Templeton Prize goes to the palaeontologist who thinks minds are inevitable

Hallucigenia is worth pausing on, because it is the best joke in the story and it is at Conway Morris's expense. He named it in 1977, from the dreamlike look of the thing, and reconstructed it stilting along on seven pairs of rigid spines with a row of waving tentacles on its back. In 1991 Lars Ramsköld and Hou Xianguang turned it over: the tentacles were legs, the spines were armour, and a second row of legs had been hidden in the rock. The head and eyes were not settled until 2015, and the animal whose strangeness did most to persuade readers that the Cambrian was a lost world of failed experiments turned out to be a relative of the velvet worms, which are still with us.

Conway Morris's reply to Gould was The Crucible of Creation: The Burgess Shale and the Rise of Animals (Oxford University Press, 1998). Reviewing it in Science on 26 June 1998, the palaeontologist Michael Foote praised the anatomy and the ecology — nobody disputes Conway Morris on the animals — and then recorded, with visible irritation, that the book dismisses Gould's emphasis on contingency as "trivial." Foote thought that overstated. That verdict has held up: the descriptive work is admired, the polemic much less so.

Nobody disputes the convergence. The dispute is over what it licenses#

Evolutionary convergence is the independent arrival of the same solution in unrelated lineages. Conway Morris did not invent the examples; what he did was collect them at a scale nobody else had attempted, in Life's Solution: Inevitable Humans in a Lonely Universe (Cambridge University Press, 2003) and The Runes of Evolution (2015).

Eyes are the standard case. The figure people still cite is Salvini-Plawen and Mayr's, from 1977: somewhere between forty and sixty-five independent origins. Vertebrates and cephalopods both built a camera eye with a lens, an iris and a retina, and wired the retina the opposite way round from each other, which is what you would expect of two engineers solving the same optical problem without consulting. Powered flight arose separately in insects, pterosaurs, birds and bats. Echolocation arose in bats and in toothed whales, and the convergence reaches into the molecules: the hearing gene prestin carries parallel substitutions in both — though a 2013 Nature paper reporting genome-wide convergence in echolocators was contested two years later in Molecular Biology and Evolution, which found none. Agriculture, with cultivars, monocultures and crop disease, arose in fungus-farming attine ants, in macrotermitine termites and in ambrosia beetles, tens of millions of years before anyone in the Fertile Crescent thought of it. Minds of a sort have arisen in corvids, parrots, cetaceans, primates and, on an entirely separate branch of the animal tree, in cephalopods — the case Peter Godfrey-Smith made in Other Minds (2016).

Two complications sit awkwardly with the strong reading. The first is deep homology. Shubin, Tabin and Carroll, writing in Nature in 2009 (volume 457, pages 818–823), set out the evidence that apparently independent innovations are repeatedly built by redeploying the same ancient genetic toolkit. If forty origins of eyes all run through the same conserved regulatory genes, the word "independent" carries less weight than the argument needs, and convergence may be telling us less about the constraints of physics than about what happened to be lying around in the genome of the last common ancestor.

The second is that contingency has been demonstrated in a laboratory and convergence has not. In Richard Lenski's long-term experiment, twelve populations of Escherichia coli have been propagated in parallel since 1988. In one of them, and only one, the ability to metabolise citrate in oxygen appeared after more than thirty thousand generations. Blount, Borland and Lenski reported in PNAS in 2008 that replaying the population from frozen samples produced the innovation again only from later time points — it depended on earlier mutations that had no value at the time. That is Gould's thesis, tested, with the tape actually rewound. Jonathan Losos, whose Improbable Destinies (2017) is the fairest book-length treatment of the quarrel, concludes that convergence is common where selection pressures are strong and the options few, and that Gould was right about the rest.

Direction without a director is the hardest sentence in the argument#

Conway Morris's philosophical claim is that evolution is blind but channelled. Natural selection has no foresight; the possibilities it explores are nonetheless limited by physics, chemistry and the geometry of bodies, so that a search without a target still arrives repeatedly at the same small set of answers. Stated that way the thesis is unremarkable and probably true. The question is whether it bears the weight he puts on it.

He separates himself carefully from intelligent design, which he has criticised for years: the ID argument is that natural processes cannot generate complexity, and his is the opposite, that they generate the same complexity over and over. He is also a practising Christian who has not hidden what he takes the pattern to mean. He gave the 2005 Boyle Lecture at St Mary-le-Bow in the City of London — a series founded under Robert Boyle's will to defend Christianity — under the title "Darwin's Compass: How Evolution Discovers the Song of Creation," and delivered the Gifford Lectures at Edinburgh in 2007 under the same title. Convergence, he has said, points to "very intriguing deep structures in the universe — that this is a true Creation, with a capital C."

Jerry Coyne's objection is the sharpest on offer, and it owes nothing to theology. Convergence is an argument about traits that arose more than once; human-grade intelligence arose once. "Complex human intelligence — and certainly religious belief and practice — is not convergent on anything," Coyne wrote in 2012. "It is an evolutionary one-off, like the elephant's trunk." He had never understood, he added, how documenting convergences bears on the inevitability of a feature that appeared once.

The reply, made more carefully by others on Conway Morris's behalf than by him, is that the convergent target is not Homo sapiens but a capacity: large brains for body size, manipulative behaviour, social learning, flexible problem-solving. Those have appeared in lineages that separated hundreds of millions of years ago. Russell Powell's Contingency and Convergence: Toward a Cosmic Biology of Body and Mind (MIT Press, 2020) defends a qualified version of this while rejecting the theological gloss entirely; George McGhee's Convergent Evolution: Limited Forms Most Beautiful (MIT Press, 2011) makes the constraint argument in the language of morphospace, with no God in it anywhere. The biology stands or falls on the biology, and Conway Morris's churchgoing is a fact about him rather than an objection to his argument. It is fair to note that the conclusion is convenient for him; it is not fair to treat convenience as refutation.

Where the argument does overreach is in the astrobiology literature that has taken it up. Convergence on Earth tells us what happens when carbon chemistry meets liquid water on a rocky planet with this particular ancestry; it is a single run, and using it to firm up the terms of a Drake-style calculation is inference from one instance dressed as inference from many. Conway Morris knows this better than his borrowers do. "The almost-certainty of ET being out there means that something does not add up, and badly," he told Cambridge in July 2015. "We should not be alone, but we are." His formulation on the day of the prize was flatter and better: "Life may be a universal principle, but we can still be alone."

The machine-mind literature is circling the same question in different words#

Ask whether intelligence is a convergent solution and you are close to the question this site has been covering all year in another form: whether minds are the kind of thing many different physical systems can implement, or whether ours is one contingent arrangement of meat among unimaginably many. Eric Schwitzgebel and Jeremy Pober have argued that consciousness is substrate-flexible rather than carbon-specific; Schwitzgebel has also argued that we will be making consequential decisions about machine minds long before we know enough to make them well.

The parallel is real but it should not be pushed. Convergence is an empirical claim about what selection actually produces under a given chemistry; multiple realisability, in the form Hilary Putnam gave it in 1967, is a claim about what could in principle realise a given mental state. They can even pull against each other. If evolution keeps funnelling towards a narrow set of cognitive architectures because the physics permits few, that is mild evidence that minds are harder to build in arbitrary substrates than the functionalist picture suggests.

The prize deserves scrutiny rather than deference#

The John Templeton Foundation sits on an endowment of about $3.6 billion and spends a substantial part of it funding research at the join between science and religion. The standing criticism is that this money does not so much fund a question as underwrite an answer. Richard Dawkins has said the prize goes "usually to a scientist who is prepared to say something nice about religion." Coyne told Nature that the Foundation's purpose is "to eliminate the wall between religion and science, and to use science's prestige to validate religion." John Horgan, who took a Templeton-Cambridge fellowship and then wrote about it, said the programme favoured speakers already tilted theologically.

The Foundation's answer, put most effectively by Martin Rees — an atheist who took the prize in 2011 — is that the laureate list includes non-believers, and that work in psychology, evolutionary biology and economics honoured by the prize cannot reasonably be described as the promotion of religion. Rees's point stands. So does a smaller and more awkward one: three of Conway Morris's books, The Runes of Evolution among them, were published by the Foundation's own press. That does not make the convergence argument wrong; it does mean the prize is rewarding work it helped bring into print.

His scientific standing was established long before any of this: the Walcott Medal in 1987, election to the Royal Society in 1990 at thirty-nine, the Lyell Medal in 1998. No ceremony date or venue has been announced. His own comment on the award was characteristically indirect: "What a journey. As somebody once said — 'Be careful when you step onto the unending road.' A Ph.D. on fossil worms might logically lead to field-work in Greenland, but to an absorption with evolutionary convergence and thence the Fermi Paradox?"

Neither tape has been rewound. Gould's cannot be, and Lenski's flasks run at the scale of bacteria rather than brains. Until somebody finds a second biosphere, whether we were bound to happen remains the most interesting question in biology on which no experiment is possible — which is the sort of question the Templeton Prize exists to fund, and the reason it is worth arguing about who gets it.

Michael Sandel, Martha Nussbaum, Plato