Perhaps the best way to see this is to go back to the recipe analogy. It will be agreed that you can’t divide a cake up into its component crumbs and say ‘This crumb corresponds to the first word in the recipe, this crumb corresponds to the second word in the recipe’, etc. In this sense it will be agreed that the whole recipe maps onto the whole cake. But now suppose we change one word in the recipe; for instance, suppose ‘baking-powder’ is deleted or is changed to ‘yeast’. We bake 100 cakes according to the new version of the recipe, and 100 cakes according to the old version of the recipe. There is a key difference between the two sets of 100 cakes, and this difference is due to a one-word difference in the recipes. Although there is no one-to-one mapping from word to crumb of cake, there is one-to-one mapping from word difference to whole-cake difference. ‘Baking-powder’ does not correspond to any particular part of the cake: its influence affects the rising, and hence the final shape, of the whole cake. If ‘baking-powder’ is deleted, or replaced by ‘flour’, the cake will not rise. If it is replaced by ‘yeast’, the cake will rise but it will taste more like bread. There will be a reliable, identifiable difference between cakes baked according to the original version and the ‘mutated’ versions of the recipe, even though there is no particular ‘bit’ of any cake that corresponds to the words in question. This is a good analogy for what happens when a gene mutates.
An even better analogy, because genes exert quantitative effects and mutations change the quantitative magnitude of those effects, would be a change from ‘350 degrees’ to ‘450 degrees’. Cakes baked according to the ‘mutated’, higher-temperature version of the recipe will come out different, not just in one part but throughout their substance, from cakes baked according to the original lower-temperature version. But the analogy is still too simple. To simulate the ‘baking’ of a baby, we should imagine not a single process in a single oven, but a tangle of conveyor belts, passing different parts of the dish through 10 million different miniaturized ovens, in series and in parallel, each oven bringing out a different combination of flavours from 10,000 basic ingredients. The point of the cooking analogy, that the genes are not a blueprint but a recipe for a process, comes over from the complex version of the analogy even more strongly than from the simple one.
It is time to apply this lesson to the question of the inheritance of acquired characteristics. The thing about building something from a blueprint, as opposed to a recipe, is that the process is reversible. If you have a house, it is easy to reconstruct its blueprint. Just measure all the dimensions of the house and scale them down. Obviously, if the house were to ‘acquire’ any characteristics — say an interior wall were knocked down to give an open-plan ground floor — the ‘reverse blueprint’ would faithfully record the alteration. Just so if the genes were a description of the adult body. If the genes were a blueprint, it would be easy to imagine any characteristic that a body acquired during its lifetime being faithfully transcribed back into the genetic code, and hence passed into the next generation. The blacksmith’s son really could inherit the consequences of his father’s exercise. It is because the genes are not a blueprint but a recipe that this is not possible. We can no more imagine acquired characteristics being inherited than we can imagine the following. A cake has one slice cut out of it. A description of the alteration is now fed back into the recipe, and the recipe changes in such a way that the next cake baked according the altered recipe comes out of the oven with one slice already neatly missing.
Lamarckians are traditionally fond of calluses, so let us use that example. Our hypothetical bank clerk had soft, pampered hands except for a hard callus on the middle finger of his right hand, his writing finger. If generations of his descendants all write a great deal, the Lamarckian expects that genes controlling the development of skin in that region will be altered in such a way that babies come to be born with the appropriate finger already hardened. If the genes were a blueprint this would be easy. There would be a gene ‘for’ each square millimetre (or appropriate small unit) of skin. The whole surface of the skin of an adult bank clerk would be ‘scanned’, the hardness of each square millimetre carefully recorded and fed back into the genes ‘for’ that particular square millimetre, in particular the appropriate genes in his sperms.
But the genes are not a blueprint. There is no sense in which there is a gene ‘for’ each square millimetre. There is no sense in which the adult body could be scanned and its description fed back into the genes. The ‘coordinates’ of a callus could not be ‘looked up’ in the genetic record and the ‘appropriate’ genes altered. Embryonic development is a process, in which all working genes participate; a process which, if correctly followed in the forward direction, will result in an adult body; but it is a process that is inherently, by its very nature, irreversible. The inheritance of acquired characteristics not only doesn’t happen: it couldn’t happen in any life-form whose embryonic development is epigenetic rather than preformationistic. Any biologist that advocates Lamarckism is, though he may be shocked to hear it, implicitly advocating an atomistic, deterministic, reductionistic embryology. I didn’t want to burden the general reader with that little string of pretentious jargon words: I just couldn’t resist the irony, for the biologists who come closest to sympathizing with Lamarckism today also happen to be particularly fond of using those same cant words in criticizing others.
This is not to say that, somewhere in the universe, there may not be some alien system of life in which embryology is preformationistic; a life-form that really does have ‘blueprint genetics’, and that really could, therefore, inherit acquired characteristics. All that I have shown so far is that Lamarckism is incompatible with embryology as we know it. My claim at the outset of this chapter was stronger: that, even if acquired characteristics could be inherited, the Lamarckian theory would still be incapable of explaining adaptive evolution. This claim is so strong that it is intended to apply to all life-forms, everywhere in the universe. It is based upon two lines of reasoning, one concerned with difficulties over the principle of use and disuse, the other with further problems with the inheritance of acquired characteristics. I shall take these in reverse order.
The problem with acquired characteristics is basically this. It is all very well inheriting acquired characteristics, but not all acquired characteristics are improvements. Indeed, the vast majority of them are injuries. Obviously evolution is not going to proceed in the general direction of adaptive improvement if acquired characteristics are inherited indiscriminately: broken legs and smallpox scars being passed down the generations just as much as hardened feet and suntanned skin. Most of the characteristics that any machine acquires as it gets older tend to be the accumulated ravages of time: it wears out. If they were gathered up by some kind of scanning process and fed into the blueprint for the next generation, successive generations would get more and more decrepit. Instead of starting afresh with a new blueprint, each new generation would begin life encumbered and scarred with the accumulated decay and injuries of previous generations.
This problem is not necessarily insuperable. It is undeniable that some acquired characteristics are improvements, and it is theoretically conceivable that the inheritance mechanism might somehow discriminate the improvements from the injuries. But in wondering how this discrimination might work, we are now led to ask why acquired characteristics sometimes are improvements. Why, for instance, do areas of skin that are used, like the soles of a barefoot runner, become thicker and tougher? On the face of it, it would seem more probable that the skin would become thinner: on most machines, parts that are subject to wear and tear become thinner, for the obvious reason that wear removes particles rather than adding them.
The Darwinian, of course, has a ready answer. Skin that is subject to wear and tear gets thicker, because natural selection in the ancestral past has favoured those individuals whose skin happened to respond to wear and tear in this advantageous way. Similarly, natural selection favoured those members
of ancestral generations who happened to respond to sunlight by going brown. The Darwinian maintains that the only reason even a minority of acquired characteristics are improvements is that there is an underpinning of past Darwinian selection. In other words, the Lamarckian theory can explain adaptive improvement in evolution only by, as it were, riding on the back of the Darwinian theory. Given that Darwinian selection is there in the background, to ensure that some acquired characteristics are advantageous, and to provide a mechanism for discriminating the advantageous from the disadvantageous acquisitions, the inheritance of acquired characteristics might, conceivably, lead to some evolutionary improvement. But the improvement, such as it is, is all due to the Darwinian underpinning. We are forced back to Darwinism to explain the adaptive aspect of evolution.
The same is true of a rather more important class of acquired improvements, those that we lump together under the heading of learning. During the course of its life, an animal becomes more skilled at the business of making its living. The animal learns what is good for it and what is not. Its brain stores a large library of memories about its world, and about which actions tend to lead to desirable consequences and which to undesirable consequences. Much of an animal’s behaviour therefore comes under the heading of acquired characteristics, and much of this type of acquisition — ‘learning’ — really does deserve the title of improvement. If parents could somehow transcribe the wisdom of a lifetime’s experience into their genes, so that their offspring were born with a library of vicarious experience built in and ready to be drawn upon, those offspring could begin life one jump ahead. Evolutionary progress might indeed speed up, as learned skills and wisdom would automatically be incorporated into the genes.
But this all presupposes that the changes in behaviour that we call learning are, indeed, improvements. Why should they necessarily be improvements? Animals do, as a matter of fact, learn to do what is good for them, rather than what is bad for them, but why? Animals tend to avoid actions that have, in the past, led to pain. But pain is not a substance. Pain is just what the brain treats as pain. It is a fortunate fact that those occurrences that are treated as painful, for instance violent puncturing of the body surface, also happen to be those occurrences that tend to endanger the animal’s survival. But we could easily imagine a race of animals that enjoyed injury and other occurrences endangering their survival; a race of animals whose brain was so constructed that it took pleasure in injury and felt as painful those stimuli, such as the taste of nutritious food, which augur well for their survival. The reason we do not in fact see such masochistic animals in the world is the Darwinian reason that masochistic ancestors, for obvious reasons, would not have survived to leave descendants that inherited their masochism. We could probably, by artificial selection in padded cages, in pampered conditions where survival is assured by teams of vets and minders, breed a race of hereditary masochists. But in nature such masochists would not survive, and this is the fundamental reason why the changes that we call learning tend to be improvements rather than the reverse. We have again arrived at the conclusion that there must be a Darwinian underpinning to ensure that acquired characteristics are advantageous.
We now turn to the principle of use and disuse. This principle does seem to work rather well for some aspects of acquired improvements. It is a general rule that does not depend upon specific details. The rule says simply, ‘Any bit of the body that is frequently used should grow larger; any bit that is not used should become smaller or even wither away altogether’. Since we can expect that useful (and therefore presumably used) bits of body in general will benefit by being enlarged, while useless (and therefore presumably unused) bits might as well not be there at all, the rule does seem to have some general merit. Nevertheless, there is a big problem about the principle of use and disuse. This is that, even if there were no other objection to it, it is much too crude a tool to fashion the exquisitely delicate adaptations that we actually see in animals and plants.
The eye has been a useful example before, so why not again? Think of all the intricately cooperating working parts: the lens with its clear transparency, its colour correction and its correction for spherical distortion; the muscles that can instantly focus the lens on any target from a few inches to infinity; the iris diaphragm or ‘stopping down’ mechanism, which fine-tunes the aperture of the eye continuously, like a camera with a built-in lightmeter and fast special-purpose computer; the retina with its 125 million colour-coding photocells; the fine network of blood vessels that fuels every part of the machine; the even finer network of nerves — the equivalent of connecting wires and electronic chips. Hold all this fine-chiselled complexity in your mind, and ask yourself whether it could have been put together by the principle of use and disuse. The answer, it seems to me, is an obvious ‘no’.
The lens is transparent and corrected against spherical and chromatic aberration. Could this have come about through sheer use? Can a lens be washed clear by the volume of photons that pour through it? Will it become a better lens because it is used, because light has passed through it? Of course not. Why on earth should it? Will the cells of the retina sort themselves into three colour-sensitive classes, simply because they are bombarded with light of different colours? Again, why on earth should they? Once the focusing muscles exist, it is true that exercising them will make them grow bigger and stronger; but this will not in itself make images come into sharper focus. The truth is that the principle of use and disuse is incapable of shaping any but the crudest and most unimpressive of adaptations.
Darwinian selection, on the other hand, has no difficulty in explaining every tiny detail. Good eyesight, accurate and true down to pernickity detail, can be a matter of life and death for an animal. A lens, properly focused and corrected against aberration, can make all the difference, for a fast-flying bird like a swift, between catching a fly and smashing into a cliff. A well-modulated iris diaphragm, stopping down rapidly when the sun comes out, can make all the difference between seeing a predator in time to escape and being dazzled for a fatal instant. Any improvement in the effectiveness of an eye, no matter how subtle and no matter how deeply buried in internal tissues, can contribute to the animal’s survival and reproductive success, and hence to the propagation of the genes that made the improvement. Therefore Darwinian selection can explain the evolution of the improvement. The Darwinian theory explains the evolution of successful apparatus for survival, as a direct consequence of its very success. The coupling between the explanation, and that which is to be explained, is direct and detailed.
The Lamarckian theory, on the other hand, relies on a loose and crude coupling: the rule that anything that is used a great deal would be better if it were bigger. This amounts to relying on a correlation between the size of an organ and its effectiveness. If there is such a correlation, it is surely an exceedingly weak one. The Darwinian theory in effect relies on a correlation between the effectiveness of an organ and its effectiveness: a necessarily perfect correlation! This weakness of the Lamarckian theory does not depend upon detailed facts about the particular forms of life that we see on this planet. It is a general weakness that applies to any kind of adaptive complexity, and I think it must apply to life anywhere in the universe, no matter how alien and strange the details of that life may be.
Our refutation of Lamarckism, then, is a bit devastating. First, its key assumption, that of the inheritance of acquired characteristics, seems to be false in all life-forms that we have studied. Second, it not only is false but it has to be false in any life-form that relies upon an epigenetic (‘recipe’) rather than a preformationistic (‘blueprint’) kind of embryology, and this includes all life-forms that we have studied. Third, even if the assumptions of the Lamarckian theory were true, the theory is in principle, for two quite separate reasons, incapable of explaining the evolution of serious adaptive complexity, not just on this earth but anywhere in the universe. So, it isn’t that Lamarckism is a rival to the Darwinian theory that
happens to be wrong. Lamarckism isn’t a rival to Darwinism at all. It isn’t even a serious candidate as an explanation for the evolution of adaptive complexity. It is doomed from the start as a potential rival to Darwinism.
There are a few other theories that have been, and even occasionally still are, advanced as alternatives to Darwinian selection. Once again, I shall show that they are not really serious alternatives at all. I shall show (it is really obvious) that these ‘alternatives’ — ‘neutralism’, ‘mutationism’, and so on — may or may not be responsible for some proportion of observed evolutionary change, but they cannot be responsible for adaptive evolutionary change, that is for change in the direction of building up improved devices for survival like eyes, ears, elbow joints, and echo-ranging devices. Of course, large quantities of evolutionary change may be non-adaptive, in which case these alternative theories may well be important in parts of evolution, but only in the boring parts of evolution, not the parts concerned with what is special about life as opposed to non-life. This is especially clear in the case of the neutralist theory of evolution. This has a long history, but it is particularly easy to grasp in its modern, molecular guise in which it has been promoted largely by the great Japanese geneticist Motoo Kimura, whose English prose style, incidentally, would shame many a native speaker.