Vision in Dogs: Acuity, Color and What a Dog Can Resolve

1. What the Canine Eye Is Built For

1.1 A Retina Weighted Toward Low Light

The dog retina is rod-dominated. Rods respond at low light levels, do not carry color information, and pool their signals onto shared downstream cells, which raises sensitivity and costs resolution. Cones do the opposite: they need more light, carry color, and support fine detail.

Behind the retina sits the tapetum lucidum, a reflective layer that returns light that has already passed the photoreceptors for a second pass. That is the eyeshine in photographs, and it is one reason dogs can detect light and movement at levels where a human struggles; detecting is not the same as resolving detail, as chapter 2 shows. It also scatters, which blurs the returning image slightly, so the gain in sensitivity is paid for in sharpness.

The design is a set of trade-offs made for a crepuscular predator, not a defect measured against human eyes. Almost every apparent shortcoming in the sections below is the other side of a gain somewhere else.

1.2 Two Cone Types

The foundational measurement is now more than three decades old and has held up. Neitz, Geist and Jacobs (1989) tested three dogs behaviorally on increment-threshold spectral sensitivity and on direct color matching, and concluded that the dog retina contains two classes of cone pigment, computed to peak at roughly 429 nm and 555 nm. Dogs are dichromats.

Two cone classes divide the spectrum at a single crossover point. Wavelengths on one side are seen as one hue category, wavelengths on the other as the second, and somewhere between them lies a neutral point where a dog cannot separate the stimulus from gray. In practical terms the dog's world is built from blue and yellow, with reds and greens collapsing toward the same muddled band.

1.3 Field of View and Where the Eyes Sit

Eye position sets the geometry. Dogs have laterally placed eyes compared with humans, which widens the total field and narrows the region seen by both eyes at once. The standard veterinary review puts the dog's total field around 240 degrees with a binocular overlap far smaller than the human 120 degrees, varying substantially with skull shape (Miller & Murphy, 1995).

Wide field, less overlap. A dog notices movement in places a human would have to turn to see, and has less of the binocular disparity that supports fine depth judgment straight ahead. Anyone who has watched a dog miss a treat lying directly under its nose and find it instantly by smell has seen the arrangement at work.

1.4 Judging Distance

Depth information comes from two sources. Binocular disparity — the small difference between the images from two forward-facing eyes — is precise but works only where the fields overlap. Monocular cues such as motion parallax, occlusion, relative size and the way objects shift as the head moves are less precise but available across the whole field.

A narrower binocular overlap means less of the first and correspondingly more reliance on the second. This is one reason a dog often moves its head or takes a step sideways before committing to a jump or a catch: movement generates the parallax that a static view does not supply.

It also means that asking a dog to judge a distance while standing still, in flat light, against a low-contrast background is asking for the least reliable version of the judgment. Handlers in dog sports discover this empirically and adjust their setups without necessarily knowing why the adjustment works.

2. Acuity: The Number Everyone Quotes

2.1 The Familiar Claim

Almost every popular account states that a dog sees at 20 feet what a human sees at 75, sometimes citing a range of estimates without noting how wide that range is. Published acuity estimates for dogs vary widely depending on method, spanning roughly 2 to nearly 50 cycles per degree, and the method, the breed and the individual dog all contribute to that spread, which is not a small disagreement — it is the difference between a moderately blurry animal and one approaching human performance.

The reason for the spread is method. Electrophysiological measures, anatomical predictions from ganglion cell density, and behavioral discrimination tests do not measure the same thing, and only the last one asks what the dog can actually use.

2.2 What a Behavioral Test Found

Lind, Milton, Andersson, Jensen and Roth (2017) trained whippets, pugs and a Shetland sheepdog on a two-choice discrimination and measured the finest grating each animal could resolve. In bright light the dogs discriminated spatial frequencies between 5.5 and 19.5 cycles per degree, higher resolution than previously reported, with large individual variation. Humans in the same setup reached 32.1 to 44.2 cycles per degree. In dim light the dogs ranged from 1.8 to 3.5 and the humans from 5.9 to 9.9.

The authors' summary is the useful one: humans make visual discriminations from roughly three times the distance dogs can, in both bright and dim conditions. Not twenty times worse, not equivalent — about a third of the working distance, with the ratio holding as light drops.

2.3 What Follows for Training

A hand signal that is unmistakable at three meters is not necessarily legible at fifteen. If a recall signal works close in and fails at distance, the first hypothesis should be that the dog cannot resolve it, not that the dog is ignoring it (and distance is a dimension along which trained behavior routinely fails to transfer anyway).

Two design consequences follow. Signals should be large, involve whole-arm movement rather than finger positions, and differ from each other in shape rather than in fine detail. And they should be tested at the distance they are meant to work at, which is a measurement rather than an assumption (testing a signal is a measurement like any other).

2.4 The Individual Variation Matters as Much as the Average

The spread within the Lind study is the finding people skip. Some dogs resolved nearly four times what others did, in the same apparatus on the same task. A statement about "the dog" conceals a range that is large enough to change what a given animal can be asked to do.

Breed differences in eye and retinal structure are part of that story, and the sample included both a sighthound and a brachycephalic breed. But individual variation within any group is substantial, and no owner knows where their dog sits without testing.

3. Color: Less Than Human, More Than Assumed

3.1 Dichromacy Confirmed Behaviorally

Physiology says two cone classes. A behavioral confirmation came from an adapted version of the test used to diagnose human deuteranopia. Siniscalchi, d'Ingeo, Fornelli and Quaranta (2017) presented dogs with a modified Ishihara-style stimulus and found that they responded as red–green blind humans do.

This closes a gap that mattered. Cone pigments establish what the hardware could support; a behavioral test establishes what the animal does with it. Both now point the same way.

3.2 Color Beats Brightness

The more interesting result reverses a long-standing assumption. Early work suggested chromatic cues were unimportant to dogs in normal activity, and training practice followed: trainers avoided color as an information channel and relied on brightness or position instead.

Kasparson, Badridze and Maximov (2013) tested eight previously untrained dogs on stimuli that differed in both brightness and chromaticity, with brightness available as a usable cue. The dogs went by color. Under natural photopic lighting, chromatic information can be predominant even in an animal with only two cone types.

That is a direct, testable practical claim. Colored discriminanda are usable in training, provided the colors chosen lie on opposite sides of the dog's spectrum rather than being red-versus-green variants that collapse into the same category.

3.3 What to Choose, Practically

Blue and yellow are the working pair. A blue object against grass, or a yellow one against a dark surface, is separable for a dog; a red toy on a green lawn is a brightness problem, not a color problem, and often disappears entirely.

The same logic applies to equipment: agility contacts, target markers, scent-work containers and food-puzzle components are all more discriminable in blue and yellow than in the red, orange and green that dominate the pet market. This is one of the rare cases where a physiological finding translates directly into a purchasing decision.

3.4 What This Does Not Mean

Dichromacy is not colorblindness in the sense of seeing grayscale, and the standard "dog vision" filters circulating online are illustrations rather than measurements. They render a plausible transformation of the spectrum; they do not show what a dog experiences, and no image on a human screen can.

4. Not Every Dog Has the Same Retina

4.1 Visual Streak and Area Centralis

This is the finding that should be better known than it is. McGreevy, Grassi and Harman (2004) mapped retinal ganglion cell distribution across dogs of different head shapes and found a strong correlation with nose length. Long-nosed dogs tend toward a horizontally aligned visual streak of fairly even density; short-nosed dogs tend toward an area centralis, a concentration of cells in the middle of the field.

The two arrangements support different tasks. A streak gives sharp vision along the horizon across a wide sweep, which suits an animal scanning open ground for movement. An area centralis gives high resolution in a small central region, closer in function to the human fovea, and less resolution in the periphery.

4.2 What That Means in Practice

Two dogs standing side by side are not receiving comparable images. A long-nosed dog is better equipped to pick up a moving object at the edge of its field; a short-nosed dog is anatomically better equipped to examine something directly in front of it, including a human face at conversational distance. Whether either dog actually uses that difference is the question the next paragraph takes up.

The temptation is to build a behavioral theory on this immediately, and it should be resisted. What was measured is retinal anatomy correlated with skull morphology. The step from there to differences in attention, trainability or human-directed gaze involves several further assumptions, and the behavioral work that would license it is thin (which is the recurring problem with breed-level behavioral claims generally).

4.3 The Cost of the Conformation

The same skull shortening that produces a central retinal specialization also produces exposed eyes. Packer, Hendricks and Burn (2015) examined 700 dogs at a UK referral hospital and found 31 with corneal ulcers, most of them small brachycephalic breeds, with the Pug the most commonly affected. Dogs with nasal folds were nearly five times more likely to be affected, and affected dogs had a much shorter mean craniofacial ratio than unaffected ones.

A corneal ulcer is painful and can scar or perforate the cornea. So the conformation question is not only about what a dog can see but about whether the eye survives intact (which is one item on a longer list of conformational costs). Pigmentation genetics adds a separate set of ocular risks in some breeding practices (as the merle evidence documents).

4.4 A Behavioral Reading Worth Making

Where a dog's visual system is weighted toward peripheral motion detection along a horizontal band, a distant moving object is highly salient and a stationary object straight ahead is less so. That maps onto what handlers of long-nosed working breeds describe, and it is a plausible part of an account of why some dogs orient to movement at a distance before their handler has registered anything (which is where predatory sequences begin).

Stated carefully: retinal organization is one plausible contributor among several, and no study has isolated it as a cause of the behavior.

5. Time: Flicker, Motion and Screens

5.1 Temporal Resolution

A visual system has a speed limit. Above a certain flicker rate, an intermittent light is perceived as continuous, and that critical flicker fusion frequency has been measured behaviorally in dogs (Coile, Pollitz & Smith, 1989).

The comparative claim usually made from this work is that dogs resolve flicker at somewhat higher rates than humans do. Treat that as a real but modest difference rather than the dramatic one it becomes in popular writing, and note that flicker thresholds depend heavily on light intensity, stimulus size and which photoreceptor system is doing the work, which is exactly why published values vary.

5.2 The Television Question

The practical version of this comes up constantly. Older cathode-ray and low-refresh displays updated slowly enough that a visual system with higher temporal resolution could plausibly perceive flicker where a human saw a stable image. Modern displays refresh far faster, and the argument for dogs perceiving them as flickering is correspondingly weaker.

What has not been established is what dogs get out of screen content generally. Two-dimensional images lack depth, smell and the motion parallax that comes with moving the head, and a dog that ignores a television is not necessarily failing to see it.

5.3 Motion Sensitivity

The claim that dogs detect motion far better than humans is repeated everywhere and is supported less firmly than its confidence suggests. The anatomical argument is reasonable: a rod-dominated retina with heavy convergence and, in many dogs, a horizontal streak is well suited to detecting movement across a wide field in low light.

The behavioral measurement that would quantify the advantage against humans under matched conditions is largely missing. Directionally probable, quantitatively unestablished — which is how it should be described until someone runs the study.

5.4 Movement Registers Before Identity

A wide field with lower peripheral resolution produces a characteristic pattern: something moving at the edge of the field is detected reliably, and what it is has to be resolved afterwards, usually by turning to bring it into the better part of the field or by using another sense entirely.

That two-stage process explains a great deal of everyday behavior. The dog that startles at a distant figure and settles when it comes closer has not changed its mind; it has completed the identification. Treating the first response as the dog's assessment of the situation, rather than as the beginning of one, leads handlers to intervene at the wrong moment (which is also why distance is a central variable in graded exposure).

It also explains why stationary objects are missed. A motionless novel object in the periphery generates little signal, which is why dogs sometimes appear to notice an object only when it moves or when they have passed it and caught its odor.

6. What Dogs Actually Look At

6.1 Eye-Tracking Evidence

Perception is selective, and eye tracking shows what a dog's visual system prioritizes when nothing is asked of it. Somppi, Törnqvist, Hänninen, Krause and Vainio (2014) recorded eye movements in 23 pet dogs and eight kennel dogs viewing facial images.

Three results stand out. The dogs preferred conspecific faces over human ones. They concentrated fixations on the eye region. And the eye area of upright faces attracted longer and proportionally greater fixation than the eye area of inverted faces, regardless of whether the face was canine or human, while familiar faces and familiar eyes attracted more fixation than unfamiliar ones.

6.2 What That Supports

The inversion effect and the familiarity effect together indicate that face scanning in dogs is guided by more than the physical properties of the image. Something about facial configuration and about prior acquaintance shapes where the eyes go.

This is the evidential basis for a claim trainers make constantly: that dogs read faces, and the eyes in particular. The evidence supports the attention part firmly. What the dog extracts from the eye region — emotional state, intention, direction of attention — is a further question that eye tracking cannot answer (and the gap between looking and understanding has to be respected).

6.3 The Limitations Are Structural

These were static two-dimensional images on a screen, presented in a laboratory to dogs trained to rest their heads still and view freely rather than to physically restrained animals. Pet dogs fixated facial images longer than kennel dogs did, which means life experience shapes even this basic measure and complicates generalizing from any one sample.

It also means the results describe how dogs look at pictures of faces, and the step to how dogs look at faces attached to living bodies producing smells and movement is not free.

6.4 Watching Other Dogs

Most writing about canine vision concerns dogs looking at humans, which reflects who funds and designs the studies rather than what dogs spend their attention on. The eye-tracking work found the opposite priority: conspecific faces attracted more attention than human ones (Somppi et al., 2014).

That fits what field observation shows about dog–dog communication, where the informative signals are postural and directional — height, orientation, the direction of a gaze, the slow approach — and are exchanged at distances where fine detail is unavailable to both animals (as the free-ranging literature describes).

The design constraint runs the other way too. A visual signaling system used by an animal with moderate acuity and a wide field will favor gross, high-contrast, movement-based displays over subtle ones, and that is broadly what the ethograms contain.

6.5 Learning What to Look At

Where attention goes is not fixed at birth. Dogs acquire the habit of monitoring human faces and hands through a history in which those things predicted outcomes, and the eye-tracking difference between pet and kennel dogs is consistent with exactly that (Somppi et al., 2014).

It follows that visual attentiveness is at least partly shaped by experience rather than fixed, and that a dog which does not watch its handler has often not had a history in which watching paid. It also means the information a dog picks up by watching others is bounded by what it has learned to attend to (which is one limit on what social learning can deliver).

7. Seeing Is Not Reproducing

7.1 Dogs and Visual Illusions

Geometrical illusions are a way of asking how a visual system constructs a scene, because susceptibility reveals the processing rules rather than the optics. Byosiere and colleagues (2017) trained dogs on a fine size discrimination and then tested them on the Ebbinghaus–Titchener and Delboeuf illusions.

The dogs were susceptible to the Ebbinghaus–Titchener illusion, but in the reverse direction from humans and most tested mammals: the circle that looks larger to a human looked smaller to them, and the circle that looks smaller to a human looked larger. On the Delboeuf illusion they showed no consistent susceptibility, with marked individual differences throughout.

7.2 Why This Is More Than a Curiosity

A reversed illusion is not a weaker version of the human effect; it suggests the underlying processing differs. One reading is that dogs weight local features where humans weight the global configuration, which would produce exactly this reversal.

The consequence for practice is a general one. A dog looking at the same arrangement as its handler is not necessarily extracting the same relationships from it, and "he can see it perfectly well" is a claim about optics being used to settle a question about perception (which is a distinction that matters wherever we infer what a dog understood).

8. Vision Over the Lifespan

8.1 What Declines

Visual discrimination performance in dogs declines with age, and the effect is well documented because the paradigm was built for it. Head, Callahan, Muggenburg, Cotman and Milgram (1998) related visual discrimination learning ability to beta-amyloid accumulation in the aging dog brain, establishing the model still in use.

Later work with family dogs found young animals learning both discriminations and reversals significantly faster than old animals, with most old dogs failing to reach criterion on the reversal at all within the trial limit (Piotti et al., 2018).

8.2 Optics Versus Processing

Two things decline together and are easy to confuse. The optical apparatus ages: lens transparency decreases, light reaching the retina drops, and pupil responsiveness changes. Separately, the processing behind the eye slows and the ability to update a learned rule degrades.

A dog that stops responding to a hand signal it has known for a decade may be failing to see it, failing to process it, or neither — pain and hearing loss produce similar pictures (which cognitive dysfunction also has to be distinguished from). The distinction is clinical and not one an owner can make by observation.

8.3 What Helps

For an aging dog, the leverage is in the environment rather than in the dog. More light rather than less, since the dim-light gap between dogs and humans is proportionally as large as the bright-light one (Lind et al., 2017). High contrast at edges, steps and thresholds. Consistent furniture placement. Larger, slower signals given from a position the dog is already oriented toward. And a shift of weight onto channels that have not degraded (with olfaction the obvious one, and hearing worth checking rather than assuming, since age takes both).

8.4 The Other End of the Lifespan

Puppies do not start with an adult visual system. The eyes open at around two weeks and the retina and its central connections continue to develop for weeks afterwards, so early visual experience arrives at a system still being built.

The general principle that early experience shapes later sensory and social processing is well supported (which the developmental work sets out). What is not established for dogs is anything as specific as a schedule for visual development, and claims that a particular kind of visual enrichment at a particular week produces a particular adult outcome go beyond the canine evidence.

The defensible practical version is unremarkable: varied, non-overwhelming visual environments during the period when puppies are forming their expectations about the world, and no assumption that a puppy is seeing what an adult dog sees.

9. Working With the Visual System a Dog Actually Has

9.1 Signals

Make them large, make them differ in gross shape, and give them within the distance at which they have been tested. Whole-arm movements outperform hand positions; a signal held still against a plain background outperforms one made in front of a busy one; and a signal given while the handler is moving is a different stimulus from the same signal given while standing.

Backlighting is the common own goal. A handler standing with a bright sky behind them is a silhouette, and the difference between two arm positions collapses.

9.2 Distance and Contrast

Contrast, not color alone, does the heavy lifting at distance. A dark target on a light background remains discriminable when the color difference has ceased to matter, and blue or yellow against a contrasting background is better still (Kasparson, Badridze & Maximov, 2013).

Where equipment has to be found at speed — agility, retrieving, scent-work containers, field markers — choosing for both contrast and canine-visible hue costs nothing and removes a variable.

9.3 Light

Dogs detect light at lower levels than humans do, but in dim light humans still resolved finer detail than dogs, by a ratio similar to the one in bright light (Lind et al., 2017). Dusk training therefore makes visual signals harder to resolve for the dog as well as for the handler, even where the dog still notices movement that the handler misses.

This matters most in the situations where handlers assume the dog has an advantage: dark parks, evening recalls, indoor halls with poor lighting. A recall that has only ever been trained in daylight has not been tested in the conditions where it will be needed most.

9.4 Position and Orientation

Because binocular overlap is narrower than a human's and peripheral resolution is lower, where the handler stands changes what the dog can resolve. A signal given from behind or from the far periphery may register as movement without resolving into a shape.

The practical rule is to secure orientation before giving a visual signal, rather than treating the dog's failure to respond as a training failure. That also applies in encounters where a dog reacts to something before its handler sees it — the dog's field is wider (and what looks like an unprovoked reaction often is not).

9.5 The Multisensory Reality

Vision is one input among several, and often not the dominant one. A dog checking a novel object typically approaches, looks, then noses it, and the visual pass frequently functions as an orienting step toward the olfactory one.

Two consequences follow. Visual signals compete for attention against odor, which is why a hand signal fails on a lamppost that a verbal cue survived. And what looks like a visual discrimination task may be being solved by smell, which is a persistent confound in canine cognition research and in everyday judgments of what a dog "recognized" (as the work on human gestures has had to control for).

9.6 Testing It Instead of Assuming It

None of the above requires equipment. Pick the signal, pick a distance at which it works reliably, and then step back in fixed increments until it fails, recording the distance at which the failure appears. Repeat in the light conditions and against the backgrounds where the signal actually has to work.

Two things usually surface. The distance at which a signal stops working is shorter than handlers expect, and it moves substantially with background and light — the same signal that carries across a mown field disappears against a hedgerow. Both are useful numbers to hold, and neither can be estimated by watching a dog succeed at close range.

The same procedure identifies the more common failure, which is that the dog is not oriented toward the handler when the signal is given. A signal delivered into the periphery of a system with low peripheral resolution has not been given in any useful sense, and no amount of repetition fixes that.

9.7 Arousal Narrows What Gets Processed

Perception is not constant across states. A dog at high arousal processes its visual environment differently from the same dog at rest, and the practical version is familiar: signals that work in a quiet room stop working in a busy one, at a distance and detail level that were adequate five minutes earlier.

Some of that is competition for attention rather than a change in the eye. It does not much matter which, because the handling response is the same — reduce what the signal has to compete with, increase its size and contrast, and shorten the distance (with arousal acting as a constraint on what a dog can do at all).

The cumulative version matters too: a dog that has already absorbed several demanding events processes the next one from a worse starting position, which is a claim about load rather than about vision (and one whose evidence base is thinner than its popularity).

10. Summary at a Glance

Dogs are dichromats with cone pigments peaking near 429 and 555 nm. Established behaviorally in three dogs by increment-threshold and color-matching tests (Neitz, Geist & Jacobs, 1989), and confirmed with a modified deuteranopia test showing dogs respond as red–green blind humans do (Siniscalchi et al., 2017).

Color is used, and used preferentially. Eight untrained dogs chose by chromaticity rather than brightness when both cues were available (Kasparson, Badridze & Maximov, 2013), reversing the assumption that built a generation of training practice.

Acuity is roughly a third of human working distance. Behavioral testing gave 5.5–19.5 cycles per degree in bright light against 32.1–44.2 for humans, and 1.8–3.5 against 5.9–9.9 in dim light (Lind et al., 2017), with large individual variation.

Retinal organization tracks skull shape. Long-nosed dogs tend toward a horizontal visual streak, short-nosed dogs toward a central concentration of ganglion cells (McGreevy, Grassi & Harman, 2004).

The same conformation carries an ocular cost. Among 700 referral-hospital dogs, corneal ulcers were concentrated in small brachycephalic breeds, and nasal folds raised the odds nearly fivefold (Packer, Hendricks & Burn, 2015).

Dogs prioritize faces and eyes when free-viewing. They preferred conspecific faces, fixated the eye region, and responded to inversion and to personal familiarity (Somppi et al., 2014).

Canine visual processing is not a scaled-down human version. Dogs showed reversed susceptibility to the Ebbinghaus–Titchener illusion and no consistent susceptibility to the Delboeuf illusion (Byosiere et al., 2017).

11. Research Gaps and Critical Appraisal

The foundational color work rests on three dogs. Neitz, Geist and Jacobs (1989) is careful psychophysics on a sample of three, and the later behavioral confirmations (Kasparson et al., 2013, with eight dogs; Siniscalchi et al., 2017) are also small. The conclusion is consistent across methods, which is what carries it, not sample size.

Acuity estimates still vary widely. Published values differ with testing method, anatomy and the individual animal, spanning roughly 2 to nearly 50 cycles per degree, and the best behavioral study to date used a small, breed-heterogeneous sample with wide individual spread (Lind et al., 2017). Any single figure quoted for "dog vision" is a simplification of a genuinely unsettled measurement.

The retinal-to-behavioral step is not made. The correlation between ganglion cell distribution and nose length is solid anatomy (McGreevy, Grassi & Harman, 2004). What long-nosed and short-nosed dogs consequently do differently in the field has not been established, and the inference is frequently made as though it had.

Motion detection is asserted more than measured. The anatomical case is plausible; a controlled comparison of motion thresholds in dogs and humans under matched conditions is not part of the canine literature.

Screen-based findings inherit screen-based limits. Eye tracking and illusion work both present two-dimensional images to animals whose natural visual input includes depth, parallax and concurrent odor (Somppi et al., 2014; Byosiere et al., 2017). The results are real; their transfer to living scenes is untested.

Flicker and temporal resolution rest on a small older literature. Behavioral flicker fusion in dogs (Coile, Pollitz & Smith, 1989) is a narrow evidence base for the confident claims made about dogs and screens, and values depend strongly on stimulus conditions.

Almost no training recommendation here has been tested. Signal size, contrast choice, lighting and handler position follow from the perceptual findings rather than from trials comparing them. That study would be straightforward to run and has not been reported in the sources reviewed here.

12. Conclusion

Dog vision is usually described by subtraction: fewer cones, less acuity, no red. That framing gets the facts roughly right and the conclusion wrong. The canine eye is organized for a different problem — detecting movement across a wide field in poor light — and it solves that problem while giving up detail and part of the spectrum. Where dogs use what they have, they use it more than the older literature credited: color turned out to be informative, faces and eyes attract attention, and acuity is better than the standard figure implies.

The useful shift for practice is from asking whether a dog can see something to asking what it can resolve, from where, in what light, against what background, at what age. Those are measurable, and they change what a handler does. The claim that a dog knew a signal and chose to ignore it should be the last hypothesis on the list, not the first.

Key Insights (Takeaways)

• Dogs are dichromats, and color is usable information rather than a lost channel. Two cone classes peaking near 429 and 555 nm (Neitz, Geist & Jacobs, 1989), with dogs choosing by chromaticity over brightness when both were available (Kasparson, Badridze & Maximov, 2013). Blue and yellow are the working pair; red on green collapses.

• Acuity is about a third of human working distance, not a twentieth. 5.5–19.5 cycles per degree in bright light against 32.1–44.2 in humans (Lind et al., 2017). A signal that works at three meters may be unresolvable at fifteen, and that is a perception problem rather than a compliance problem.

• Retinal organization differs by skull shape. Long-nosed dogs tend toward a horizontal visual streak, short-nosed dogs toward a central concentration (McGreevy, Grassi & Harman, 2004). Two dogs side by side are not receiving the same image — though what they consequently do differently has not been established.

• Dogs look at faces and at eyes, and familiarity changes where they look. Conspecific faces preferred, eye region prioritized, inversion and personal familiarity both altering fixation (Somppi et al., 2014). What they extract from the eye region remains a separate question.

• Canine visual processing is not a downscaled human version. Reversed susceptibility to the Ebbinghaus–Titchener illusion and none to the Delboeuf (Byosiere et al., 2017). "He can see it fine" answers a question about optics, not about perception.

• The lever for an aging dog is the environment. More light, higher contrast at edges and steps, larger and slower signals, and a check of the other senses before concluding anything about the dog's willingness.

References

Byosiere, S.-E., Feng, L. C., Woodhead, J. K., Rutter, N. J., Chouinard, P. A., Howell, T. J., & Bennett, P. C. (2017). Visual perception in domestic dogs: Susceptibility to the Ebbinghaus–Titchener and Delboeuf illusions. Animal Cognition, 20(3), 435–448. https://doi.org/10.1007/s10071-016-1067-1

Coile, D. C., Pollitz, C. H., & Smith, J. C. (1989). Behavioral determination of critical flicker fusion in dogs. Physiology & Behavior, 45(6), 1087–1092. https://doi.org/10.1016/0031-9384(89)90092-9

Head, E., Callahan, H., Muggenburg, B. A., Cotman, C. W., & Milgram, N. W. (1998). Visual-discrimination learning ability and beta-amyloid accumulation in the dog. Neurobiology of Aging, 19(5), 415–425. https://doi.org/10.1016/S0197-4580(98)00084-0

Kasparson, A. A., Badridze, J., & Maximov, V. V. (2013). Colour cues proved to be more informative for dogs than brightness. Proceedings of the Royal Society B, 280(1766), 20131356. https://doi.org/10.1098/rspb.2013.1356

Lind, O., Milton, I., Andersson, E., Jensen, P., & Roth, L. S. V. (2017). High visual acuity revealed in dogs. PLoS ONE, 12(12), e0188557. https://doi.org/10.1371/journal.pone.0188557

McGreevy, P., Grassi, T. D., & Harman, A. M. (2004). A strong correlation exists between the distribution of retinal ganglion cells and nose length in the dog. Brain, Behavior and Evolution, 63(1), 13–22. https://doi.org/10.1159/000073756

Miller, P. E., & Murphy, C. J. (1995). Vision in dogs. Journal of the American Veterinary Medical Association, 207(12), 1623–1634.

Neitz, J., Geist, T., & Jacobs, G. H. (1989). Color vision in the dog. Visual Neuroscience, 3(2), 119–125. https://doi.org/10.1017/S0952523800004430

Packer, R. M. A., Hendricks, A., & Burn, C. C. (2015). Impact of facial conformation on canine health: Corneal ulceration. PLoS ONE, 10(5), e0123827. https://doi.org/10.1371/journal.pone.0123827

Piotti, P., Szabó, D., Bognár, Z., Egerer, A., Hulsbosch, P., Carson, R. S., & Kubinyi, E. (2018). Effect of age on discrimination learning, reversal learning, and cognitive bias in family dogs. Learning & Behavior, 46(4), 537–553. https://doi.org/10.3758/s13420-018-0357-7

Siniscalchi, M., d'Ingeo, S., Fornelli, S., & Quaranta, A. (2017). Are dogs red–green colour blind? Royal Society Open Science, 4(11), 170869. https://doi.org/10.1098/rsos.170869

Somppi, S., Törnqvist, H., Hänninen, L., Krause, C. M., & Vainio, O. (2014). How dogs scan familiar and inverted faces: An eye movement study. Animal Cognition, 17(3), 793–803. https://doi.org/10.1007/s10071-013-0713-0