Eyes are a primary organ of the human experience: Not only are we profoundly visual animals, but eyes are organs of expression, they are indicators of health and beauty, and we have a cultural tradition of regarding them as windows into the mind. We value eyes in other animals, too. If you have a pet, you know that one of the signs of its domesticity and your friendliness toward one another is a willingness to look into each other's eyes and acknowledging each other's gaze, while looking into the eyes of a wild animal is a challenge and a threat. The eyes that we like most are those most similar to our own, that share an evolutionary affinity with us, and the farther we move from our own lineage, the more disquieting the face and eyes of an animal can be. Eyes are actually remarkably diverse and, although one might think that visual function is fairly straightforward, and that there wouldn't be that many ways to put together a visual sensor, nature has done it.
While eyes are common in larger animal species, about a third of all animal phyla lack eyes altogether; sea urchins do not bother with them, nor do many worms. Another third have eyes that look rudimentary to us; spots and patches and pits that can sense whether it's night or day or whether a shadow is passing overhead, but that do not form any kind of image. The final third have true image-forming eyes that can capture a picture of what's going on around them and pass that on to some kind of brain or nerve net. The phyla that have true eyes are a diverse subset of the multicellular animals, including jellyfish and sea anemones, molluscs, annelid worms, onychophora (velvet worms), arthropods, and us chordates, which is a strange distribution. It's as if eyes popped up in scattered lineages interspersed with groups that lack them. For a long time, one of the hypotheses to explain all these eyes was that they evolved independently, multiple times within the animal kingdom.
If you look at the deep structure of eyes and their cellular components, that impression is reinforced. For instance, from the outside, octopus eyes look remarkably human. They are fluid-filled eyeballs with an iris, a pupil, and a lens. On the inside, they are fundamentally different. The light-sensitive layer, the retina, is inside out in humans relative to the octopus. Our photoreceptor cells all point to the back of the eye, while theirs point forward toward the lens. Octopus eyes also have all the nervous wiring exiting out the back, rather than being draped over the photoreceptors. They have an organization that is visually superior to ours, but the important point is that the structural details are so radically different that we know cephalopod eyes did not evolve from chordate eyes or vice versa. They evolved independently from simpler precursors along different lines, and the external similarities are a result of convergence, not evolutionary relationship.
The compound eyes of arthropods are even more radically different. Instead of a single lens focusing an image on an array of photoreceptors, they have compound eyes with multiple lenses, each focusing a part of the visual field on a small set of photoreceptors. Furthermore, the photoreceptors are a different kind of cell. We chordates have ciliary photoreceptors, where cells have modified a kind of motile appendage called a cilium into an antenna for collecting light. Arthropods have rhabdomeric photoreceptors, which instead fold up one side of the cell into deeply corrugated furrows to increase the surface area for collecting light. Ciliary and rhabdomeric photoreceptors also use G proteins—signal transduction proteins that activate enzymes in the cell in response to the reception of light and different opsins (a c-opsin vs. an r-opsin)—proteins that carry the photoreceptive pigments. Different cells, different arrangements, different optical elements, different proteins—eyes are overwhelming in their diversity.
To make matters worse, cephalopods with their human-looking eyes use rhabdomeric photoreceptors, like the ones used in the compound eyes of insects. At the same time, those cephalopod cousins, clams, have mantle eyes that use ciliary receptors. It begins to look like genetic chaos, as if animals just slapped together eyes with any old components on hand, and without much respect for any kind of evolutionary unity. Eyes must be incredibly easy to evolve, or we're missing some important unifying principle in their construction. The answer has been found by reaching farther and farther back into evolutionary history, using analysis of the diversity to find the core, common elements of eyes.
One clue can be found in the polychaete worm, Platynereis. Not all animals are limited to just two eyes, and Platynereis is typical. It has one set of eyes for the larval form, and another for the adult form. As an adult it has, in addition to a pair of lens-type eyes with rhabdomeric photoreceptors, another pair, of the ciliary type, embedded in its brain. It has both!
Lest you think this is an obscure condition found only in bizarre marine invertebrates, humans have been found to carry a vestige of a similar condition: Our photoreceptors use c-opsin, but we also have cells in our retinas that use the rhabdomeric form, r-opsin, thought to be important in light/dark detection for setting circadian rhythms, but which don't form images. This tells us something about the last common ancestor of animals—that it might possibly have had multiple kinds of receptors and eyes, and that what we observe in the diversity of extant eyes is not that it is easy to evolve an eye, but that it is easy to lose one or the other kind of eye in a lineage.
The key to figuring out the evolutionary relationships is to look at the most distantly related group in the set of eyed animals, which in this case is the cnidarians, jellyfish and anemones. Recent work by Plachetzki, Degnan, and Oakley has found that that clade have both the ciliary and rhabdomeric opsins, making the division into two kinds of photoreceptors an ancient one, occurring at least 600 million years ago.
This ancient animal probably had very simple eye spots with no image-forming ability, but still needed some diversity in eye function. It needed to be able to sense both slow, long-duration events such as the changing of day into night, and more rapid events, such as the shadow of a predator moving overhead. These two forms arose by a simple gene duplication event and concomitant specialization of association with specific G proteins, which has also been found to require relatively few amino acid changes. This simple molecular divergence has since proceeded by way of the progress of hundreds of millions of years and amplification of a cascade of small changes into the multitude of diverse forms we see now. There is a fundamental unity that arose early, but has been obscured by the accumulation of evolutionary change. Even the eyes of a scorpion carry an echo of our kinship, not in their superficial appearance, but deep down in the genes from which they are built.