight? Possibly, especially when one more factor is taken into consideration: the smaller area of binocular vision in dogs compared to humans.
In humans, visual acuity tends to be a little bit better when both eyes are used. People have frontally placed eyes and a lot of overlap between the fields of vision of the two eyes. This area of binocular overlap allows for depth perception, the ability to gauge distance based upon slightly disparate images from the two eyes. Depth perception is especially essential in animals that need to jump or reach with great accuracy; it is more often well developed in tree-dwelling species or predators. A wide range of view is more essential in prey animals that must constantly scan the horizon.
The most obvious difference between canine and human eyes is in the placement of the eyes in the head. In general, dogs tend to have eyes placed more laterally than humans, giving them a more panoramic view at the expense of binocular vision. Although no fixed relationship exists between head type and eye placement, in general brachycephalic (flat nosed) breeds have more frontally placed eyes, and dolichocephalic (long nosed) breeds have more laterally placed eyes. In addition, long noses may partially obscure some of the visual field that would normally be covered by both eyes in these dogs, which would be the area right in front of their noses. Reported values of binocular fields in dogs range from 116 degrees in a "ratter" to 78 degrees in a setter 23. These values were recorded in the 1930s, and dog head conformation may have changed significantly since that time. Despite this, no modern estimates comparing breed visual fields exist. You can estimate your dog’s field of view by holding a tidbit directly in front of him so that it attracts his attention, and then quietly and slowly moving an object from behind your dog forward around the side of his head, making note of the point at which it is first noticed. Another simple method is to observe how far to the side of your dog you can move until your dog’s pupils are no longer visible to you, which would indicate the farthest possible extent of lateral vision. Dog breeders may wish to consider the role of eye placement for breeds in which depth perception is critical, such as coursing, retrieving, herding, or guiding dogs.
Of course, in real life dogs move their head and eyes, and so effectively increase their visual fields. Dogs do not have the range of eye movements that humans do, but they can move their eyes in one direction people can not: backwards! They have a muscle that humans do not, the retractor bulbi, which enables them to retract their eyes back into their sockets---an ability you may be reminded of when you try to put drops in your dog’s eyes. This brings us back to the topic of accommodation, the ability of the lens to "fine-tune" the focus on the retina for objects at various distances.
In young humans, accommodation can achieve as much as 15D change in power. Although modern values aren’t available for dogs, one early study obtained a range of accommodation in the dog of only 1D, compared to about 11D in monkeys using the same technique of electrically contracting the muscles that control lens tension 24. In humans lens tension affects how spherical the lens is, and thus its power. The dog lens, which is already more spherical than the human lens, as well as stiffer, would not have its shape changed as readily. In addition, the dog lacks some of the distinct groups of muscle fibers controlling lens shape that primates have. These observations have caused people to assume that dogs have virtually no accommodative abilities. This may be the case, but some researchers feel that accommodation may be accomplished in the cat (an animal with lens anatomy similar to the dog’s) not by changing the shape of the lens but by actually moving the position of the entire lens back and forth---similar to how a camera focuses! Finally, it has also been suggested that the relatively thin wall of the eyeball and well-developed retractor bulbi muscle may enable accommodation by slightly changing the entire shape of the eye. Until recently such questions have only been of theoretical interest, but as advances are made in intraocular lens prostheses, ultimately a lens that could accommodate like the natural lens would be desirable. Before that day comes, we need to know how much dogs can really accommodate, and what mechanisms they employ to achieve it. Finally, we will want to know whether it really matters if an image is slightly unfocused on the eye. Can the dog’s retina discern small enough differences that a slight degree of blur from unfocussed light would really matter?
This article has dealt with the dog’s ability to focus, detect, and perceive detail---that is, its sense of acuity. As stated at the offset, in all eyes compromises must be made. The dog has given up the ability to perceive very fine detail, but what has it gotten in return? It turns out that the dog may have made a good trade afterall. Next time this other major dimension of vision will be explored: the ability to detect light and color.