The Ghost in the Machine

  * I would like to remind the reader that a summary of these principles can be found in Appendix I.

  X

  EVOLUTION:

  THEME AND VARIATIONS

  I refuse to believe that God plays dice with the world. Albert Einstein

  In the last chapter we were concerned with ontogeny -- the development of the individual. We can now turn to phylogeny, and the crucial problem of evolutionary progress.

  The orthodox ('neo-Darwinian' or 'synthetic') theory attempts to explain all evolutionary changes by random mutations (and re-combinations) of genes; most mutations are harmful, but a very small proportion happens to be useful and is retained by natural selection. As already mentioned, 'randomness' means in this context that the hereditary changes wrought by mutation are totally unrelated to the animal's adaptive needs -- that they may alter its physique and behaviour 'in any and every direction.' In this view, evolution appears as a game of blind man's buff. Or, in the words of Professor Waddington -- a quasi-Trotskyite member of the Establishment whom I shall have occasion to quote repeatedly in this chapter: 'To suppose that the evolution of the wonderfully adapted biological mechanisms has depended only on a selection out of a haphazard set of variations, each produced by blind chance, is like suggesting that if we went on throwing bricks together into heaps, we should eventually be able to choose ourselves the most desirable house.' [1]

  To illustrate the point, here is a simple example. The giant panda -- mascot of the World Wildlife Fund -- has on its forelimbs an added sixth finger, which comes in very 'handy' for manipulating the bamboo-shoots which are its principal food. But that added finger would be a useless appendage without the proper muscles and nerves. The chances that among all possible mutations those which produced the additional bones, muscles and nerves should have occurred independently in the same population are of course infinitesimally small. And yet in this case there are only three variable factors involved. If we have, say, twenty factors (which is still a modest estimate for the evolution of a complex organ), the odds against their simultaneous alteration by chance alone become absurd, and instead of scientific explanations, we should be trading in miracles.

  Let us look at a less primitive example. The vertebrates' conquest of dry land started with the evolution of reptiles from some primitive amphibian form. The amphibians reproduced in the water, and their young were aquatic. The decisive novelty of the reptiles was that, unlike amphibians, they laid their eggs on dry land; they no longer depended on the water and were free to roam over the continents. But the unborn reptile inside the egg still needed an aquatic environment: it had to have water or else it would dry up long before it was born. It also needed a lot of food: amphibians hatch as larvae who fend for themselves, whereas reptiles hatch fully developed. So the reptilian egg had to be provided with a large mass of yolk for food, and also with albumen -- the white of egg -- to provide the water. Neither the yolk by itself, nor the egg-white itself, would have had any selective value. Moreover, the egg-white needed a vessel to contain it, otherwise its moisture would have evaporated. So there had to be a shell made of a leathery or limey material, as part of the evolutionary package-deal. But that is not the end of the story. The reptilian embryo, because of this shell, could not get rid of its waste products. The soft-shelled amphibian embryo had the whole pond as a lavatory; the reptilian embryo had to be provided with a kind of bladder. It is called the allantois, and is in some respects the forerunner of the mammalian placenta. But this problem having been solved, the embryo would still remain trapped inside its tough shell; it needed a tool to get out. The embryos of some fishes and amphibians, whose eggs are surrounded by a gelatinous membrane, have glands on their snouts: when the time is ripe, they secrete a chemical which dissolves the membrane. But embryos surrounded by a hard shell need a mechanical tool: thus snakes and lizards have a tooth transformed into a kind of tin-opener, while birds have a caruncle -- a hard outgrowth near the tip of their beaks which serves the same purpose. In some birds -- the honey-guides -- which lay their eggs like cuckoos in alien nests, the caruncle serves yet another purpose: it grows into a sharp hook with which the newly hatched invader kills off his foster-brethren, after which it amiably sheds the hook.

  All this refers to one aspect only of the evolution of reptiles; needless to say, countless other essential transformations of structure and behaviour were required to make the new creatures viable. The changes could have been gradual but at each step, however small, all the factors involved in the story had to co-operate harmoniously. The liquid store in the egg makes no sense without the shell. The shell would be useless, in fact murderous, without the allantois and without the tin-opener. Each change, taken in isolation, would be harmful, and work against survival. You cannot have a mutation A occurring alone, preserve it by natural selection, and then wait a few thousand or million years until mutation B joins it, and so on, to C and D. Each mutation occurring alone would be wiped out before it could be combined with the others. They are all interdependent. The doctrine that their coming together was due to a series of blind coincidences is an affront not only to commonsense but to the basic principles of scientific explanation.

  The propounders of the orthodox theory may have been uneasily aware that something essential was missing, and paid occasional lip service to 'unsolved problems', then hurriedly swept them under the carpet. To quote one authority, Sir Peter Medawar (himself not excessively given to tolerance of other people's opinions): 'Twenty years ago it all seemed easy: with mutation as a source of diversity, with selection to pick and choose. . . . Our former complacency can be traced, I suppose, to an understandable fault of temperament: scientists tend not to ask themselves questions until they can see the rudiments of an answer in their minds. Embarrassing questions tend to remain unasked or, if asked, to be answered rudely. . . . ' [1a]*

  * Compare this with Sir Julian Huxley's ex cathedra pronouncement: 'In the field of evolution, genetics has given its basic answer, and evolutionary biologists are free to pursue other problems.' [2]

  A convenient way to evade these embarrassing questions was to concentrate attention on the statistical treatment of mutations in large populations of the fruit fly, Drosophila melanogaster -- the pet animal of geneticists because it propagates so fast and has only four pairs of chromosomes. The method is based on the measurement of the variations of some isolated, and mostly trivial, characteristic, such as the colour of the eyes or the distribution of bristles on the fly's body. Steeped in the atomistic tradition, the upholders of the theory were apparently unable to see that these mutations of a single factor -- virtually all of them deleterious -- were quite irrelevant to the central problem of evolutionary progress, requiring simultaneous changes in all the factors affecting the structure and function of a complex organ. The geneticist's obsession with the bristles of the fruit fly, and the Behaviourist's obsession with the bar-pressing of the rat, show a more than superficial analogy. Both derive from a mechanistic philosophy which regards the living creature as a collection of elementary bits of behaviour (S-R units) and of elementary bits of heredity (Mendelian genes).

  Internal Selection

  The alternative proposed here is the concept of the open hierarchy. Let us see whether it can be applied to the evolutionary process. I shall start by quoting Waddington's answer to problems of the type posed by the giant panda's finger:

  There are still some of us for whom the orthodox modern explanations do not seem very satisfying. One well-known problem is this: many organs are very complex things, and in order to bring about any improvement in their functioning, it would be necessary to make simultaneous alterations in several different characters . . . and that, it might appear, is something which one would not expect to occur under the influence of chance alone. There have always been, and still are, reputable biologists who feel that such considerations make it doubtful whether random hereditary changes can provide a sufficient basis for evolution. But I believe that the difficu
lty largely disappears if one remembers that an organ like an eye is not simply a collection of elements, such as a retina, a lens, an iris, and so on, which are put together and happen to fit. It is something which is gradually formed while the adult animal is developing out of the egg; and as the eye forms, the different parts influence one another. Several people have shown that if, by some experimental means, the retina and eyeball are made larger than usual, that in itself will cause a larger lens to appear, of at least approximately the appropriate size for vision. There is no reason, therefore, why a chance mutation should not affect the whole organ in a harmonious way; and there is a reasonable possibility that it might improve it. . . . A random change in a hereditary factor will, in fact, not usually result in an alteration in just one element of the adult animal; it will bring about a shift in the whole developmental system, and may thus alter a complex organ as a whole. [3] (my italics)

  We remember from the previous chapter that the growing eye-bud of the embryo is an autonomous holon, which, if part of its tissue is taken away, will nevertheless develop into a normal eye, thanks to its self-regulating properties. It is by no means surprising that it should display the same self-regulating powers, or 'flexible strategies' of growth, if the disturbance is caused not by a human agent, but by a mutated gene, as Waddington suggests. The chance mutation merely triggers off the process; the 'prenatal skills' of the embryo will do the rest, in every successive generation. The enlarged eye has become an evolutionary novelty.*

  * It should be added that the example of the enlarged mutant eye is typical of the sort of thing a mutating gene will do. Genes regulate chemical reaction rates, including the rates of growth; and one of the most frequent effects of gene mutations is to alter the speed of growth of one part relative to others, and thus to modify the proportions of the organ.

  But embryonic development is a many-levelled hierarchic process; and this leads one to assume that selective and regulative controls operate on several levels to eliminate harmful mutations and to co-ordinate the effects of acceptable ones. Various authors* have suggested that this screening process might start at the very base of the hierarchy, on the level of the molecular chemistry of the gene-complex. Mutations are chemical changes, presumably caused by the impact of cosmic radiations and other factors, on the germ cells. The changes consist in alterations in the sequence of the chemical units in the chromosomes -- the four letters of the genetic alphabet. Mostly they are the equivalents of misprints. But there seems to be again a hierarchy of correctors and proof-readers at work to eliminate these; 'The struggle for survival of mutations begins at the moment mutation occurs', writes L.L. Whyte. 'It is obvious that entirely arbitrary changes will not be physically, chemically or functionally stable. . . . Only those changes which result in a mutated system that satisfies certain stringent physical, chemical and functional conditions will be able to survive. . . . ' [4] All others will be eliminated, either by the death of the mutated cell and its offspring at an early stage or, as we shall presently see, by the remarkable self-repairing properties of the gene-complex as a whole.

  * Von Bertalanffy, Darlington, Spurway, Lima da Faria, L.L. Whyte. See footnote on p. 147.

  In the orthodox theory, natural selection is entirely due to the pressures of the environment, which kills off the unfit and blesses the fit with abundant progeny. In the light of the preceding considerations, however, before a new mutation has a chance to be submitted to the Darwinian tests of survival in the external environment, it must have passed the tests of internal selection for its physical, chemical and biological fitness.

  The concept of internal selection, of a hierarchy of controls which eliminate the consequences of harmful gene-mutations and co-ordinates the effects of useful mutations, is the missing link in orthodox theory between the 'atoms' of heredity and the living stream of evolution. Without that link, neither of them makes sense. There can be no doubt that random mutations do occur: they can be observed in the laboratory. There can be no doubt that Darwinian selection is a powerful force. But in between these two events, between the chemical changes in a gene and the appearance of the finished product as a newcomer on the evolutionary stage, there is a whole hierarchy of internal processes at work which impose strict limitations on the range of possible mutations and thus considerably reduce the importance of the chance factor. We might say that the monkey works at a typewriter which the manufacturers have programmed to print only syllables which exist in our language, but not nonsense syllables. If a nonsense syllable occurs, the machine will automatically erase it.* To pursue the metaphor, we would have to populate the higher levels of the hierarchy with proof-readers and then editors, whose task is no longer elimination, but correction, self-repair and co-ordination -- as in the example of the mutated eye.

  * This metaphor is almost literally applicable to mistakes made in the protein manufacture in micro-organisms due to 'nonsense syllables' appearing in the RNA code. [5]

  That was an example of harmonising the consequences of a potentially favourable mutation. Let me now quote another example of evolutionary self-repair after a potentially harmful mutation.

  The Case of the Eyeless Fly

  The fruit fly has a mutant gene which is recessive, i.e., when paired with a normal gene, has no discernible effect (it will be remembered that genes operate in pairs, each gene in the pair being derived from one parent). But if two of these mutant genes are paired in the fertilised egg, the offspring will be an eyeless fly. If now a pure stock of eyeless flies is made to inbreed, then the whole stock will have only the 'eyeless' mutant gene, because no normal gene can enter the stock to bring light into their darkness. Nevertheless, within a few generations, flies appear in the inbred 'eyeless' stock with eyes that are perfectly normal. The traditional explanation of this remarkable phenomenon is that the other members of the gene-complex have been 'reshuffled and recombined in such a way that they deputise for the missing normal eye-forming gene'. [6] Now re-shuffling, as every poker player knows, is a randomising process. No biologist would be so perverse as to suggest that the new insect-eye evolved by pure chance, thus repeating within a few generations an evolutionary process which took hundreds of millions of years. Nor does the concept of natural selection provide the slightest help in this case. The re-combination of genes to deputise for the missing gene must have been co-ordinated according to some overall plan which includes the rules of genetic self-repair after certain types of damage by deleterious mutations. But such co-ordinative controls can only operate on levds higher than that of individual genes. Once more we are driven to the conclusion that the genetic code is not an architect's blueprint; that the gene-complex and its internal environment form a remarkably stable, closely knit, self-regulating micro-hierarchy; and that mutated genes in any of its holons are liable to cause corresponding reactions in others, co-ordinated by higher levels. This micro-hierarchy controls the prenatal skills of the embryo, which enable it to reach its goal, regardless of the hazards it may encounter during development. But phylogeny is a sequence of ontogenies, and thus we are confronted with the profound question: is the mechanism of phylogeny also endowed with some kind of evolutionary instruction-booklet? Is there a strategy of the evolutionary process comparable to the 'strategy of the genes' -- to the 'directiveness' of ontogeny (as E.S. Russell has called it)?

  Let me recapitulate. The eyes in the normal fruit fly, and the eyes which suddenly appear in the 'eyeless stock', are homologous organs, identical in appearance, and yet produced by a different combination of genes; and this is only one of many similar phenomena. Genetic atomism is dead. Hereditary stability and hereditary change are both based, not on a mosaic of genes, but on the action of the gene-complex 'as a whole'. But this face-saving expression -- which is now coming into increased use -- is empty, like so many other holistic formulations, unless we interpolate between the gene-complex as a whole, and the individual gene, a hierarchy of genetic subassemblies -- self-regulating holons of heredity,
which control the development of organs, and also control their possible evolutionary modifications, by canalising the effects of random mutations. A hierarchy with its built-in, self-regulatory safeguards is a stable affair. It cannot be pulled in here, pulled out there, like Patou belabouring his model. It is capable of variation and change, but only in co-ordinated ways and only in limited directions. Can we say anything about the general principles which determine that direction?

  The Puzzle of Homology

  The most fundamental principle of evolutionary strategy, related to the watchmakers' parable, is the standardisation of sub-assemblies. But since most of us have no very clear idea of the mechanism of our time-pieces, we might look under the hood of a motor car instead. Here the sub-assemblies are easily named: chassis, engine, battery, steering, brakes, differential, and so on to the distributor and heating system. Each of these component parts is a more or less self-contained unit, a mechanical holon in its own right. A V8 engine, or a standard battery, can be taken out of the car and made to function by itself, like an organ in vitro. It can be transferred to another type of car, and even to a different species of machine, such as a motor boat. But how do automobiles evolve?