Why vertebrate eyes are so different: A new 'cyclops ancestor' hypothesis

Background

Biologists have long marveled at how many ways evolution has solved the problem of seeing. Octopuses and humans both have camera-style eyes with a lens and retina, yet the two designs differ in critical ways: cephalopod retinas face the incoming light and don’t produce a blind spot, while vertebrate retinas are inverted, wiring sits on top of the light-sensitive cells, and the optic nerve exits through a literal hole in the retina. Arthropods, meanwhile, mostly rely on compound eyes built from many repeating facets.

The vertebrate eye is especially odd in molecular terms. Across the animal kingdom, light-sensing cells broadly fall into two lineages:

Rhabdomeric photoreceptors, which dominate vision in many invertebrates and use a Gq/PLC signaling cascade.
Ciliary photoreceptors, which in most non-vertebrates handle non-image tasks like circadian rhythms—but in vertebrates, they became the basis of rods and cones and use a Gi/Transducin cascade.

This “lineage switch”—using ciliary cells for image-forming vision—sets vertebrates apart. Add in the inverted retina and our characteristic blind spot, and an obvious question follows: why are vertebrate eyes so different from those of other animals?

A new hypothesis brings together developmental biology, comparative anatomy, and genomics to suggest a counterintuitive answer: our ancestors may have lost sophisticated lateral eyes and later rebuilt image-forming vision—starting from a single, median light-sensing organ that eventually split into a pair. In other words, vertebrate vision might trace back to a cyclops-like beginning.

What happened

The idea rests on several well-established pieces of evidence that only recently have been integrated into a single evolutionary story.

Embryos start with one eye field

Early in vertebrate development, the forebrain contains a unified “eye field.” A midline signal—largely Sonic hedgehog (SHH)—splits this field into two optic vesicles that later become the paired eyes. If that signaling fails, a rare birth defect called cyclopia can occur, in which the eyes do not separate. Developmentally, then, a single median eye is not an alien concept: it is the ground plan before the midline divides it.

Median light-sensing organs are common in vertebrates

Many vertebrates carry a light-detecting organ on the top of the head: the pineal or parietal complex (sometimes called a “third eye” in lizards). In some lineages this organ remains photosensitive and even lens-bearing; in others, like mammals, it has retreated into the skull as the hormone-secreting pineal gland that regulates daily and seasonal rhythms. The existence of a median photoreceptive system indicates that chordates have long harbored midline light detectors separate from the lateral eyes.

The molecular toolkit points to a ciliary origin

Vertebrate rods and cones are ciliary photoreceptors. Their development depends on a familiar set of transcription factors (Pax6, Six3, Rx/Rax, Otx2, Crx, Vsx2) and an opsin/transduction cascade distinct from invertebrate visual pathways. Intriguingly, median photoreceptive organs in non-vertebrate chordates—like the ocellus of tunicate larvae or the frontal eye of amphioxus—express overlapping gene networks and c-opsins that resemble vertebrate rods/cones more than invertebrate rhabdomeric eye cells.

Chordate relatives show reduced lateral eyes—and tiny midline organs

The closest living relatives of vertebrates, tunicates and amphioxus, have light-sensing structures but not the elaborate paired camera eyes of jawed vertebrates. Tunicate tadpoles have a single pigmented ocellus that helps them orient to light before they settle; amphioxus carries a small frontal eye at the midline, plus other photoreceptors along its nerve cord. These structures are not obviously image-forming cameras, but they share developmental gene signatures with vertebrate retinas.

Transitional vertebrates carry mixed signals

Jawless vertebrates underscore the story. Lampreys possess fully formed lateral eyes with layered retinas and a pineal/parapineal complex on the midline. Hagfish, by contrast, have rudimentary, skin-covered photoreceptive organs that look like eyes in decline. Together they hint at a complicated early vertebrate history in which light detection existed in multiple layouts and levels of complexity.

Putting this all together, the new hypothesis sketches a scenario like this:

Stage A: An ancient bilaterian ancestor had rhabdomeric visual organs on the sides of the head and separate ciliary photoreceptors for non-visual light sensing.
Stage B: A chordate lineage passed through a lifestyle bottleneck—think a tunicate-like, filter-feeding adult—that shed large, lateral, image-forming eyes as unnecessary. A small median ocellus or photoreceptive field remained for orienting to day-night cycles.
Stage C: Early vertebrates re-entered active predation during the Cambrian, needing high-acuity vision. They elaborated the surviving median, ciliary-based photoreceptor system into an image-forming organ—a primitive camera eye sited on the midline.
Stage D: Developmental patterning (notably SHH) split the single eye field into two, creating paired lateral eyes. Subsequent innovations—lens focusing, layered interneuron circuits, and the vertebrate-specific phototransduction toolkit expanded by early whole-genome duplications—yielded the camera eyes we recognize today.

This “cyclops-first, then split” proposal would elegantly explain three longstanding puzzles:

Why vertebrate vision uses ciliary photoreceptors: because the system was rebuilt from a ciliary median organ rather than inherited from rhabdomeric lateral eyes.
Why the retina is inverted: because the developmental geometry of the median forebrain photoreceptor field and its evagination produces an inside-out layering once the eye field balloons outward to form an optic cup.
Why a median photoreceptive system (pineal/parietal) persists in many vertebrates: it is a vestige of the ancestral midline photoreceptor domain that seeded the paired eyes.

Crucially, the hypothesis does not deny multiple origins of eyes across animals; rather, it proposes that vertebrate eyes are a special case of secondary invention—recruiting a different photoreceptor lineage and developmental pathway than those used by most invertebrates. It also fits comfortably with known gene regulatory networks: the same eye-building genes crop up in diverse organisms, but the wiring of those networks—and the cell types they assemble—can shift over evolutionary time.

Evidence threads in more detail

Developmental echoes: The single early eye field and the cyclopia defect show that a midline-to-paired transformation is a natural feature of vertebrate development. Evo-devo often treats such embryonic stages as historical fingerprints.
Gene expression homologies: Tunicate ocelli and amphioxus frontal eyes express Pax6, Rx, c-opsins, and other markers in patterns suggestive of homology with vertebrate retinas, though their anatomy is much simpler.
Opsin repertoire shifts: Vertebrates carry multiple c-opsins for rods and cones and a small set of rhabdomeric opsins in intrinsically photosensitive retinal ganglion cells (ipRGCs), which govern pupil responses and circadian timing. That inversion—rhabdomeric cells relegated to non-image roles—aligns with a scenario in which a ciliary system took over vision.
Genomic expansions: Two rounds of whole-genome duplication in early vertebrates (the “2R” hypothesis) likely fueled the elaboration of lenses, phototransduction cascades, and retinal interneuron diversity, helping a once-modest organ scale into a sophisticated camera.
Comparative morphology: The pineal/parietal organ’s variable complexity across vertebrates—from a true, lens-bearing photoreceptive organ in some lizards and lampreys to a hormone gland in mammals—maps neatly onto a story of an ancient midline photoreceptor domain with lineage-specific outcomes.

Caveats and open questions

Fossils are sparse: Soft tissues like eyes rarely fossilize, so we often rely on indirect clues. That leaves room for alternative scenarios.
Median-to-lateral mapping is not straightforward: Demonstrating that a midline organ directly seeded the paired eyes requires more than shared genes; it needs precise, testable homologies of tissue origins and circuitry.
Convergence vs. homology: Similar gene networks can be re-used independently. It remains possible that vertebrate eyes co-opted ciliary photoreceptors in parallel without a true ancestral median camera.
Life history inference: The idea that a tunicate-like ancestor lost sophisticated eyes is plausible, but reconstructing ancient ecologies is tricky. Multiple bouts of simplification and re-elaboration may have occurred.

Even so, the cyclops-first hypothesis integrates many otherwise disconnected facts. It offers a coherent narrative for why vertebrate eyes look and work the way they do.

Key takeaways

Vertebrate eyes differ profoundly from most animal eyes: they use ciliary photoreceptors, have an inverted retina, and create a blind spot.
A new evolutionary scenario proposes that our distant ancestors rebuilt image-forming vision from a single median photoreceptor organ, which later split into a pair during development.
Developmental biology provides a key clue: embryos begin with one eye field that SHH signaling separates into two—failures of this process yield cyclopia.
Comparative chordate biology supports a midline origin: tunicates and amphioxus possess small median photoreceptive organs with gene signatures similar to vertebrate retinas; many vertebrates retain a light-sensing pineal/parietal complex.
Genomic and molecular evidence (opsins, transcription factors, whole-genome duplications) fits a story of secondary elaboration and diversification of a ciliary-based visual system.
The hypothesis is testable with modern tools: single-cell atlases, CRISPR lineage tracing in amphioxus and tunicates, and functional circuit analyses in lampreys and lizards.

What to watch next

Single-cell and spatial atlases across chordates: Mapping photoreceptor types and developmental trajectories in tunicates, amphioxus, lampreys, and jawed vertebrates could reveal whether the same cell lineages truly underlie median organs and lateral eyes.
Gene regulatory network perturbations: Targeted manipulations of Pax6, Rx, Six3, SHH, and opsins in non-vertebrate chordates could test whether a median field can be coaxed into a camera-like organ.
Comparative connectomics: Do circuits downstream of pineal/parietal organs and early retinal circuits share organizational motifs that imply a common origin?
Developmental geometry: High-resolution imaging of optic cup formation may show how a midline photoreceptor sheet could generate an inverted, layered retina after evagination.
Lamprey and hagfish models: As survivors of early vertebrate lineages, they can illuminate transitional states of eye complexity and the relationship between median and lateral photoreception.
Environmental drivers: Paleoecological work on Cambrian food webs could clarify the selective pressures that might have favored rapid re-elaboration of visual systems in early vertebrates.

FAQ

What is an “inverted retina,” and why does it matter?
The vertebrate retina places photoreceptors behind layers of neurons and blood vessels, so light passes through wiring before hitting rods and cones. This arrangement creates a blind spot where the optic nerve exits. It’s functionally fine (brains fill in the gap), but it’s a distinctive architectural choice compared to cephalopods, where wiring sits behind the photoreceptors.
What is cyclopia in development?
Cyclopia is a rare developmental defect in which the embryonic eye field fails to split into two, often due to disrupted SHH signaling. Its existence shows that vertebrate embryos pass through a single-eye-field stage, which the new hypothesis leverages as an evolutionary clue.
Do humans have a “third eye”?
Not externally. Our pineal gland is buried deep in the brain and no longer forms an image. It still responds indirectly to light via retinal input and helps regulate daily rhythms through melatonin secretion.
Didn’t eyes evolve many times independently?
Yes. Evolution has repeatedly produced light-sensing organs—from simple eyespots to camera eyes and compound eyes—using different building blocks. The new hypothesis concerns specifically how vertebrate camera eyes arose and why they differ from other lineages.
What are ciliary and rhabdomeric photoreceptors?
They’re two ancient photoreceptor cell lineages that use different signaling molecules and cell architectures. Vertebrate rods and cones are ciliary; many invertebrate visual systems rely on rhabdomeric photoreceptors. Vertebrates retain a few rhabdomeric cells (ipRGCs) for non-image tasks like circadian entrainment.
How could a median eye turn into two eyes?
Development offers a template: a single eye field forms first and then splits left-right under midline signals, including SHH. Over evolutionary time, a median photoreceptor domain could have been elaborated into a camera-like structure and then patterned into bilateral organs.
Is there direct fossil evidence for a cyclops ancestor?
Not for soft tissues like eyes. The case is built from comparative anatomy, genetics, embryology, and physiology. Future molecular and developmental studies in living basal chordates can provide stronger tests.

Source & original reading

Original article: https://arstechnica.com/science/2026/03/the-vertebrate-eye-may-have-begun-as-a-cyclops/

Did vertebrate vision start with a single cyclops-like eye? A new hypothesis for why our eyes are so different