Science in the Soul

by Richard Dawkins

  I also wonder whether an injustice is being done to Darwin, owing to this same failure of critics to come to grips with the distinction between DC-8 and 747 saltation. It is frequently alleged that Darwin was wedded to gradualism, and therefore that, if some form of evolution by jerks is proved, Darwin will have been shown wrong. This is undoubtedly the reason for the ballyhoo and publicity that has attended the theory of punctuated equilibrium. But was Darwin really opposed to all jerks? Or was he, as I suspect, strongly opposed only to 747 saltation?

  As we have already seen, punctuated equilibrium has nothing to do with saltation, but anyway I think it is not at all clear that, as is often alleged, Darwin would have been discomfited by punctuationist interpretations of the fossil record. The following passage, from later editions of the Origin, sounds like something from a current issue of Paleobiology: ‘The periods during which species have been undergoing modification, though very long as measured by years, have probably been short in comparison with the periods during which these same species remained without undergoing any change.’ I believe we can reach a better understanding of Darwin’s general gradualistic bias if we invoke the distinction between 747 and DC-8 saltation.

  Perhaps part of the problem is that Darwin himself did not have the distinction. In some anti-saltation passages it seems to be DC-8 saltation that he has in mind. But on those occasions he does not seem to feel very strongly about it: ‘About sudden jumps,’ he wrote in a letter in 1860, ‘I have no objection to them – they would aid me in some cases. All I can say is, that I went into the subject and found no evidence to make me believe in jumps [as a source of new species] and a good deal pointing in the other direction.’ This does not sound like a man fervently opposed, in principle, to sudden jumps. And of course there is no reason why he should have been fervently opposed, if he only had DC-8 saltations in mind.

  But at other times he really is pretty fervent, and on those occasions, I suggest, he is thinking of 747 saltation. As the historian Neal Gillespie puts it: ‘For Darwin, monstrous births, a doctrine favored by Chambers, Owen, Argyll, Mivart, and others, from clear theological as well as scientific motives, as an explanation of how new species, or even higher taxa, had developed, was no better than a miracle: “it leaves the case of the co-adaptation of organic beings to each other and to their physical conditions of life, untouched and unexplained”. It was “no explanation” at all, of no more scientific value than creation “from the dust of the earth” ’.

  Darwin’s hostility to monstrous saltation, then, makes sense if we assume that he was thinking in terms of 747 saltation – the sudden invention of new adaptive complexity. It is highly likely that that is what he was thinking of, because that is exactly what many of his opponents had in mind. Saltationists such as the Duke of Argyll (though presumably not Huxley) wanted to believe in 747 saltation, precisely because it did demand supernatural intervention. Darwin did not believe in it, for exactly the same reason.

  I think this approach provides us with the only sensible reading of Darwin’s well-known remark that ‘if it could be demonstrated that any complex organ existed, which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down’. That is not a plea for gradualism, as a modern paleobiologist uses the term. Darwin’s theory is falsifiable, but he was much too wise to make his theory that easy to falsify! Why on earth should Darwin have committed himself to such an arbitrarily restrictive version of evolution, a version that positively invites falsification? I think it is clear that he didn’t. His use of the term ‘complex’ seems to me to be clinching. Gould describes this passage from Darwin as ‘clearly invalid’. So it is invalid if the alternative to slight modifications is seen as DC-8 saltation. But if the alternative is seen as 747 saltation, Darwin’s remark is valid and very wise. His theory is indeed falsifiable, and in the passage quoted he puts his finger on one way in which it might be falsified.

  There are two kinds of imaginable saltation, then, DC-8 saltation and 747 saltation. DC-8 saltation is perfectly possible, undoubtedly happens in the laboratory and the farmyard, and may have made occasional contributions to evolution.*16 747 saltation is statistically ruled out unless there is supernatural intervention. In Darwin’s own time, proponents and opponents of saltation often had 747 saltation in mind, because they believed in – or were arguing against – divine intervention. Darwin was hostile to (747) saltation, because he correctly saw natural selection as an alternative to the miraculous as an explanation for adaptive complexity. Nowadays saltation either means punctuation (which isn’t saltation at all) or DC-8 saltation, neither of which Darwin would have strong objections to in principle, merely doubts about the facts. In the modern context, therefore, I do not think Darwin should be labelled a strong gradualist. In the modern context, I suspect that he would be rather open-minded.

  It is in the anti-747 sense that Darwin was a passionate gradualist, and it is in the same sense that we must all be gradualists, not just with respect to life on Earth, but with respect to life all over the universe. Gradualism in this sense is essentially synonymous with evolution. The sense in which we may be non-gradualists is a much less radical, although still quite interesting, sense. The theory of evolution by jerks has been hailed on television and elsewhere as radical and revolutionary, a paradigm shift. There is, indeed, an interpretation of it which is revolutionary, but that interpretation (the 747 macro-mutation version) is certainly wrong, and is apparently not held by its original proponents. The sense in which the theory might be right is not particularly revolutionary. In this field you may choose your jerks so as to be revolutionary, or so as to be correct, but not both.

  Theory 5. Random evolution

  Various members of this family of theories have been in vogue at various times. The ‘mutationists’ of the early part of the twentieth century – de Vries, W. Bateson and their colleagues – believed that selection served only to weed out deleterious freaks, and that the real driving force in evolution was mutation pressure. Unless you believe mutations are directed by some mysterious life force, it is sufficiently obvious that you can be a mutationist only if you forget about adaptive complexity – forget, in other words, most of the consequences of evolution that are of any interest! For historians there remains the baffling enigma of how such distinguished biologists as de Vries, W. Bateson and T. H. Morgan could rest satisfied with such a crassly inadequate theory. It is not enough to say that de Vries’s view was blinkered by his working only on the evening primrose. He only had to look at the adaptive complexity in his own body to see that ‘mutationism’ was not just a wrong theory; it was an obvious non-starter.

  These post-Darwinian mutationists were also saltationists and anti-gradualists, and Mayr treats them under that heading, but the aspect of their view that I am criticizing here is more fundamental. It appears that they actually thought that mutation, on its own without selection, was sufficient to explain evolution. This could not be so on any non-mystical view of mutation, whether gradualist or saltationist. If mutation is undirected, it is clearly unable to explain the adaptive directions of evolution. If mutation is directed in adaptive ways we are entitled to ask how this comes about. At least Lamarck’s principle of use and disuse makes a valiant attempt at explaining how variation might be directed towards improvement. The ‘mutationists’ didn’t even seem to see that there was a problem, possibly because they underrated the importance of adaptation – and they were not the last to do so. The irony with which we must now read W. Bateson’s dismissal of Darwin is almost painful: ‘The transformation of masses of populations by imperceptible steps guided by selection is, as most of us now see, so inapplicable to the fact that we can only marvel…at the want of penetration displayed by the advocates of such a proposition.’

  Nowadays some population geneticists describe themselves as supporters of ‘non-Darwinian evolution’. They believe that a substantial number of the gene replacements that occ
ur in evolution are non-adaptive substitutions of alleles, whose effects are indifferent relative to one another. This may well be true. But it obviously has nothing whatever to contribute to solving the problem of the evolution of adaptive complexity. Modern advocates of neutralism admit that their theory cannot account for adaptation.

  The phrase ‘random genetic drift’ is often associated with the name of Sewall Wright, but Wright’s conception of the relationship between random drift and adaptation is altogether subtler than the others I have mentioned. Wright does not belong in Mayr’s fifth category, for he clearly sees selection as the driving force of adaptive evolution. Random drift may make it easier for selection to do its job by assisting the escape from local optima, but it is still selection that is determining the rise of adaptive complexity.*17

  Recently paleontologists have come up with fascinating results when they perform computer simulations of ‘random phylogenies’. These random walks through evolutionary time produce trends that look uncannily like real ones, and it is disquietingly easy, and tempting, to read into the random phylogenies apparently adaptive trends which, however, are not there. But this does not mean that we can admit random drift as an explanation of real adaptive trends. What it might mean is that some of us have been too facile and gullible in what we think are adaptive trends. That does not alter the fact that there are some trends that really are adaptive – even if we don’t always identify them correctly in practice – and those real adaptive trends can’t be produced by random drift. They must be produced by some non-random force, presumably selection.

  So, finally, we arrive at the sixth of Mayr’s theories of evolution.

  Theory 6. Direction (order) imposed on random variation by natural selection

  Darwinism – the non-random selection of randomly varying replication entities by reason of their ‘phenotypic’ effects – is the only force I know that can, in principle, guide evolution in the direction of adaptive complexity. It works on this planet. It doesn’t suffer from any of the drawbacks that beset the other five classes of theory, and there is no reason to doubt its efficacy throughout the universe.

  The ingredients in a general recipe for Darwinian evolution are replicating entities of some kind, exerting phenotypic ‘power’ of some kind over their replication success. In The Extended Phenotype I referred to these necessary entities as ‘active germ-line replicators’ or ‘optimons’. It is important to keep their replication conceptually separate from their phenotypic effects, even though, on some planets, there may be a blurring in practice. Phenotypic adaptations can be seen as tools of replicator propagation.

  Gould disparages the replicator’s-eye view of evolution as preoccupied with ‘book-keeping’. The metaphor is a superficially happy one: it is easy to see the genetic changes that accompany evolution as book-keeping entries, mere accountant’s records of the really interesting phenotypic events going on in the outside world. Deeper consideration, however, shows that the truth is almost the exact opposite. It is central and essential to Darwinian (as opposed to Lamarckian) evolution that there shall be causal arrows flowing from genotype to phenotype, but not in the reverse direction. Changes in gene frequencies are not passive book-keeping records of phenotypic changes: it is precisely because (and to the extent that) they actively cause phenotypic changes that evolution of the phenotype can occur. Serious errors flow, both from a failure to understand the importance of this one-way flow,*18 and from an over-interpretation of it as inflexible and undeviating ‘genetic determinism’.

  The universal perspective leads me to emphasize a distinction between what may be called ‘one-off selection’ and ‘cumulative selection’. Order in the non-living world may result from processes that can be portrayed as a rudimentary kind of selection. The pebbles on a seashore become sorted by the waves, so that larger pebbles come to lie in layers separate from smaller ones. We can regard this as an example of the selection of a stable configuration out of initially more random disorder. The same can be said of the ‘harmonious’ orbital patterns of planets around stars, and electrons around nuclei, of the shapes of crystals, bubbles and droplets, even, perhaps, of the dimensionality of the universe in which we find ourselves. But this is all one-off selection. It does not give rise to progressive evolution because there is no replication, no succession of generations. Complex adaptation requires many generations of cumulative selection, each generation’s change building upon what has gone before. In one-off selection, a stable state develops and is then maintained. It does not multiply, does not have offspring.

  In life the selection that goes on in any one generation is one-off selection, analogous to the sorting of pebbles on a beach. The peculiar feature of life is that successive generations of such selection build up, progressively and cumulatively, structures that are eventually complex enough to foster the strong illusion of design. One-off selection is a commonplace of physics and cannot give rise to adaptive complexity. Cumulative selection is the hallmark of biology and is, I believe, the force underlying all adaptive complexity.

  Other topics for a future science of Universal Darwinism

  Active germ-line replicators together with their phenotypic consequences, then, constitute the general recipe for life; but the form of the system may vary greatly from planet to planet, both with respect to the replicating entities themselves, and with respect to the ‘phenotypic’ means by which they ensure their survival. Indeed, as Leslie Orgel has pointed out to me, the very distinction between ‘genotype’ and ‘phenotype’ may be blurred. The replicating entities do not have to be DNA or RNA. They do not have to be organic molecules at all. Even on this planet it is possible that DNA itself is a late usurper of the role, taking over from some earlier, inorganic crystalline replicator.*19

  A full science of Universal Darwinism might consider aspects of replicators transcending their detailed nature and the timescale over which they are copied. For instance, the extent to which they are ‘particulate’ as opposed to ‘blending’ probably has a more important bearing on evolution than their detailed molecular or physical nature. Similarly, a universe-wide classification of replicators might make more reference to their dimensionality and coding principles than to their size and structure. DNA is a digitally coded one-dimensional array. A ‘genetic’ code in the form of a two-dimensional matrix is conceivable. Even a three-dimensional code is imaginable, although students of Universal Darwinism will probably worry about how such a code could be ‘read’. (DNA is, of course, a molecule whose three-dimensional structure determines how it is replicated and transcribed, but that doesn’t make it a three-dimensional code. DNA’s meaning depends upon the one-dimensional sequential arrangement of its symbols, not upon their three-dimensional position relative to one another in the cell.) There might also be theoretical problems with analogue, as opposed to digital codes, similar to the theoretical problems that would be raised by a purely analogue nervous system.*20

  As for the phenotypic levers of power by which replicators influence their survival, we are so used to their being bound up into discrete organisms or ‘vehicles’ that we forget the possibility of a more diffuse extra-corporeal or ‘extended’ phenotype. Even on this Earth a large amount of interesting adaptation can be interpreted as part of the extended phenotype. There is, however, a general theoretical case that can be made in favour of the discrete organismal body, with its own recurrent life-cycle, as a necessity in any process of evolution of advanced adaptive complexity, and this topic might have a place in a full account of Universal Darwinism.

  Another candidate for full discussion might be what I shall call divergence, and convergence or recombination, of replicator lineages. In the case of Earth-bound DNA, ‘convergence’ is provided by sex and related processes. Here the DNA ‘converges’ within the species after having very recently ‘diverged’. But suggestions are now being made that a different kind of convergence can occur among lineages that originally diverged an exceedingly long time ago. For in
stance, there is evidence of gene transfer between fish and bacteria. The replicating lineages on other planets may permit very varied kinds of recombination, on very different timescales. On Earth, apart from in bacteria, the rivers of phylogeny are almost entirely divergent: if main tributaries ever recontact each other after branching apart it is only through the tiniest of trickling cross-streamlets, as in the fish/bacteria case. There is, of course, a richly anastomosing delta of divergence and convergence due to sexual recombination within the species, but only within the species. There may be planets on which the ‘genetic’ system permits much more cross-talk at all levels of the branching hierarchy, one huge fertile delta.

  I have not thought enough about the fantasies of the previous paragraphs to evaluate their plausibility. My general point is that there is one limiting constraint upon all speculations about life in the universe. If a life form displays adaptive complexity, it must possess an evolutionary mechanism capable of generating adaptive complexity. However diverse evolutionary mechanisms may be, even if there is no other generalization that can be made about life all around the universe, I am betting it will always be recognizable as Darwinian life. The Darwinian Law may be as universal as the great laws of physics.

  AFTERWORD

  I promised in one of my footnotes in this piece to return to ‘poetic science’. Steve Gould was so in love with his own rhetoric that he allowed his readers to confuse three kinds of discontinuity: macro-mutation, mass extinction and rapid gradualism. They have nothing else in common and the suggestion of a connection between them is worthless and profoundly misleading. Such is the danger of ‘poetic science’. For the most extreme example I know of Gould’s poetic rhetoric misleading even expert scientists, see the footnote to ‘The “Alabama Insert” ’ on this page.