Arriving Twice — Convergence and the Shape of the Possible

An octopus and a human being last shared an ancestor somewhere in the region of six hundred million years ago. That ancestor had, at most, a patch of light-sensitive cells. It had no lens, no iris, no retina, no chamber. Everything that makes an eye an eye was built twice, separately, by lineages that had already parted.

And the two results are the same device. A transparent lens focusing an image onto a sheet of photoreceptors at the back of a fluid-filled chamber, with an adjustable aperture in front. Not similar in spirit — the same optical solution, arrived at from opposite directions.

The octopus version is, on one point, better engineered. In the vertebrate retina the photoreceptors face backwards, away from the light, so the nerve fibres run across the front of the retina and must exit through it, leaving a hole in the visual field. You have a blind spot; your brain conceals it from you. The cephalopod retina is the right way round and has no such hole. Two independent solutions to the same problem, and the one that came out cleaner belongs to the mollusc.

Eyes of some kind have arisen independently at least forty times. That number gets quoted loosely, so hold it carefully — we will come back to what "independently" is doing in that sentence, because it is the weakest joint in the argument and it deserves attention rather than concealment.

The Same Protein, Twice

Resemblance in an organ can be argued about. Resemblance in a molecule is harder to wave away.

Bats and toothed whales both echolocate. They are both mammals, but their common ancestor did not echolocate, and the two lineages developed the ability separately — one in air, one in water, with nothing in common but the physics of sound and the problem of hunting where you cannot see.

Echolocation requires exquisitely sensitive high-frequency hearing, and high-frequency hearing in mammals depends on a protein called prestin, which changes the length of the hair cells in the inner ear. When the prestin genes of echolocating bats and echolocating whales were compared, they were found to have converged — not merely in function but in the specific amino acid substitutions, at the same sites, in animals separated by the whole span of mammalian evolution and by the boundary between air and water.

What that means and what it does not

It does not mean the mutations were directed. Each substitution arose as substitutions do, and was kept or lost by selection. What it means is that the space of ways to build a high-frequency-sensitive prestin is small — small enough that two lineages searching it at random, sixty million years apart, landed on the same residues. The randomness is real and the outcome is still nearly forced. That combination is the entire subject of this book.

Crabs, Repeatedly

There is a joke among biologists that everything eventually becomes a crab. It is not much of a joke, because it is roughly true. The crab body plan — flattened, broad carapace, abdomen tucked underneath, walking sideways — has evolved at least five separate times from ancestors that were not crabs. King crabs are descended from hermit crabs. Porcelain crabs are squat lobsters. The process has a name, coined in 1916: carcinisation.

The same pattern runs through the rest of the record once you look for it. C₄ photosynthesis — a substantial biochemical rebuild of how a plant fixes carbon — has arisen independently something over sixty times. Powered flight, four. The sabre-tooth morphology appeared in the placental Smilodon of North America and again in Thylacosmilus, a metatherian in South America, on a continent the other never reached, with the two lineages separated by more than a hundred million years of divergence. Anteater morphology, several times. Streamlined fusiform bodies in fish, ichthyosaurs and dolphins — a reptile, a fish and a mammal converging on the shape that water permits.

The Tape

In 1989 Stephen Jay Gould proposed a thought experiment that has organised the argument ever since. Rewind the tape of life to the Cambrian and play it again. What comes out? Gould's answer, in Wonderful Life, was: something unrecognisable. The history of life is a history of accidents — which lineages happened to survive a mass extinction, which happened to be in the wrong ocean at the wrong moment. Replay it and intelligence, vertebrates, perhaps animals themselves might simply not appear. We are here because of a chain of contingencies, not because anything was aiming at us.

In 2003 Simon Conway Morris answered, in Life's Solution, with the evidence above. Replay the tape, he argued, and you get something remarkably similar. Not the same species — but eyes, and flight, and echolocation, and sociality, and probably intelligence, because those are not accidents. They are the small number of workable answers to problems the physical world keeps setting. The tape can be replayed and the destinations are constrained even when the route is not.

Both men had the same data. They were, on inspection, answering different questions.

The reading this book takes

Gould is right about lineages. Which clade survives the asteroid is contingent; nothing about mammalian biology earned the Cretaceous impact, and a different Tuesday produces a different inheritor. Conway Morris is right about forms. Whichever lineage inherits, the animal that ends up hunting in the dark will echolocate, and the one that ends up hunting by sight will build a camera eye, because those are the solutions available.

Contingency chooses the runner. Constraint chooses the shape. The two claims were never in competition, and the appearance of a fight between them comes from a single ambiguity in the question — whether "what would happen" means who or what.

Why This Is Not the Old Error

A reader who knows the history will already be uneasy, and should be. Convergence was once the central exhibit for orthogenesis: the doctrine, respectable until the middle of the last century, that evolution proceeds along predetermined lines driven by some internal tendency. The repeated appearance of the same forms looked like proof of an inner directive. The Modern Synthesis destroyed the doctrine, and rightly.

The distinction is not subtle and everything here depends on it. Orthogenesis located the direction inside the organism — a drive, a momentum, a tendency toward a form. Nothing in the evidence above requires or permits that. The bat's prestin mutations were not aimed at echolocation; they occurred, and almost all of their siblings were discarded. What is doing the directing is entirely outside: the physics of sound, the optics of a lens, the mechanics of a carapace, the chemistry of carbon fixation. The organism proposes at random. The world disposes, and the world's dispositions are few.

That is why the observation survived the death of the theory. Nineteenth-century naturalists saw a real pattern and attributed it to the wrong cause. Rejecting the cause did not oblige anyone to stop seeing the pattern — though for several decades it very nearly did, which is its own lesson about how disciplines handle a discredited idea.

The Weak Joint

Now the part that ought to trouble you, because it troubles the specialists.

Every number in this chapter — forty eyes, sixty origins of C₄, five crabs — depends on a judgement about what counts as independent. And the judgement is not always clean. The eyes of a fly, a squid and a mouse all deploy the same deep regulatory machinery: the Pax6 gene and its relatives switch on eye development across the animal kingdom. So the organs were built separately while drawing on a genetic toolkit inherited from a common ancestor. Is that convergence, or is it the same latent instruction expressed twice?

[OPEN] · what would sharpen this

The honest answer is that "independent origin" is a threshold, not a fact, and different specialists set it differently. Deep homology — shared developmental machinery underneath convergent structures — means some of the counting in this chapter is a matter of where you draw a line. That does not dissolve the pattern: prestin's substitutions are convergence at the sequence level with no shared latent instruction to appeal to, and carcinisation is a body plan, not a switch. But a reader who wants to know how strong the case is should know that its headline numbers are softer than they sound, and that this book has not done the work of re-deriving them. Recorded as open.

The Door This Opens

Return to Pirangi. One organism, growing outward without aim, arriving at the form of a forest — not because a forest was the goal but because that is what a tree becomes when its branches are heavy and the ground is close and the soil will take a root. The shape was in the situation, not in the seed.

The claim of this book is that the same thing is true one level up, in the history of what people have thought. If a structure is the only place the reasoning can go, then unconnected traditions will arrive at it separately, and the fact that they did is not evidence of contact between them. It is evidence about the structure.

Which makes the gallery of this book something other than a courtesy. Sixteen figures, across traditions that mostly could not read one another, converging on the same small set of ratios and the same ladder. Read as a list of precursors, it is a gesture of respect. Read as convergence data, it is the same argument as the octopus eye, in a different substrate — and it is testable in the same way, by asking whether the arrivals are genuinely independent or whether a channel of transmission explains them. That question is live for every figure in the gallery, and it is asked chapter by chapter rather than assumed.

The formal machinery underneath all of this — what it means for a system to have a small number of stable destinations, and how one proves that a ladder of forms converges — is the subject of Chapter Five, and of the mathematics the rest of this series reports. The route down begins at the Ladder of Ascent; the route up begins at τ = 2. This chapter is the evidence that something is there to climb.

Sources: Stephen Jay Gould, Wonderful Life (Norton, 1989); Simon Conway Morris, Life's Solution (Cambridge University Press, 2003); on prestin convergence, Y. Li and colleagues and separately Z. Liu and colleagues, both in Current Biology (2010), and J. Parker and colleagues, Nature (2013), on genome-wide convergence in echolocating mammals; on carcinisation, L. A. Borradaile (1916) for the term and the anomuran literature since; on C₄ origins, R. F. Sage's surveys; on deep homology and Pax6, the developmental-genetics literature following Walter Gehring's 1990s work.

Next: if the same structure is found twice in bodies, is it also found twice in minds? In the Air takes the convergence test to the history of ideas — Darwin and Wallace, Newton and Leibniz, and the 264 catalogued cases — and turns it on the gallery of this book.