How to build an eye
June 22, 2025
Ingredients:
1 functioning universe (ours will do)
1 medium-sized star (such as the Sun)
1 rocky planet with a stable orbit in the habitable zone
Liquid water (your source of two of the CHNOPS), abundant and persistent
CHNOPS elements (Carbon, Hydrogen, Nitrogen, Oxygen, Phosphorus, Sulfur)
Natural laws (physics, chemistry, and, super importantly, evolution by natural selection)
Radiation (UV, cosmic, solar. Not too much, not too little)
Billions of years of uninterrupted time
How to cook:
1. Prepare the Universe.
First of all, we need a universe that hosts stars and celestial bodies such as planets. The one we have right now will suffice.
Let stars forge heavy elements in their cores. Allow for stellar deaths (especially supernovae) to scatter these elements into the cosmos. This cosmic debris will later accrete into rocky planets.
2. Select a planet in the habitable zone.
We need one planet to orbit a star at the right distance, during the right phase of the star’s life. Earth and the Sun make a good pair.
This planet must host persistent bodies of liquid water, ideally stabilized by a thick atmosphere and magnetic field to protect early organic chemistry from excessive radiation.
3. Add foundational chemistry.
Ensure the presence of the essential elements for life: C, H, N, O, P, and S. These elements support the formation of amino acids, nucleotides, and lipid membranes. Water provides a solvent for reactions and participates in many of them.
4. Set up biological laws.
Implement evolution by natural selection as a governing principle. Introduce the possibility of heritable variation. For instance, through replication errors (mutations), recombination, and others.
5. Apply radiation, entropy, and time.
Expose surface or near-surface molecules to selective doses of UV light, geothermal gradients, and chemical disequilibria. Let the chemistry run until you get self-assembly, self-replication, and metabolism.
A sign that it’s working: lipid bilayers enclosing RNA or DNA-like molecules. You now have the precursors to biological life.
6. Let cells emerge and diversify.
Over hundreds of millions of years, allow cells to evolve increasing diversity and complexity: DNA genomes, protein enzymes, ribosomes, membranes, and (optionally) nuclei.
Early photosynthetic organisms will evolve pigment-protein complexes (like bacteriorhodopsin or chlorophyll) to harvest photons. Some pigments, especially retinal-based ones, can isomerize in response to light. This will set up photoreception.
7. Encourage evolution of sensory photoreception.
Some cells will eventually evolve opsins, light-sensitive proteins bound to retinal. These opsins activate signal transduction pathways in response to photons. With selection favoring light-detecting behavior (e.g., moving toward or away from light), these systems evolve into dedicated photoreceptor cells.
8. Promote multicellularity and division of labor.
As multicellular organisms evolve, some cells specialize for sensing light. Simple animals like flatworms and cnidarians develop eyespots: clusters of photoreceptive cells often backed by pigment to block light from certain directions. This creates directional sensitivity.
9. Shape the geometry.
Natural selection now works on spatial arrangement. A depression in the eyespot enhances directionality. Deepen the depression, and you get a pinhole eye capable of forming rough images. This geometry emerges independently in several animal lineages.
10. Add optical components.
Once a pinhole exists, a transparent covering can emerge to protect it. Slight thickening of this covering, followed by fine-tuning its curvature and refractive index, produces a lens. The lens improves brightness and focus.
11. Refine neural circuitry.
What good does a complex light-sensing mechanism without processing power? Photoreceptors connect to nerve cells, and eventually to centralized processing units (ganglia, brains). These pathways extract features like contrast, motion, and orientation.
And voilá, habemus oculos.
You now have a functioning camera-like eye. This recipe has been independently cooked multiple times. Camera eyes evolved separately in vertebrates and cephalopods; compound eyes in arthropods. Evolution, it seems, finds similar solutions under similar conditions.
Now, let’s dive a bit into the nitty-gritty of the above steps, shall we?
I mentioned cnidarians in the repice, which might strike as odd to some of you. Not many of us think of jellyfishes and suddenly think of eyes. Personally, to start thinking of eyes is akin to thinking of the evolution of eyes. I cannot do it otherwise, it must be understood through macroevolutionary lenses (pun intended). Of course, this small text is not supposed to cover the evolution of eyes, though you are welcome to explore a bit of it in good papers like Schwab’s Keeler lecture of 2017. Truth is, eyes are a fascinating example of evolution in action. Take a quick look at the eyes below:
There’s eagles, chameleons, parrots, geckos, turtles, pythons, cockatoo, and more in the collage above. In a small sample, diversity blooms greatly. How remarkable is that? Now, let’s note that these are all vertebrate eyes. If we start moving around in the tree of life, landing in different groups, we will start to see that eyes only increase in their shapes and variety. Insects have eyes too, but they are called compound eyes. Instead of forming a cup, the light-sensing cells in insects protrude convexically, with a double layer of lens sitting on top of each other. That’s the ommatidium, and thousands upon thousands of these will then form the eye we see in insects.
What makes life see? It all revolves around the Sun
Perhaps we ought to take a step back, and think of things before eyes. Reflecting on the early evolution of life is similar to trying to put together a puzzle of millions of pieces with very tiny connections between them. An overwhelming task, so I will try to keep it very simple. At the cost of sounding pedantic, life started around 4 billion years ago (give or take a few hundred million years), and since then been set on a non-stop fight against entropy. Life takes energy and then uses such energy for its many processes, and even the assembly of a molecule of RNA roaming free in whatever goo was the first pond that sourced life, sets energy in motion. The most basic chemistry of life requires energy to push its reactions, facilitate the creation of its necessary components, which in turn basically keeps life living. Luckily, our planet sits in a very suitable place in the universe. It orbits a star at just the right distance, a star that keeps throwing these little packets of energy, the photons. Some of them at our planet.
So every incoming photon is going to add energy to the systems on Earth. One of the earliest biological process that emerged (now we are thinking well beyond individual molecules, but in cells) takes these little packets of energy and convert them into the type of energy that is required to fuel chemical reactions. Photosynthesis is then the most ingenious systems to help life obtain some of that much needed energy, but here, I’d like to focus on the precedent it sets. Photosynthesis is the process that illustrates how tightly life interacts with light(¡).
And sunlight spans a broad spectrum. At one end lie short wavelengths like gamma and X-rays — energetic enough to ionize atoms and create reactive free radicals. These radicals, in turn, disrupt cellular structures by stealing electrons from nearby molecules. On the other end are long wavelengths, such as radio waves, which carry too little energy to drive biological processes. The sweet spot lies in between: the visible spectrum — where light is energetic enough to excite electrons but not so much as to destroy molecules — making it ideal for life.
It may seem like a digression, but it’s not. What does sunlight and photosynthesis have to do with eyes? Quite a lot, in fact. We’ve just seen how not all wavelengths of light are suitable for life to use. Photosynthesis, for instance, only taps into a narrow slice of the visible spectrum. A pattern begins to emerge: although sunlight delivers energy across a vast range of wavelengths, only a fraction is biologically useful — and from that fraction, even less is actually utilized.
The most well-known pigment involved in this selective energy capture is chlorophyll, which absorbs primarily in the blue and red regions of the visible spectrum, leaving out much of the green (and then organisms with chlorophyll appear green to us). As photosynthetic systems evolved, nature found ways by which efficiency is improved. Among these were accessory pigments like the carotenoids — long chains of conjugated double bonds that give them their distinctive orange color. These molecules expanded the usable range of sunlight by absorbing in the blue-violet region, where chlorophyll is less effective, and safely transferring that energy into the photosynthetic machinery.
So without photosynthesis, there would be no evolutionary pressures driving and sustaining the production of carotenoids such as β-carotene.. No β-carotene, no retinal. No retinal, no opsins. No opsins, no fundamental units of the most basic structure of an eye. This is finally where the eye gets back to the spotlight of the article.
The first “eye” is no eye at all, but sets the tone of evolution
If even the most basic cellular forms of life use light as an energy source, then there’s already a foundation for a selective pressure — or more precisely, for an evolutionary advantage. Cells that are capable of moving toward more intense sources of light, such as the ocean surface compared to deeper waters, can activate their photosynthetic systems more efficiently. This gives them greater access to the biochemical products of photosynthesis, which in turn fuels growth and replication. In evolutionary terms, that’s success.
This sets the stage for the evolution of phototaxis: the ability to move in response to light.
The simplest known structures to accomplish this are found in organisms like bacteria, euglenids, and even some planarians. These are called eyespots. Far from what we think of as eyes, eyespots are clusters of light-sensitive proteins (such as flavoproteins and opsins) paired with a light-absorbing pigment. While they cannot form images the way our eyes do, they are enough to detect the presence of light — and for many organisms, that’s more than enough. Often reddish-orange in appearance, these spots are typically located near the flagellum, helping steer the organism toward light sources.
Exemplifying some other intermediate steps
Even chlorophyll alone can be understood as responsive to light and, in some organisms, may be sufficient to drive phototaxis. But not all light-sensitive molecules are benign — many classes of photoreceptors can generate reactive byproducts, potentially damaging the very cells that host them. Rhodopsins, however, proved to be a breakthrough: evolutionarily refined, they became one of the most efficient and stable photoreceptor proteins. Flavoproteins, particularly the group known as cryptochromes, come second in both prevalence and efficiency. As a result, most known photoreceptors across the tree of life fall into these two families: rhodopsins and flavin-based proteins.
Once cells capable of processing light began to cluster spatially — that is, localize into specific regions — the next evolutionary step was surprisingly simple: they caved in. Literally. It’s believed that at the dawn of true optical systems, these patches of light-sensitive cells depressed into a shallow cup. This concave arrangement allowed not only the already existing detection of light, but perception of its direction, laying the foundation for spatial vision.
To illustrate how different organisms have run with this innovation, let’s examine three animals and the types of eyes they’ve developed. Then, we’ll explore what cnidarians — some of the earliest diverging animals — have devised to sense and interpret the world around them.
Sponges have eyes for swimming
Sponges, as a group, lack many of the diagnostic features seen in upper Phyla. They have no mouth, no digestive tract, no eyes. These are animals that lack true tissues and organs. Instead, these are usually porous constructions composed of several distinct cell types (excluding sex-related cells), some of them being: the pinacocytes, resembling some sort of flat plate-like that covers sponge’s outer surface; the choanocytes, flagellated cells resembling the living extant choanoflagellates. These drive most of the water movement within the porous cavity of their bodies; and the amoeba-like cells lophocytes and amoebocytes, that the name already suggests to be amoeba-like. The function of these cells is mainly transporting nutrients and secreting the collagen that sustain their mesohyl (between pinacocyte and choanocyte layers).
They can reproduce asexually or sexually, where a free-swimming larvae is ultimately released in the body of water, and here’s where things get interesting: these larvae do not lack eye-like structures. Instead, they again show some cluster of light-sensing cells that guide phototaxis during that stage of life. These eyes are made of chryptochromes, without opsins at all! This class of flavoproteins is sensitive to blue light. In this way, sponges challenge the expectation that opsins are necessary for light-guided behavior. Even without true eyes or nervous systems, they demonstrate how light perception can evolve independently, through molecular pathways distinct from those used in most animals.
Before we think about what these eyes actually look like, a quick note on evolution. While it might seem that we’re moving from “simpler” systems to “more complex” ones, this is not a progression toward perfection. The eye is not a grand experiment, inching closer to the human eye as its ultimate goal.
Adding to any biological system is costly, in both time and resources. The more sophisticated a system becomes, the more molecular and developmental components must be orchestrated, like adding new instruments to a growing symphony. Some evolutionary steps are easier to achieve than others. For instance, adding a lens is, in some lineages, as accessible as adding a bass line to an ensemble of violins. But not every eye will evolve a lens.
That’s because evolution isn’t a quest for optimization. It’s the preservation of what works on average through the forwarding of successful alleles to the next generation. The preservation, in other words, of whatever allows an organism to survive and reproduce within the genetic constraints and environmental pressures it faces. We’ll see how this plays out in the various kinds of eyes that follow.
The Nautilus pinhole
The most logical concept of an eye in animals is then a small pit (we can call this “camera-like” eyes, too), and in this pit you place whatever photoreceptor and pigment you can squeeze in to some efficiency and form the retina (or the layer of photoreceptors and pigments). Light will hit the retina, and the animal will process light, but no image. Good enough for groups such as flatworms, green algae, sponge larvae, slugs and snails, and some bivalves.
This basic structure won’t even be able to process the direction of the source of light, but will be good enough to spot the intensity. Now, what happens if we squeeze the entry of this pit and make it very narrow?
Light now can be directionally processed. This is a huge step from the wide-opening eye pit! It is the organization you see in a few mollusks, abalones, and nautilus. This way, if a nervous system is present, some blurry image reconstruction is doable. The big advantage here, from an optics perspective, is that the narrow opening causes the incoming light to concentrate in one area relative to its source, recapitulating its source direction. Resolution of the image is directly related to the size of the hole, where larger will allow light from a single point hit multiple places in the back of the pit, causing blurrier images. Narrower holes will do the opposite, and crispier images are allowed.
However, if you need depth perception, you’re out of luck! If three objects are aligned in front of this eye, but at different distances, the eye won’t process anything but a juxtaposition of all three objects. Nevertheless, we are slowly moving from a cluster of cells to a pit, then a pit with a narrow entry to allow for localization of light.
Nautilus possess the most sophisticated version of the pinhole eye. Muntz & Raj (1984) measured the nautilus eyes and created a fascinatng model of what they could be able to see:
The nautilus did not change much over the course of hundreds of millions of years. It is what we call a “living fossil”. If eyes were akin to parties in an arms race towards creating decent resolution images of the world around it, then nautilus defies participation in it with extreme power. Whenever the pinhole eye emerged in this group, it stayed. The pinhole eye is not the greatest system ever: small fluctuations on the pinhole opening (the pupil) come at a great cost to retinal illuminance. Equally so, as we can see the model above, resolution frankly sucks!
Some mollusks, such as the octopus, some marine snails, and most subsequent Phyla will display a very elegant solution to the resolution problem.
What’s left for an eye
Let’s go back to the image of the different types of eyes in mollusks, and tell yourself, what differs from our human eyes? The most obvious difference is the cornea — or, in the case of some mollusks, the lack of one. Take the nautilus, for example: its eye has a pinhole opening rather than a lens, and its internal cavity is filled with seawater. There is no transparent “window” separating the outside environment from the inner eye. The same water flows both inside and out.
This system works well in aquatic environments, where the refractive index of water (1.33) closely matches that of biological tissue (eyes have 1.35 to 1.44). But imagine applying the same model to a human eye: with air as the external medium and fluid inside, the optical challenges would be enormous. Without a cornea or a refractive boundary, light would scatter inefficiently, and the image would blur. Our eye relies on the sharp interface between air and cornea to properly bend incoming light and form a clear image on the retina.
A small layer of transparent cells now allows for the fluid inside the eye cavity to be different than the exterior. Depending on what fluid we use, we will change the index of refraction (how much the ray of light changes it trajectory as it traverses from one medium to another) of the eye. This is yet another powerful step towards resolution. “Reverse engineering” our eyes show that all rays of light concentrate in one single point in the retina!
Apparently, however, the lens appear to be the first structure of the two to emerge. Some marine slugs, such as the members of the Onchidium genus, have two types of eyes: the stalk eye — composed of cornea, lens, and retina — and the dorsal eye. Curiously, the dorsal eye lacks a cornea, but contains lens. The lens in an eye are achieving the same thing as the cornea, only slightly different in its very specific purpose. Embriologically, it is thought that, in the process of curving the eye pit towards a pinhole, the ectoderm (the outer embryonic layer) hardens and detaches into a free-floating structure, giving rise to the lens. Biochemically, different things create these two structure: lens are composed of different materials in different groups. For example, trilobites are believed to have calcite lens, while extant invertebrates and vertebrates lens are composed by a variety of crystallins and heat-shock proteins. This is all to say that cornea and lenses are the taxon-specific latecomers to the building of an eye.
Evidence that there is no linearity in eye evolution
Early, I showed the Euglena eyespot. We can incorrectly assume that this is the best we will find in these unicellular eukaryotes, but evolution is in action all around, and all the time. We can see this in the dinoflagellate Nematodinium. This little protist has a proto-cornea, a proto-lens, and a pigment cup. It is as if certain steps in the building of an eye are universal. Maybe they are, as the pieces required to build an eye are doing other things in the cell and can be co-oped to this new function. It is a leading hypothesis to the opsin origins, as we saw before. In the case of these dinoflagellates, there’s evidence that the ocelloid (the eye) is composed of structures that came from symbiontic events with red algae, perhaps even other protists. Chloroplast is one such structure acquired by symbiosis. They might be able to switch between autotrophy (photosynthesis) and heterotrophy (predatory behavior) based on the environment cues. How they process lights (or image, if at all) is really a mystery. Nevertheless, this pattern of eye emergence across life is starting to be clear.
To put into perspective, we started by looking at the early stages of life on Earth. We move to the celled organisms, saw eyespots in Euglena, a camera-like eye in
The structure we see in Cnidarians
Finally, we will circle back to jellyfishes. Cnidarians are very old, 635-543 million-years old, to be precise. Around the same time, sponges and ctenophores showed up. These, alongside Placozoans, comprise the four major basal metazoans. One particular group of jellyfishes, the class Cubozoa, have what is called camera-style eyes. Anatomically, these cubozoans are indeed cube-like, with four sides, and each side has a structure called rhopalia. Rhopalia is basically the location in which six eyes can be found. Four of those eyes are analogous to the eyepits we were discussing previously, while two are camera-like with cornea, lens, and retina. Six eyes per side, to a total of 24 eyes per animal!
Scientists knew previously that the Medusazoan (those with medusa life stages) clade contains several classes with species that have eyes. While it is uncertain how images are processed in cnidarian nervous systems, eyes seems not to necessarily need a brain to process these external inputs. Some scientists argue that different eyes have different purpose (either looking up, which is especially good for discerning complex images, such as the composition of a mangrove forest, or down, to look at the underwater environment).
It is truly remarkable. These cnidarians have completely different regimes of development for each type of eye. The lens-containing eye development shows pretty much the same hypothesis we discussed previously being accurate: the lens cell invaginate at first, when there’s no differentiated cornea cells. Those cells then differentiate, overlaying the lens which is now globular in shape. The image below can speak way better than I can
Final Remarks
In this small glimpse at the evolution of the eye, we have celebrated diversity to the exponential. Multiple independent origins of the eye in different groups tell us that interacting with the world around us is a highly successful strategy for survival. By looking at different animals and the development of their optical systems, we can draft fairly accurate recipes for building an eye. Mind you, I have not dived into the genetics, cell signaling, or developmental pathways at all, as the rabbit hole would be insufferably deep.
But we did learn one thing: to build an eye, you first must have a planet at the right distance of the right star, at the right time. Everything else can then be recycled and optimized.
Notes:
¡ If you want to read more on the origins of photosynthesis, take a look at this paper.
Other references:
Jonasova & Kozmik (2008). Eye evolution: Lens and cornea as an upgrade of animal visual system.
Lamb (2011). Evolution of the Eye.
Berzins et al. (2021). Cnidaria.
Sandmann (2021). Diversity and origin of carotenoid biosynthesis: its history of coevolution towards plant photosynthesis.
Serb & Eernisse (2008). Charting Evolution’s Trajectory: Using Molluscan Eye Diversity to Understand Parallel and Convergent Evolution
Land & Fernald (1992).The Evolution of Eyes.
Nature (2012). Vision with no nervous system.