A median-eye origin for vertebrate retinas remains a testable hypothesis
The proposal concerns the ancestry of cell types and circuits, not a literal replay of evolutionary history in every embryo.
AI-assisted, source-linked analysis. Product testing is only claimed when explicitly documented. How we work.
A Current Biology review proposes that the vertebrate retina arose by repurposing an ancestral median light-sensing system. George Kafetzis, Michael Bok, Tom Baden and Dan-Eric Nilsson combine comparative evidence into an evolutionary scenario, rather than reporting the discovery of a fossilized cyclops.
That distinction is central to the story. A hypothesis can make a useful connection between observations and generate new tests without becoming a directly observed historical sequence. The article's earlier version sometimes treated a proposed sequence as if embryonic development had already demonstrated it.
What the authors propose
The review describes the vertebrate retina as a circuit combining ciliary and rhabdomeric cell lineages. Its scenario follows loss of lateral eyes in an ancestral burrowing, suspension-feeding lineage, retention of median photoreceptors and later diversification into retinal and pineal systems.
The authors also propose evolutionary relationships for bipolar cells linking the retinal circuits. These relationships are part of the hypothesis being argued, not observations of ancestral cells preserved through time.
The primary article is explicitly labelled a review, with DOI 10.1016/j.cub.2025.12.028, and is available from the authors' laboratory. Reading that designation prevents a misleading summary in which the paper becomes an experiment that created eyes or a fossil discovery that identified the ancestor itself.
Cell identity is more informative than a nickname
Calling the proposal cyclops-like is an accessible image, but it can conceal the real explanatory question: how did different cell types and their connections come to form the vertebrate retinal circuit? A single visible eye is not a sufficient description of that molecular and anatomical problem.
The earlier article reduced the rhabdomeric contribution to a small non-image-forming residue. That was misleading. The review discusses ganglion, amacrine and horizontal cell lineages as part of the retinal circuit, not merely an isolated clock function.
This correction matters because an evolutionary explanation must account for the relationships among components. Replacing one oversimplified label with another can make a complicated proposal sound intuitive while removing the features it was designed to explain.
Development is evidence to interpret, not a historical recording
The old article leaned heavily on an embryonic eye field and developmental disorders to support a literal single-eye-to-two-eye history. It also added detailed gene and signalling claims without verifying each against the cited source.
A developmental observation and an ancestral reconstruction are different propositions. A process occurring in a living organism can constrain an evolutionary explanation without showing that every stage occurred as a free-living ancestral adult. Treating development as a recording of history skips that inferential step.
This revision therefore removes the clinical discussion of cyclopia and the unverified gene-by-gene narrative. They are unnecessary to explain the review's proposed relationship between ancient sensory systems and the modern retina.
Similarity needs a comparison that can distinguish alternatives
Two tissues sharing molecular features can suggest a relationship, but the interpretation still needs to address what kind of relationship is proposed. Common ancestry, reuse of an existing component and later functional similarity are different explanations that should not be silently merged.
A useful test asks what an alternative explanation would predict. If two explanations accommodate the same observation equally well, that observation alone cannot select between them. This is a general criterion for evaluating an inference, not a claim that the authors have ignored alternatives.
The authors describe comparisons of transcriptomic profiles across retinal, pineal and other median photoreceptor systems as a route to testing their hypotheses. The value of that proposal lies in the possibility of comparing predicted relationships with additional evidence, rather than declaring the story complete because it is coherent.
What the review does not establish
It does not identify a fossil individual as the exact common ancestor of modern vertebrate eyes. Nor does this article claim that the review settles every question about retinal inversion, every animal's eye or the function of a human pineal gland.
The older text expanded into whole-genome duplication claims, named experimental manipulations and broad assurances about what developmental evidence proves. Those additions have been removed unless required for the narrower, supported account.
The result is still a significant scientific idea: the retina may be better understood as a reorganized combination of pre-existing sensory circuits than as an entirely new invention. Presenting the claim at that level preserves both its explanatory ambition and the need for tests. Readers can appreciate the proposed evolutionary connection without being told that a vivid nickname is itself proof.
Sources
Primary review, author-hosted copy: https://badenlab.org/wp-content/uploads/2026/02/2026-Kafetzis-et-al-Current-Biology.pdf
DOI: https://doi.org/10.1016/j.cub.2025.12.028
Original reporting: https://arstechnica.com/science/2026/03/the-vertebrate-eye-may-have-begun-as-a-cyclops/