Molecules are not enough

John Maynard Smith

This book contains a collection of essays about biology, most of which have been published before, in varied and often inacessible places, together with a new concluding chapter on dialectics. The authors have at least four things in common: they are Harvard professors, they have made distinguished contributions to the theory of ecology and evolution, they are dialectical materialists, and they write with wit and insight. Their thesis is that their philosophy is a valuable aid in the practice and understanding of biology. It is not only that Marxism helps in analysing the history and sociology of science: if you are a working biologist, they are saying, Marxism will help you to plan and to interpret the results of research. Crudely, Marxism is good for you. They argue that their own work has been helped by their philosophy. The claim is not only brave but necessary: I would not take them seriously unless they were willing to make it.

I have known and admired both authors for thirty years. During that time, I have worked on many of the same problems that fascinate them. More relevant, I too have been deeply influenced by Marxism, although in my case the start of my second career, as a biologist, coincided with a growing disillusion with Communist politics, and to a lesser extent with Marxist philosophy. Inevitably, therefore, this review is in part autobiographical: it is a debate with a past self.

Any discussion of the value of dialectics in biology must take in Lysenkoism, and it might as well start there. Lysenko is the millstone round the neck of the dialectical biologist. The acceptance of his views, under Party pressure, caused Russian biology damage from which it has not yet recovered, even thirty years later. Levins and Lewontin address these events, as they must. They do not (unlike some Maoists) claim that Lysenko was right: ‘Far from overthrowing traditional genetics and creating a new science,’ they write, Lysenkoism ‘cut short the pioneering work of Soviet genetics and set it back a generation. Its own contribution to contemporary biology was negligible.’ They also recognise that Marxist arguments were used to support Lysenko: ‘we cannot dismiss the obviously pernicious use of philosophy by Lysenko and his supporters as simply an aberration ... Unless Marxism examines its failures, they will be repeated.’

However, they point out, correctly in my view, that Lysenkoism is not to be explained merely as an error of Marxist philosophy, or as a crime committed by a bureaucratic government headed by a paranoid lunatic. The state of Soviet agriculture, the position of academic genetics at that time, the class differences between academic geneticists and agronomists, and the political reactions to collectivisation and to the German invasion, all contributed. It was, they argue, a misuse of Marxism to apply it in support of Lysenko: ‘Dialectical materialism is not, and has never been, a programmatic method for solving particular physical problems.’ Instead, they suggest that the philosophy provides a ‘set of warning signs’ against theories that are too narrow, mechanical or abstract.

Now I don’t think this will quite do. To see why, one must first understand why at that time Marxists saw Mendelian genetics as undialectical. The orthodox view was that genes influence development, but are themselves unaltered in the process, and hence that the ‘Lamarckian’ process of the inheritance of acquired characters is an impossibility. Hence the gene is a metaphysical and undialectical entity. Even if a sophisticated Marxist would not reject Mendelism on such apriorist grounds alone, his philosophy would certainly issue ‘warning signs’ against such a theory. I felt that way myself in 1950, and even went to the length of carrying out an experiment, on temperature acclimatisation in the fruitfly Drosophila, which I hoped – I cannot say expected – would demonstrate Lamarckian inheritance (it didn’t). Those of my friends who were both Marxists and biologists tended to be similarly hostile to orthodox genetics.

The paradox about genes was cleared up when their molecular structure was discovered. The current view is that it is information that is transmitted in one direction only: from gene to gene and from gene to protein but not from protein to gene. There is nothing implausible about a device that transmits information unidirectionally. A record-player is such a device: you cannot cut a disc by singing at the speaker. But this solution has emerged from a philosophical approach that has nothing to do with Marxism – and is still rejected by some Marxists. Levins and Lewontin point out that a dialectical picture of the relation between organism and environment can be constructed which is fully compatible with the non-Lamarckian nature of inheritance. This may be so, but it is not the point. It is always possible, after the event, to argue that a correct application of some philosophy would have led to the right solution. Unfortunately, when the issue was still open, most Marxists were not on the side of the angels.

I have spent some time on Lysenkoism because it constituted by far the most important application of Marxism to science. One could argue that the real damage was done, not because a group of scientists held erroneous views, but because those views were imposed by a dogmatic bureaucracy. This is true enough, but raises the question of whether Marxist philosophy predisposed that bureaucracy to think itself justified in imposing a theory on the scientific community. In any case, Lysenko must count as a strike against Marxism: are there positive achievements on the other side? The authors suggest that the contribution of Haldane and Oparin to an understanding of the origin of life is such an achievement. I would agree, but I am not sure that Haldane was significantly influenced by Marxism when he had these ideas – they were published in 1932.

Perhaps a better test is to ask how Levins and Lewontin have been influenced in their own scientific work. A different answer must be given for the two men. Levins was a Marxist before he was a biologist, and all his work shows it. His book Evolution in a Changing Environment, although it avoids the usual jargon, is the work of a conscious Marxist. I also think that it was a major contribution to ecology. In it, he faces up to the fact that ecology cannot be a science without theories, and yet any theory of ecology that is simple enough to be comprehensible will be too simple fully to reflect reality. To make matters worse, he recognises that ecological theory cannot be constructed without taking evolution into account, and cannot be applied except in a political and economic context. His attempt to construct models that are at once simple, general and ‘robust’ (i.e. the conclusions are not too sensitive to slight changes in the model) has influenced all subsequent work in the field. It is perhaps ironic that he made extensive use of mathematical techniques borrowed from capitalist economic theory: I cannot criticise because I have done the same. Since that time, he has worked more on applications of ecological theory. The essays in this book on pesticides, on Latin community health, and on applied biology in the Third World, reflect these interests. They illustrate the power of Marxism in the right hands. I have long thought of Levins as a rare example of a scientist whose work has been strengthened by adherence to a philosophy – Marxism or any other – and this book has confirmed that view.

Lewontin is harder to evaluate. He is one of a handful of contemporaries whose work has altered the way we see evolution. But how far has his success been influenced by his present philosophy? Unlike Levins, his career as a biologist was well established before he became a Marxist. Since that time, he has been an outspoken critic of various applications of biology to man, in particular of sociobiology and of hereditarian views about intelligence. But my impression is that the influence of Marxism on his own research has so far been slight. His most individual attitude to biology has been a wish to make it more like physics. In particular, he has hungered after theories that enable the future to be predicted from the past, as the best theories in physics do. This is an admirable, if somewhat optimistic aim, but it is not peculiarly Marxist.

His own work has been largely devoted to answering the following question: what is the precise nature of the genetic variability of natural populations? I was amused, on a recent visit to New Zealand, to meet a very bright Marxist graduate student who expressed astonishment that, when he visited Lewontin’s laboratory, he found everyone grinding up DNA. This reaction, although understandable, is unfair. Marxists are materialists, and there is nothing un-Marxist about studying the material basis of heredity, even by grinding it up. But if it is not un-Marxist, neither is it peculiarly Marxist. What is peculiar about Lewontin is not his use of physical techniques in biology, but the questions he has used those techniques to answer. It seems likely that those questions were in his mind long before he became a Marxist.

There are ways, however, in which his biology has been affected. Perhaps the most interesting is illustrated by his book, The Genetic Basis of Evolutionary Change. Its closing words are: ‘The fitness of a single locus ripped from its interactive context is about as relevant to real problems of evolutionary genetics as the study of the psychology of individuals isolated from their social context is to an understanding of man’s sociopolitical evolution. In both cases context and interaction are not simply second-order effects to be superimposed on a primary monadic analysis. Context and interaction are of the essence.’ These are impeccably dialectical words, but what are the interactions he wants us to take into account?

This question could have two answers. Thus suppose we want to know whether one gene at a locus will replace another in evolution. The new gene will have an effect: in the first instance, it will cause the appearance of an altered protein, or perhaps of the same protein at a different time or place in development. But whether this will increase ‘fitness’ depends on how its effects interact with everything else going on in the organism, and on how the organism interacts with its environment. This is true, but it is also obvious to almost everyone: even Lewontin’s favourite enemy, Richard Dawkins, is well aware of it. In fact, however, the last part of Lewontin’s book is devoted to a different kind of interaction. In a chapter entitled ‘The Genome as the Unit of Selection’ he argues that genes at different loci (i.e. at different places on a chromosome) will become preferentially linked together in co-operating units, and that it is these units, rather than the individual genes, that are relevant in evolution. Now this might be true: that is to say, one can construct models of evolving populations in which it is true. However, observations made after the publication of Lewontin’s book suggest that in actual populations it is rarely the case (technically, genes in natural populations tend to be in linkage equilibrium).

What this means is that, for most purposes, it is correct to think of evolution occurring ‘one gene at a time’. Of course, since newgenes only succeed through their interactions with all others, we expect the genes in a given organism to work together to produce a functional whole. But it seems that it usually not the case that groups of co-operating genes are held together, so that they can collectively spread through a species, replacing an earlier group of genes. By analogy, the genome evolves as a language might evolve if it incorporated one new word at a time, and not as a language would evolve if it incorporated a set of words and grammatical structures as a unit. On this matter, I think dialectics may have led Lewontin to espouse a hypothesis which might have been correct, but which probably is not.

I cannot leave this topic without a digression. A consistent proponent of the idea of co-operating groups of genes has been Ernst Mayr, whose earlier writings helped to educate the generation to which Levins, Lewontin and I belong. No one could be less Marxist than Mayr, but there are other sources for dialectical ideas. I remember once asking him whether the geneticist Richard Goldschmidt had been a Marxist, because his writings were permeated (for the worse, in my view) by dialectics. His reply was to remind me that only illiterate Anglo-Saxons have to get their dialectics from Marx and Engels: he and Goldschmidt had been raised on a diet of Hegel. Although chastened, I am unrepentant. It is often fruitful to think of evolution ‘one gene at a time’. Lewontin’s closing words are absurdly exaggerated. If you doubt this, remember that, from a practical viewpoint, one of the most important evolutionary events of recent years has been the spread of insecticide resistance, and that this can be well understood as the result of selection acting on single genes.

There is much in these essays that I found infuriating, but more that I found illuminating. Let me mention two of the pleasures. One is the first appearance in print of ‘Isadore Nabi on the Tendencies of Motion’, which has been circulating in samizdat for many years. It is an account of what might have happened if Newton had been replaced by a committee of statisticians. It is the only genuinely funny scientific article known to me, essentially because it is making a serious point. Why is it, by the way, that statisticians bring the best out in their critics? Remember Barnet Woolf’s definition of statistics as that branch of mathematics which enables a man to do 20 experiments a year and publish one false result in Nature.

A second pleasure is the analogy they draw between the present sweeping advance of molecular genetics and the advance of a Medieval army. The latter left behind unreduced castles under siege, because it lacked the weapons for direct assault. Molecular genetics also leaves unsolved problems in its wake. Most notably, it has left unsolved the problem of how eggs turn into animals. Genes may code for proteins, but how does this enable them to control development? It is now clear to everyone, as it was not ten years ago, that the castle of development must be reduced. But how is it to be done? The commonest view, I think, is that the molecular approach will do the job. Sequence enough genes, and the meaning of the message will at last become clear. The authors’ opinion, which I share, is that the phenomena of development will have to be studied at their own level, although they would probably agree with me that the laws of development, once understood, will prove to be reducible to molecular biology in the same sense that the laws of heredity have been so reduced. This seems to me one of the central strategic, and therefore philosophical, problems in biological research today: how is development best studied?

A problem that the authors mention obliquely, but do not address directly, is the following. How far have Hegel’s dialectical categories been rendered obsolete by advances in mathematics? One interpretation of dialectical materialism would be as follows. Marx and Engels wished to analyse the behaviour of highly complex systems. At that time, mathematics was adequate only for the description of simple dynamical systems. Therefore they were obliged to borrow from Hegel a set of verbal concepts, such as the negation of the negation and the change of quantity into quality. Today developments in mathematics make reliance on such vague verbal concepts less necessary.

This argument can be made more explicit by considering the change of quantity into quality. We now have a mathematical language for describing such changes. Imagine a dynamical system described by a set of differential equations. If we gradually change the parameters in the equations, the behaviour of the system will also change gradually: for example, if the behaviour is to oscillate, then the period and amplitude of the oscillation will change gradually. But ultimately, as we continue to change the parameters, we reach a threshold, or ‘bifurcation’, at which the behaviour changes dramatically: for example, the system may cease to oscillate, and start to grow exponentially. This, I take it, is a mathematical description of the change from quantity into quality. When one has played with a few systems of this kind, one has a better feel for how things are likely to behave. Other Hegelian categories may be susceptible to a similar analysis, although I am not sure: even in my most convinced Marxist phase, I could never make much sense of the negation of the negation or the interpenetration of opposites. But I note that Levins and Lewontin use the cybernetic concept of feedback when they discuss contradiction.

I would find it easier to draw this essay to a conclusion if I knew what conclusion to draw. It is easy to argue that all scientific research requires some prior philosophical commitment, and that it is therefore better that the commitment be conscious and explicit. The snag, of course, is that too firm a commitment only too easily leads to the espousal of erroneous hypotheses. This is not an objection just to Marxism: Pearson’s positivism was just as effective as Lysenko’s Marxism in leading to wrong views about genetics. Whether a scientist is conscious or unconscious of his philosophical position, there is no guarantee against backing the wrong horse, although some philosophies make it easier than others to correct one’s errors. At the present time, because of the astonishing success of molecular biology during the past thirty years, the overwhelming tendency among biologists is towards reductionism. I therefore welcome the final conclusion reached by Levins and Lewontin: ‘We must reject the molecular euphoria that has led many universities to shift biology to the study of the smallest units, dismissing population, organismic, evolutionary and ecological studies as a form of “stamp collecting” and allowing museum collections to be neglected. But once the legitimacy of these studies is recognised, we also urge the study of the vertical relations among levels, which operate in both directions.’