The eye (any eye) is a miracle of evolution, which has filled its structure at every scale with an apparent intentionality that is the hallmark of complicated systems under intensive selective pressure. It is very much possible to look at almost every design feature and state a compelling reason why it’s there, which makes it a rich place for the engineering-minded to find fun little design features.

This post is meant to discuss topics that are less frequently discussed elsewhere, and not often discussed in introductory courses, rather than anything that can be quickly gleaned off of common summaries. Another way to describe it is that I intend to dive a bit into the kind of fridge-logic questions you might come up with randomly after opening the evolutionary eye-fridge, particularly the ones that turn out to have interesting answers.

As such, this post is going to be a bit of a grab bag of topics, rather than a focused point-by-point breakdown of the eye. I’m still going to group topics broadly into general categories, just for organizational reasons.

One thing I will not go into is visual processing in the retina or visual cortex–even though this is a fascinating topic and one of the best understood parts of neuroscience (relative to the rest of the brain). That’s simply for space considerations; including that topic would balloon this post to enormous sizes. Maybe next time.

Part I: Optics and the Lens

Biology Basics (Brief)

To talk about any of this, it is necessary to have some basic grasp of the eye’s anatomy and functioning. That is, you need to know the following key facts and have some sense of the diagrams below:

Key parts of the eye, in order from outside to inside:
1. Cornea - outermost part you can touch (don’t touch)^[1]
2. Aqueous humor - thin watery goop
3. Lens - it’s a lens, it focuses light
4. Vitreous humor - thick watery goop, most of the eye by volume
5. Retina - thin layer of tissue at back of eye, has the famous rods and cones that actually detect light
6. Optic nerve - transmits signal from retina to brain
Basic functioning: Light hits your face, is focused by the cornea^[2] and natural lens, resulting in an image projected onto the retina, which then does a ton of image processing before passing its in-progress data to the optic nerve. (which connects to the visual cortex in the brain which does the rest of the processing)

Now, let’s explore some optics-adjacent questions that might come to mind, provided one thinks about the optics.

Chromatic Aberration (refractive lens drawbacks)

Anyone who has spent any time working with refractive lenses (as opposed to reflectors) is familiar with the fact that not all frequencies of light are focused in quite the same way. Specifically, short-wavelength light focuses slightly closer to the lens than long-wavelength light. In cameras, this often results in color fringing in photographs:

Notice the purple tinge on the edge of the hair. (source)

and results in camera-makers spending considerable effort on modifications to their lenses to reduce this effect. This is also just one of the reasons why reflectors are preferred for telescopes.

You may have noticed that you never see these effects in everyday life. So does the biological lens have a very clever system for eliminating chromatic aberration?

In a word: no. Your brain has simply learned not to see it. Human eyes often exhibit a relatively strong degree of chromatic aberration (sometimes cited as an average of 1.75 Diopters). Some mechanisms in the eye compensate for this–for instance, the center of the fovea lacks high-frequency blue photoreceptors, which then reappear in a ring around the fovea–but at the end of the day, the brain simply covers for the imperfections of its visual instrument.

As a small consequence of this, at nighttime, this combines with the optimal absorption frequency of human rods (500 nm, roughly blue-green) to cause vision to be somewhat more near-sighted. When combined with a dilated pupil (=more light into the eye, but less precise focus), this can cause serious vision impairment at night, particularly in older individuals.^[3] It is more important to wear your glasses when driving at night.

Why Everything Looks Weird Underwater (but not for fish)

So, refractive lenses work by bending light at the surface of the lens between the external medium and the interior of the lens. Leaving aside complications like multimaterial lenses, Snell’s Law makes clear the degree of bending depends on the refractive index of both the external medium and the lens itself. If you’ve paid careful attention to the discussion above, then you can discern that there are three major points of refraction in the eye:

The cornea, where light passes from air to water.^[4]
The outer surface of the lens, where light passes from water into the lens.
The inner surface of the lens, where light passes from the lens back into water.^[5]

Naturally, if the eye enters water rather than air, the refraction from 1 changes drastically–in fact, it mostly disappears, since both sides of the interface are, roughly speaking, water. But recall that the cornea is two-thirds of the refraction! This is why everything looks blurry underwater unless you’re wearing swimming goggles^[6]–you’re effectively extremely far-sighted underwater.

But now the follow-on question from the title of this section suggests itself: What about aquatic animals? They need to see too!

The answer follows naturally from optics: without the cornea providing much refraction, the lens needs to do a lot more of the work. One possibility is to use a different material, but for reasons we will discuss next section, this is probably evolutionarily difficult. Much easier to simply have a much bigger lens.

Fish lenses are nearly spherical, and this introduces its own problem: thus far all our discussions about lenses have assumed that they always focus light neatly into a single focal point. This is actually not true of lenses with a spherical surface–it requires a parabolic surface. For thin lenses like in the human eye, this is not that big a deal, thanks to the thin lens approximation^[7], but for a roughly spherical lens it absolutely does matter. Rather than learn how to make parabolic surfaces, fish instead introduce a refractive index gradient within the lens itself to correct the effect.

In humans and land vertebrates, focusing the lens is done by reshaping the lens. Presumably this is more difficult for a thicker spherical lens, which may be why in fish focusing is instead done by moving the lens back and forth.

And now I’ve successfully tricked you into learning some geometric optics, assuming you read this far.^[8]

Crystallins: Building Large Functional Structures Biologically (spoiler: it’s frequently dead cells)

The overall structure of the lens resembles a motif repeated frequently everywhere in biology, often when a biological system needs to generate a large-scale object that is more important for its structural properties than whether or not it is, technically speaking, alive.^[9] A living layer of cells will sustain itself as the base layer, dividing occasionally. As the layer gets thicker, some cells are forced inward (in the case of the lens), and will gradually turn themselves into structural cement, entombing themselves in a sacrifice for the greater good.^[10]

In the case of the lens, the living layer is the outer layer, the lens capsule, which also circulates fluid and oxygen into the core (blood-vessel-free for transparency!) to keep it fresh. Moving inward, the now-lens-fiber cells lengthen themselves into long fibers, binding tightly to each other and filling themselves with transparent proteins called crystallins. Eventually, going inward, the cells eject their nuclei and other organelles entirely, becoming effectively zombie cells formed of only a membrane and crystallins, which now fill every corner of the cell with a transparent gel. The lens grows gradually this way your entire life–necessary as the eye changes shape with age–but the structure is permanent, with the deepest core laid down during embryohood.

As such, the lens, and the transparent crystallin proteins that form it, need to survive an entire lifetime of UV radiation and other insults, and by and large they do a very good job of it. The name “crystallin” is perhaps one of the worst chosen names ever, since the one thing these proteins seem to avoid at all costs is forming any kind of crystal or aggregate. These proteins stay water-soluble and transparent at extremely high density for decades.^[11]

The crystallin proteins come in three flavors: alpha, beta, and gamma. Alpha-crystallins are the most common, and seem to have as their main purpose maintaining and reinforcing protein structure–remember, the whole gel has to stay in place for decades.^[12] Beta-crystallins and gamma-crystallins have a less clear purpose, but presumably do something.

Bizarrely, all forms of crystallin are closely related to common catalytic enzymes, particularly those in the liver, and many of them frequently moonlight in other roles in other parts of the body. Human beta-crystallin also binds calcium and has some role in axon maintenance, and in other animals some crystallins straight up act as liver enzymes. Even alpha-crystallin is secreted in small quantities in other cells and seems to be protective against cellular stress.

Intriguingly, Cephalopods, which are invertebrates that evolved eyes independently, also use crystallins for their lens (derived from different but similar enzymes). The evolutionary pathway from crystallin-like enzyme to lens material thus seems to be “natural”, and may be simply co-opting an accidental property of certain enzymes to form nice transparent gels when concentrated. If so, this may be why lenses between fishes and humans only change shape and not fundamental material–crystallins already do the job quite well, and it is easier to change the shape of a lens than design an entirely new protein with a different refractive index.

Failure Modes (practical knowledge that may apply to you!)

I won’t go into these in too much detail since these are very commonly discussed, but there are a lot of people with myopia (near-sightedness), for whom some basic interesting information may be useful and welcome. Beyond that, I discuss some of the most common conditions of the eye you are likely to run into in daily life, in the vein of a PSA.

Myopia and Hyperopia

Near-sightedness and far-sightedness respectively, and these are by far the most common conditions of the eye, and chances are good^[13] that you, yes you, reading this have myopia and are wearing glasses or contacts because of it.

Both of these conditions are failures of optical calibration, resulting from a lens that focuses light too far in front of and behind the retina, respectively. Except, it’s usually not the lens’s fault! Rather, it’s the rest of the eye that’s grown too long or too short around the lens. Typically, the problem is the vitreous chamber, that is, the thing that holds the vitreous humor. However, that’s just usually, there’s variants.^[14]

Presbyopia

A fancy word for what is commonly considered just “getting old” and needing reading glasses.

The human eye adjusts focus for near and far by contracting muscles around the lens, reshaping the lens for different focal lengths. As one grows older, the proteins in the lens gradually stiffen and harden, making this focusing process increasingly difficult. Every year, the ability of your eye to adjust its focus diminishes slightly, and eventually, usually by your early 40s, this problem becomes noticeable.^[15]

Cataracts

In some cases, the crystallins of the lens also break down and begin to lose transparency, resulting in cataracts, one of the most common eye conditions of old age–by the mid-to-late 70s, about half of Americans can expect to have this. In severe cases, the problem is addressed by removing the natural lens and replacing it with an artificial lens, a shockingly common procedure given how metal it sounds. Cataracts are also the world’s leading cause of blindness, despite being addressable by what is considered a routine surgery.

As implied, some degree of reduction in transparency is normal for everyone, and as you age the world actually dims because of it–this is further amplified by the pupil constricting more and more over time as the iris muscles weaken.

Glaucoma

Comes in two flavors. Open-angle or Chronic Glaucoma, the most common kind, is what we will discuss here. The other kind, Closed-angle or Acute Glaucoma, is an ophthalmological emergency (yes, those exist outside of being stabbed in the eye or whatever) where eye pressure spikes to extremely high levels, threatening major damage to the optic nerve, requiring immediate treatment.

The chronic version is the slow version of that, where damage accumulates to the optic nerve over years or more, leading to the slow loss of peripheral and (eventually if untreated) central vision. The cause is high pressure inside the eyeball–the fluid of the inner eye is carefully filtered from the bloodstream, helping to nourish the otherwise bloodless lens and cornea, before being drained again. A normal range of eye pressures is considered 10-21 mmHg (relative to ambient pressure–contrast the 760 mmHg pressure of the atmosphere at sea level)–too high a pressure can cause gradual damage to the optic nerve. If you’ve been to the ophthalmologist and had them bring a weird machine to gently poke your eye, or perhaps blow a puff of air instead, they’re measuring intraocular pressure (IOP).

Glaucoma is treated firstly with eye drops and if that fails, with various surgical interventions.

In some cases, glaucoma damage may occur even with “normal” IOP. In those cases the treatment is similar, and it is supposed that the person might have unusually sensitive optic nerves. These cases are tougher to detect, since they pass as normal under the usual IOP measurement, and must be diagnosed by proxy from vision damage.

Macular Degeneration

A poorly understood condition where, in the chronic version (“Dry” Macular degeneration), the retina itself breaks down in old age, with photoreceptors and eventually entire sectors of the central retina failing, greatly damaging central vision. It is irreversible and progressive once it starts, but fortunately it’s not that fast. I personally think of it as Alzheimer’s for the retina–kind of a terrible analogy biologically, but it conveys the idea.

Unfortunately, prevalence in the US in 80+ year olds is 12% and it is the leading old-age cause of blindness. A number of clinical trials are underway and there is much in the research pipeline, but no sign of a silver bullet.

It is not clear what causes this, other than the body’s gradual long-term maintenance mechanisms failing. The disease has a strong genetic component, but all the usual things about good diet and exercise can help prevent it, as does not having high blood pressure, cholesterol, or obesity. Also don’t smoke.

This time it is Caucasians who have lost the genetic lottery. In fact, much of the 12% prevalence in the US I mentioned above is driven by just Caucasians.

Source: National Eye Institute (via archive.org)

There are outlines of potential reasons for this. Dopamine is synthesized by and used in the retina. Its precursor, L-dopa (also a Parkinson’s Disease drug), is synthesized from the amino acid Tyrosine in the same pathway that synthesizes melanin (the brown skin pigment). Caucasians have a lot of genetic polymorphisms affecting enzymes in this pathway, seem to have less L-dopa, and, well, have white skin. Epidemiological evidence suggests that patients on L-dopa (see: Parkinson’s) have a lower rate of macular degeneration, there are some reports that administering L-dopa helps with the disease, and there is even an outline of a mechanism: the only known direct receptor for L-dopa (GPR143) is expressed in melanocytes and the Retinal Pigment Epithelium (discussed way below).

And yes, people with lighter eyes do develop macular degeneration at a higher rate, though this on its own is not enough to explain the Caucasian effect here.

Diabetes-Related Retinopathy

The name almost entirely explains it, and I’m not going to go into it much other than to note it is blood sugar-related progressive damage to the retina. Try not to have diabetes, and if you do have it, try to keep your blood sugar well controlled.

Retinal Detachment

This is not common, but I’m inserting this as a PSA, particularly for those with a higher chance of developing this (high myopia, family history, cataract surgery). It happened once to a friend of mine.

This occurs when a portion of the retina literally detaches from the back of the eye. Left alone, this might spread to the rest of the retina, literally like a sheet tearing off, and is another bona fide ophthalmological emergency, addressable with surgery.

Signs include:

A sudden increase in the number of floaters in the eye
Persistent flashes of light in vision
A dark curtain or shadow appearing to move over the visual field
Sudden blurred vision

This is not something that should be ignored if observed and is a good reason to immediately take yourself to your ophthalmologist or, failing that, local ER. An ophthalmologist will understand that this is an emergency and treat it as such.

Part II: Photoreception

Basics II: Rods, Cones, and Color as a Matter of Information

Photoreception is the process by which the retina converts incoming light into neural signals traveling down the optic nerve.

The front lines of this process are the photoreceptors, the class of light-sensitive cells lining the retina. Each of these photoreceptors may be thought of as a single pixel on the retina, the smallest unit of light detection.

Behold the retina (in cell layers):

The photoreceptors are divided into rods and cones, with the former (~92 million in the human eye) responsible for dim light and night vision and the latter (~4.6 million) responsible for color vision. In humans, there are three types of cones, called L, M, and S (for Long, Medium, and Short wavelength) corresponding to the wavelength of light that they detect best.

The sensitivity spectrum of each class of photoreceptor, by wavelength of absorption. Black line is rods (only 1 type), colored lines are the varieties of cones. From Wikipedia.

From this, it is easy to see with a little thought how humans achieve color vision, and why our color wheels are divided into three colors (for instance red-green-blue or cyan-magenta-yellow). It is basically a sampling problem–any given frequency of pure single-frequency light will activate each category of cone a particular amount–say, 70% L, 40% M, 30% S. The brain can interpret this correctly as a blueish-cyan.

But this one-to-one ratio breaks down immediately once you consider mixtures of frequencies, or full spectrums, where infinitely many different mixtures of light can excite the same reaction in the three cone types. To the human eye, the entire space of possible colors is thus described by essentially three numbers–it cannot do any better in information theory terms. Hence the color wheel.^[16]

More fundamental than that, this is how your brain has to represent the visual spectrum, as a mixture of three points of information. This is simply what it has to work with, and our very perception of the world is shaped by it.^[17]

But wait: after all talk of three, the observant may have noticed that rods are represented on the spectrum chart as well. Sure, they’re “different” and dim-light focused, but couldn’t they be used as a sort of poor man’s fourth cone?

Well, no, apparently not. It’s a matter of optimization and saturation. In order to distinguish between colors in the regime they are optimized for (daytime), cones need to respond in a perceptibly different way over the relevant frequency range. Rods, with an absolute light sensitivity two orders of magnitude higher than cones, are saturated at all times in daylight, providing no useful informational content. To be useful in daytime cones have to be less sensitive. Moreover, taking advantage of the daytime environment, they “reset” themselves much faster, allowing for perception of fast motion and quick-changing environments.^[18]^[19]

Because of this, the brain treats signals from rods as essentially a binary white/black signal, rather than using it for any color information. In the narrow twilight range where both cones and rods provide some information, this causes blue-green parts of the world to appear a bit more bright–the Purkinje Effect.

As for why there aren’t multiple types of rods, to allow for some kind of color sensitivity at night, or some kind of advanced mechanism for tuning photoreceptor sensitivity over the course of the day, I can only say that evolution only optimizes to what is necessary. You may as well ask why Humans don’t have five cone types (like some birds), or the many nighttime vision adaptations of cats.

Detecting Photons with Cells and Chemicals (quantum mechanics!)

As discussed above, the pixel-level unit of light detection is the individual photoreceptor cell, whether it be a rod or cone.

The best picture I found on a quick search of rods and cones came from, surprisingly, Britannica. Managing to stay relevant!

As you can see, both rods and cones have an arrangement of lamellar discs at top, where the photosensitive proteins are kept, with the rest of the cell at the bottom responsible for maintenance and synapsing with the neurons of the retina. As always, some of the most interesting questions are the most childlike: Why are they shaped differently?

Recall: cones deliberately need less sensitivity, so they are shorter and contain fewer discs. The reason for the tapered shape is more subtle–the cone shape acts like a waveguide, ensuring that light hitting the cone straight on is detected more reliably than stray light rays hitting at an angle.^[20] This acts as a form of noise reduction (apodization), improving spatial acuity. This is the kind of luxury you can afford to have when photons are plentiful, as in daytime.

In any case, the proteins responsible for light detection are collectively called the opsins, with the primary rod molecule called rhodopsin and each of the three types of cone carrying their own slightly different variant.^[21]

Each opsin is a G-protein-coupled receptor clutching a copy of the molecule retinal^[22], a light-sensitive molecule that is heavily recycled, but in vertebrates ultimately manufactured from the carotenes, originally photosynthetic molecules inside plants. You may know it as Vitamin A. Thus, in a way, vertebrate vision is co-opted from plant photosynthesis.

Left: pre-photon, right: post-photon (images stolen from Wikipedia but I lost the link)

Note: the pattern of double and single bonds here is typical for organic molecules that have pigment or light absorption properties. This “resonance” structure ensures the electrons in the π-orbitals are delocalized, lowering the energy necessary to push an electron into the next orbital, into the range carried by visible light.^[23]

When a photon strikes a retinal molecule, an electron moving in the higher orbital “frees” the lower orbitals temporarily–the p-orbitals, previously wedged together, now have more degrees of freedom with an electron missing–and they quickly rotate into the more favorable straight conformation, forcing the surrounding opsin to distort as well.^[24] In the brief period before another enzyme arrives to return the retinal to its original conformation, the opsin may interact with another G-protein, transducin, which in turn activates phosphodiesterase (PDE6) which then catalyzes the conversion of cyclic GMP to GMP, which in turn signals sodium channels in the cell membrane to close.

Students of quantum mechanics may note that this makes retinal, and indeed every system involved–the opsin protein, the signaling pathway, the cell itself–a kind of single-photon quantum detector. It may be interesting to reflect on such questions as “At what point during the ensuing signaling cascade does ‘observation’ occur?” Though please read about Quantum Decoherence and don’t go down any consciousness rabbit holes here.

Nonetheless, as a detector the efficiency is remarkable, with the rod rhodopsin–optimized for dark sensing–achieving a quantum efficiency (percentage of photons detected) of 65-67% at optimal wavelength.^[25] For comparison, consumer camera CCDs have a quantum efficiency of 30-60%, while scientific equipment can be well north of 90%–with high-end cooling and the use of exotic metals. 65-67% is really good for an organic molecule at body temperature.

But back to the “transducin activates the phosphodiesterase which then catalyzes…” yadda yadda. This kind of seemingly redundant multistep cascade is all over biology and serves a clear purpose: chains of quickly resetting intermediates allow for signal amplification on a potentially massive scale, with each step of the chain serving as a several-order-of-magnitude increase in the size of the signal. In this case, it converts what may be as small as a single photon’s worth of input (in the case of a rod) to a reaction the size of the entire cell.

As a small aside that is difficult to bring up contextually, rods behave in a manner opposite to almost every other neural detector, hyperpolarizing when in the presence of light and stopping the flow of glutamate across the synapse, which removes the inhibition of subsequent neurons. This kind of reverse polarity saves considerable energy in rods during daytime–but does the opposite at night. Insect eyes behave the “normal” way and it is still an open question why this unusual reverse polarity behavior exists. There is some speculation it saves energy overall, but this is far from clear.

Also fun fact, saving energy here does matter, photoreceptors are the most energy-intensive cells in the entire body.^[26]

The Macula and Fovea (efficiency compromises)

The macula is the relatively cone-rich center of the retina, responsible for the eye’s color and higher resolution vision. It is anatomically distinguished from the more peripheral parts of the eye by the presence of two or more layers of ganglionic processing neurons and a distinct yellow coloration that serves to protect the macula from near-ultraviolet and harsh light. This yellow color comes from carotenoid pigments which, like retinal, are derived from photosynthetic plant carotenes. Another reason to eat your vegetables.

At the center of the macula is the fovea^[27], the highest-resolution, nearly pure-cone center focus of the eye.^[28] Here, the blood vessels and neurons that normally sit in front of the photoreceptors are shifted to the side, giving the photoreceptors a clearer view–the diverted blood vessels and neurons are stacked in a bulge surrounding the fovea. Moreover, the fovea is heavily prioritized by the neural processing of the retina–while in the periphery dozens or even hundreds of rods might share a single optic nerve axon, the cones of the fovea address optic nerve axons nearly one-to-one.

The pit in the center here is the fovea. (source)

Relative acuity of parts of the eye at various degrees away from the fovea. (source)

You may have noticed that I mentioned blood vessels and neurons needing to get out of the way of the photoreceptors in the fovea. An astute reader might ask: why are they in front of rather than behind the photoreceptors?

That Thing About the Retina Being Backwards (also cephalopods)

One of the most common fun facts about the vertebrate retina is that it is essentially “backward”, at least relative to what one might expect.

Traditional diagram of the retina. Despite all seeming logic, light enters from the *top*, needing to pass through layers of neural tissue and blood vessels to reach the photoreceptors. (source)

This arrangement has one other consequence: Blood needs to pass through the photoreceptors to reach the retina, and the optic nerve needs to do the same to communicate. In the vertebrate eye, the main blood vessels and optic nerve are centralized to pass through in one spot, the blind spot, where nothing can be seen.^[29]

The usual story is that this backwardness and blind spot represent a kind of evolutionary legacy code,^[30] a leftover from the vertebrate evolution of the eye that we are stuck with, because there is no simple way to “correct” the manner in which the retina grows during embryonic development.

This argument is made stronger by the observation that in Cephalopods, which evolved eyes independently, the retina is the “right way” around and doesn’t have a blind spot.

The thing is, while all that is still considered a good part of the reason, there are now known to be potential advantages to the “backwards” vertebrate arrangement. Firstly, the presence of a cellular layer in front of photoreceptors allows for protection of the photoreceptors from UV radiation and excessive light. While the glial Muller cells above the photoreceptor are shaped such as to maximize light transmission, they transmit green and red much more effectively than blue, effectively pre-filtering incoming light towards the colors most useful for daytime vision.

Secondly, vertebrates have an additional retinal pigment epithelium, a layer of support cells connecting the photoreceptors to the underlying choroid of the eye, which does have blood vessels. This layer of nearly black cells supports the photoreceptors by absorbing the light after it passes through, maintaining ionic balance, recycling retinal, transporting nutrients, and cleaning up damaged photoreceptor membrane fragments–which photoreceptors shed constantly due to photooxidative damage.^[31]

In other words, the Cephalopodan “right way around” eye has some disadvantages, albeit ones that don’t matter much for short-lived underwater creatures that have little UV radiation to worry about.

To be clear, the exact orientation of the retina in both vertebrates and Cephalopods is still evolutionarily determined and probably not “reversible”–but the human eye is not as clear-cut bad as sometimes said.

Also, notice: these are the exact cells which have a receptor for L-Dopa, mentioned back in the footnote on Macular Degeneration. In Macular Degeneration, the RPE breaks down with the photoreceptors and the rest of the retina, presumably through some kind of long-term maintenance failure.

Tapetum Lucidum (glowy cat eyes)

In many animals, particularly nocturnal animals, the retina is further backstopped by layers of reflective cells, the tapetum lucidum, which reflect light that passes through the photoreceptors back, giving it a second chance to be captured. This does not work by simple mirror reflectance–the light rays need to go directly back, regardless of the angle of arrival. This makes it a retroreflector, and while not quite done perfectly, in animals something similar is typically achieved with cells filled with iridescent crystals, or with extracellular collagen–the diversity of mechanisms suggests it has evolved several times.

This is why cat eyes seem to glow when photographed at night.

Infrared and UV vision (visible light range isn’t that arbitrary, also snakes)

Now that I’ve talked about nocturnal vision and the tapetum lucidum, a question may scratch at the mind: Why not infrared vision? After all, this is how night vision goggles work, and the advantages of seeing better at night for many animals seem clear. Why are there no photoreceptors for infrared light?

Well there are, kind of. Three different groups of snakes have infrared sensing organs below their eyes–but this is not something that can be handled by normal eyes, because infrared radiation is absorbed heavily by water, also famously known as the substance eyes are mostly made of.

Light absorption by water as a function of Wavelength. Wikipedia again. Look, they have a lot of non-copyrighted charts.

In addition, hot bodies radiate heavily in infrared, meaning that any infrared detection device needs to solve the problem of not seeing its own heat.^[32]

The manner in which snakes have solved this is instructive:

In all three snake families, a heat-sensitive membrane is draped across pockets of air (with ventilation!), with the narrow hole in front serving as a pinhole camera of sorts. The air isolates the membrane from the animal’s own heat, and the air doesn’t absorb infrared like water would. Thus, this is a sort of primitive air-based eye.

Perhaps, given enough time, snakes might evolve some kind of infrared lens, but this would be a challenge, since it certainly couldn’t use the materials or processes of the usual lens. However, some kind of infrared reflector is potentially possible–infrared reflecting biological materials exist and reflectors have evolved in two closely-related species of fish.^[33]

As for the polar opposite sensing problem, ultraviolet–well, while many bees, hummingbirds, and insects see into the near-ultraviolet,^[34] anything further runs into the problem that ultraviolet radiation is highly photochemically damaging, damaging proteins and DNA, as anyone with a sunburn (including me right now, I blame the World Cup) will attest. Any cell designed to detect high ultraviolet would take constant, serious damage doing so, and a lens might as well be a cellular death ray.

This problem is serious enough that outside of animals with a clear incentive to do so (pollinators are motivated by the special UV markings on flowers), even near-UV vision is comparatively rare. It is also a comparatively minor component of natural sunlight, especially at the higher frequency bands that insects don’t see:

So, species without a need to see near-UV opt to save their photoreceptors and filter it out.

Saccades (more efficiency tricks)

The final topic of discussion is an important one I had difficulty fitting elsewhere: saccades, or eye movements. Both eyes automatically maintain focus on the same point, through means both conscious and unconscious.

For discussion here I’m more interested in the unconscious kind, in particular the extremely rapid, jerky, unconscious movements of the eye 3 to 4 times a second as it fixates on a scene or object.

If, like most humans, you have never noticed this happening, there is a very good reason for this–the visual system and brain go to great lengths to hide all visual evidence of this, essentially freezing the visual representation of the scene in place while the eye moves, especially the finer details. This is actually rather easy to demonstrate to yourself–simply move your eyes around and observe that the world doesn’t seem to blur or distort in any way.^[35] Then reflect that your eyes are doing this 3 to 4 times a second without you even noticing.

The obvious question here is: why these constant movements? It is not entirely known, but the most widely believed reason is essentially that it’s for the mitigation of photoreceptor latency and for retinal adaptation–by randomly shifting the image across different populations of photoreceptors, fresh photoreceptors and ganglionic neurons can be brought into line while others are resting. Neurons stop responding to persistent or repeated stimuli–a phenomenon known as adaptation, likely for energy-efficiency and information-efficiency reasons–and a constant input would eventually fade away.^[36] Direct forced stabilization of an image on the retina using specialized devices also causes images to fade.^[37]

Because of the importance placed on saccades being done as fast as possible, speed is highly prioritized–burst neurons in the brainstem wire directly to the optic motor neurons, and essentially only have an on-off switch, with no intermediate level of control. Again, this can be demonstrated to yourself fairly intuitively–just try to move your eyes slowly in one direction, instead of in jumps.

Teaser - Visual Processing in the Retina

This is not actually a full section–this topic is so massive it might occupy a future post on its own. Instead a simple teaser fact: There are approximately 100 million photoreceptors in the eye, but the optic nerve has only about one million fibers.

Somewhere in there, in the retina, information is being compressed massively–more so even than the fiber count suggests, with some estimates suggesting a three order-of-magnitude loss from ~1 GB/s of visual input to 1 MB/s down the optic nerve. The retina is doing a tremendous amount of image compression work before relaying anything to the brain.^[38]

Summary

Eyes are neat.

^{^}
If you rub your eyes too hard and too much, you will gradually damage your cornea, causing permanent blurry vision. Also, because LASIK surgery involves lasering your cornea, a bad rubbing habit can prevent you from being able to safely get LASIK surgery.
^{^}
While intuitively one would imagine that the “lens” does the focusing, and this is how it’s typically diagrammed, it’s actually the round shape of the cornea that performs about two-thirds of the refractive power of the eye. Hence why vision correction surgeries like LASIK target the cornea.
But while the cornea does most of the refraction, the lens is responsible for ensuring that the final image lands correctly and sharply on the retina.
^{^}
Though many studies find that the main reason is actually this: in the dark without a target to focus, the eye tends to drift to a sort of intermediate position that is effectively somewhat myopic; it is unclear how much the chromatic aberration effect contributes to this, and some sources claim it hardly matters.
^{^}
Not water, but the inner fluid of the eye, the aqueous humor. Water is a pretty good approximation though. (Somewhere there is an ophthalmologist very mad at me right now.)
^{^}
The vitreous humor. Water is an… okay approximation.
^{^}
Which places air in front of your eye, with front surfaces carefully curved to cancel out their own refraction.
^{^}
https://en.wikipedia.org/wiki/Thin_lens, which also has fun equations.
^{^}
Diving birds, which need to see in both air and water, can adjust the focus of their eyes to an extreme amount–three times or more compared to humans. Wikipedia has some numbers and sources–I’m not going to pretend I knew these numbers offhand or from the primary sources.
^{^}
Other examples include skin (where the outer layer is dead keratinocytes stuffed to the brim with keratin) and wood (mostly tubes of lignin and cellulose formed by once-living cells). Notable non-example: Bone.
^{^}
Normally I would put a cool picture of this below, but I can't find a good one, sorry. Instead please enjoy further reading on crystallins from the Protein Data Bank.
^{^}
When this property fails, the lens gets cloudy, perhaps to the point of unusability, which is the root cause of cataracts. The system simply can’t last forever–by the mid-to-late 70s, about half of Americans have some form of cataracts, though incidence rates vary by ethnicity and region, with South Asia and other high-UV equatorial regions having the highest rates.
^{^}
Alpha-crystallins seem to be descended from heat-shock proteins, produced when a cell detects it is overheating. This makes sense if you know that one of the big risks of high temperature is proteins losing structure–denaturing–and that heat-shock proteins frequently serve as structural support, reinforcing the existing structures of proteins. Exactly what you want in a lens gel too.
^{^}
The global prevalence of myopia is ~30% and climbing, with some projections to 50% by 2050. In East Asia in particular it is practically at universal levels, with as much as 90% of city-dwellers below a certain age myopic.
The topic of why myopia is so common is a fascinating subtopic of its own, about which surprisingly little is known. Epidemiological and historical study shows a clear strong correlation between myopia prevalence and urbanization and time spent indoors, but many of the most obvious explanations (time spent reading, for instance) show a surprisingly mixed or even no effect in studies. The only one that consistently holds up is time spent outdoors during childhood and consequent exposure to sunlight.
Vision is very good at correcting for differences in brightness, so the average person would be surprised to learn that natural sunlight varies from dozens to thousands of times brighter than indoor lighting. Light exposure is known to drive dopamine release in the retina (indeed, a gradual drop in dopamine over the course of the day causes gradual retinal adaptation to lower light levels), and there is an idea that long-term lack leads to inaccurate eye growth. But this has not been clearly shown in any study I am aware of.
Regardless, while earlier I praised the eye as a carefully tuned system under enormous selection pressure, here we may be seeing the system fall down under an outside-context problem–presumably, none of our pre-industrial ancestors ever had the problem of not enough sunlight in childhood.
^{^}
Also fun fact, newborns are fairly hyperopic (farsighted), and the eye has to elongate as they grow up to focus properly (also the lens changes a bunch, it’s actually more powerful as an infant to help make up for the fact that an infant’s eye is smaller).
^{^}
Here again, East Asians lose the genetic lottery, with an onset that is more like late 30s, though this may admittedly be environmental and related to latitude/temperature/UV.
^{^}
Or even better, the color spectral curve!
^{^}
The manner in which humans actually process color is more complicated, and contains things like color opposition. Moreover, both the L and M-opsin are encoded on the X chromosome; mutations on these can lead to color-blindness, most frequently in males. However, small variances in the detection spectrum between copies of the genes can lead to some women having four cone types, with two that are slightly different. There is some evidence this can lead to improved color perception.
^{^}
I have also seen it falsely asserted that cones are less sensitive because they are “paying more attention” to a specific part of the light spectrum, and ignoring other photons. A glance at the spectrum chart above rules this notion out: cones are not particularly narrow-banded in their frequency response compared to rods.
^{^}
In numbers, rods have a refresh rate of 5-10 Hz, whereas cones are about 50-60 Hz. Before the 120 Hz monitor people jump on me, note that this does not imply that humans cannot see faster than 60 Hz–the 60 Hz number comes from measurements of how fast you have to flicker a light before people start seeing it as continuous. But what exactly the system “sees” depends on the entire system, brain included, and often depends on what you’re measuring. Physiologically, cones seem to recover in 20-100 ms. People can notice the presence of a single image shown for as low as 13 ms. Sometimes things are seen without reaching full conscious attention. And finally, most relevant to the gaming monitor types, the brain can use intermediate sub-60 Hz frames to better track motion or detect anomalies.
There is a degree to which photoreceptors adjust their refresh rate to their frequency of stimulation, and a degree to which the underlying ganglionic neurons adjust the amount of temporal integration performed over the course of the day (based on dopamine levels, mediated by overall light levels).
^{^}
Stiles-Crawford Effect
^{^}
There is also melanopsin, which is expressed inside neurons of the retina that respond directly to light. This plays no role in vision, but does play a role in the circadian rhythm and biological clock, communicating with the suprachiasmatic nucleus. Melanopsin has a peak sensitivity around 480 nm, cyan.
^{^}
Small differences in the amino acids forming the retinal binding pocket affect the electrostatic environment of the molecule, shifting its light-absorbing spectrum. All opsins are descended from the ancestor protein by gene duplication.
^{^}
https://en.wikipedia.org/wiki/Conjugated_system#In_pigments
^{^}
As always, take these folksy descriptions of quantum mechanics with a grain of salt, and actually I’m pushing up against the limits of my own understanding here.
^{^}
It is perhaps unsurprising to learn that retinal, like chlorophyll, also exploits quantum coherence effects to achieve an unusually rapid and reliable conformation change. However, on its own, retinal only has a QE of 20-24%–the opsin protein holding it physically forces it into a tuned detection configuration.
^{^}
Source. Cones are more energy-intensive than rods, especially in the daytime where rods' trick comes into play, but also at night too.
^{^}
If you enjoy your Latin, the full names are the macula lutea and fovea centralis.
^{^}
This does indeed mean that the center of your vision is effectively blind at night-time. You can sometimes tell when star-gazing–dim stars are more visible in the peripheral vision.
^{^}
As you may have noticed, the brain does not perceive a blatant missing spot in the visual field, instead patching in the missing section from a combination of extrapolation and previous images of the same location. This is aided by the fact that the eye never really stays still. See Saccades, later.
^{^}
A fun non-eye example of this: https://en.wikipedia.org/wiki/Recurrent_laryngeal_nerve
^{^}
These pigment cells also seal off the retina from the choroid, preserving the eye’s immune privilege, the phenomenon wherein certain sensitive tissues (most notably the central nervous system) are isolated from the immune system and the potential damage it can inflict.
^{^}
Which is also why the James Webb Space Telescope’s most sensitive infrared sensor needs to be cooled to below 7 K.
^{^}
The Brownsnout Spookfish has a second eye that uses reflectors. See also: Rhynchohyalus.
^{^}
In fact, the photoreceptors of the Human eye are somewhat sensitive to near-ultraviolet, but the natural lens screens these out to protect the retina. Older artificial lens replacements sometimes didn’t, and individuals with such a lens (and the rare individual without a lens) sometimes reported seeing near-ultraviolet as a blue-white or purplish color. Newer lens replacements sensibly screen this out, depriving us of the posthuman experience of seeing brand new colors.
^{^}
Note that this kind of conscious eye movement is not the same as the unconscious movements 3-4 times a second, but it illustrates the point about the visual freezing.
^{^}
Troxler's fading, sick effect name honestly.
^{^}
https://pmc.ncbi.nlm.nih.gov/articles/PMC2951333/
^{^}
For comparison, an uncompressed 1080p, 24-bit, 30 fps video is about 1.5 Gbit/s, while with h.264 compression this is more like 8-10 Mbit/s.