Consider another line of argument that the species selectionist might adopt. He might protest that it is unreasonable of me to base my combinatorial calculation on the assumption that the ten major trends in the evolution of whales are independent. Surely the number of combinations we have to consider will be drastically reduced by correlations among the different trends? It is important here to distinguish two different sources of correlation, which may be called incidental correlation and adaptive correlation. An incidental correlation is an intrinsic consequence of the facts of embryology. For example, elongation of the left foreleg is unlikely to be independent of elongation of the right foreleg. Any mutation that achieves one is intrinsically likely to achieve the other simultaneously. Slightly less obviously, the same might be partially true of elongation of the forelegs and the hindlegs. There are probably many similar cases that are less obvious still.
An adaptive correlation, on the other hand, does not follow directly from the mechanics of embryology. A lineage that moves from a terrestrial to an aquatic existence is likely to need changes to its locomotory system and its respiratory system, and there is no obvious reason to expect any intrinsic connection between them. Why should trends that convert walking limbs to flippers be intrinsically correlated with trends to boost the efficiency of the lungs in extracting oxygen? Of course it is possible that two such coadapted trends might be correlated as an incidental consequence of embryological mechanisms, but the correlation is no more likely to be positive than negative. We are back to our combinatorial calculation, albeit we must exercise care in counting our separate dimensions of change.
Finally, the species selectionist may retreat and invoke ordinary low-level natural selection to weed out ill-coadapted combinations of change, so that speciation events only serve up already tried and proved combinations to the sieve of species selection. But this ‘species selectionist’ is, by Gould’s lights, no species selectionist at all! He has conceded that all the interesting evolutionary change results from inter-allele selection and not from inter-species selection, albeit it may be concentrated in brief bursts, punctuating stasis. He has conceded the violation of Wright’s Rule. And if ‘Wright’s Rule’ now seems unfairly easy to violate, that is what I meant when I said that Gould had stuck his neck out. I should repeat that Wright himself is not responsible for naming the rule.
The theory of species selection, growing out of that of punctuated equilibria, is a stimulating idea which may well explain some single dimensions of quantitative change in macroevolution. I would be very surprised if it could be used to explain the sort of complex multidimensional adaptation that I find interesting, the ‘Paley’s watch’, or ‘Organs of extreme Perfection and Complication’, kind of adaptation that seems to demand a shaping agent at least as powerful as a deity. Replicator selection, where the replicators are alternative alleles, may well be powerful enough. If the replicators are alternative species, however, I doubt if it is powerful enough, because it is too slow. Eldredge and Cracraft (1980, p. 269) appear to agree: ‘The concept of natural selection (fitness differences, or differential reproduction of individuals within populations) appears to be a corroborated, within-population phenomenon, and constitutes the best available explanation for the origin, maintenance, and possible modification of adaptations.’ If this is indeed the opinion of ‘punctuationists’ and ‘species selectionists’ generally, it is hard to see what all the fuss is about.
For simplicity I have discussed the theory of species selection as one in which the species is treated as a replicator. The reader will have noticed, however, that this is rather like speaking of an asexually reproducing organism as a replicator. Earlier in this chapter, we saw that the test of mutilation forces us strictly to limit the title of replicator to the genome of, say, a stick insect, not the stick insect itself. Similarly, in the species selection model, it is not the species that is the replicator but the gene-pool. It is tempting, now, to say, ‘In that case, why not go the whole hog and regard the gene as replicator rather than some larger unit, even in the Eldredge/Gould model?’ The answer is that if they are right about a gene-pool being a coadapted unit, homeostatically buffered against change, it might have the same kind of right to be treated as a single replicator as has the genome of a stick insect. The gene-pool has this right, however, only if it is reproductively isolated, just as the genome has that right only if it is reproduced asexually. Even then the right is a tenuous one.
Earlier in this chapter, we established that an organism is definitely not a replicator, although its genome may be if it is asexually reproduced. We have now seen that there may be a case for regarding the gene-pool of a reproductively isolated group, such as a species, as a replicator. If we provisonally accept the logic of this case, we can visualize evolution directed by selection among such replicators, but I have just concluded that this kind of selection is unlikely to explain complex adaptation. Apart from the small genetic fragment, which we discussed in the previous chapter, are there any other plausible candidates for the title of replicator?
I have previously supported the case for a completely non-genetic kind of replicator, which flourishes only in the environment provided by complex, communicating brains. I called it the ‘meme’ (Dawkins 1976a). Unfortunately, unlike Cloak (1975) but, if I understand them aright, like Lumsden and Wilson (1980), I was insufficiently clear about the distinction between the meme itself, as replicator, on the one hand, and its ‘phenotypic effects’ or ‘meme products’ on the other. A meme should be regarded as a unit of information residing in a brain (Cloak’s ‘i-culture’). It has a definite structure, realized in whatever physical medium the brain uses for storing information. If the brain stores information as a pattern of synaptic connections, a meme should in principle be visible under a microscope as a definite pattern of synaptic structure. If the brain stores information in ‘distributed’ form (Pribram 1974), the meme would not be localizable on a microscope slide, but still I would want to regard it as physically residing in the brain. This is to distinguish it from its phenotypic effects, which are its consequences in the outside world (Cloak’s ‘m-culture’).
The phenotypic effects of a meme may be in the form of words, music, visual images, styles of clothes, facial or hand gestures, skills such as opening milk bottles in tits, or panning wheat in Japanese macaques. They are the outward and visible (audible, etc.) manifestations of the memes within the brain. They may be perceived by the sense organs of other individuals, and they may so imprint themselves on the brains of the receiving individuals that a copy (not necessarily exact) of the original meme is graven in the receiving brain. The new copy of the meme is then in a position to broadcast its phenotypic effects, with the result that further copies of itself may be made in yet other brains.
Returning, for clarification, to DNA as our archetypal replicator, its consequences on the world are of two important types. Firstly, it makes copies of itself, making use of the cellular apparatus of replicases, etc. Secondly, it has effects on the outside world, which influence the chances of its copies’ surviving. The first of these two effects corresponds to the meme’s use of the apparatus of inter-individual communication and imitation to make copies of itself. If individuals live in a social climate in which imitation is common, this corresponds to a cellular climate rich in enzymes for copying DNA.
But what about the second kind of effect of DNA, the kind conventionally called ‘phenotypic’? How do a meme’s phenotypic effects contribute to its success or failure in being replicated? The answer is the same as for the genetic replicator. Any effect that a meme has on the behaviour of a body bearing it may influence that meme’s chance of surviving. A meme that made its bodies run over cliffs would have a fate like that of a gene for making its bodies run over cliffs. It would tend to be eliminated from the meme-pool. But just as promoting bodily survival is only part of what constitutes success in genetic replicators, so there are many other ways in which memes may work phenotypically fo
r their own preservation. If the phenotypic effect of a meme is a tune, the catchier it is the more likely it is to be copied. If it is a scientific idea, its chances of spreading through the world’s scientific brains will be influenced by its compatibility with the already established corpus of ideas. If it is a political or religious idea, it may assist its own survival if one of its phenotypic effects is to make its bodies violently intolerant of new and unfamiliar ideas. A meme has its own opportunities for replication, and its own phenotypic effects, and there is no reason why success in a meme should have any connection whatever with genetic success.
This is regarded by many of my biological correspondents as the weakest point in the whole meme theory (Greene 1978; Alexander 1980, p. 78; Staddon 1981). I don’t see the problem; or, rather, I do see the problem but I don’t think it is any greater for memes as replicators than it is for genes. Time and again, my sociobiological colleagues have upbraided me as a turncoat, because I will not agree with them that the ultimate criterion for the success of a meme must be its contribution to Darwinian ‘fitness’. At bottom, they insist, a ‘good meme’ spreads because brains are receptive to it, and the receptiveness of brains is ultimately shaped by (genetic) natural selection. The very fact that animals imitate other animals at all must ultimately be explicable in terms of their Darwinian fitness.
But there is nothing magic about Darwinian fitness in the genetic sense. There is no law giving it priority as the fundamental quantity that is maximized. Fitness is just a way of talking about the survival of replicators, in this case genetic replicators. If another kind of entity arises, which answers to the definition of an active germ-line replicator, variants of the new replicator that work for their own survival will tend to become more numerous. To be consistent, we could invent a new kind of ‘individual fitness’, which measured the success of an individual in propagating his memes.
It is, of course, true that ‘Memes are utterly dependent upon genes, but genes can exist and change quite independently of memes’ (Bonner 1980). But this does not mean that the ultimate criterion for success in meme selection is gene survival. It does not mean that success goes to those memes that favour the genes of the individuals bearing them. To be sure, this will sometimes be so. Obviously a meme that causes individuals bearing it to kill themselves has a grave disadvantage, but not necessarily a fatal one. Just as a gene for suicide sometimes spreads itself by a roundabout route (e.g. in social insect workers, or parental sacrifice), so a suicidal meme can spread, as when a dramatic and well-publicized martyrdom inspires others to die for a deeply loved cause, and this in turn inspires others to die, and so on (Vidal 1955).
It is true that the relative survival success of a meme will depend critically on the social and biological climate in which it finds itself, and this climate will certainly be influenced by the genetic make-up of the population. But it will also depend on the memes that are already numerous in the meme-pool. Genetic evolutionists are already happy with the idea that the relative success of two alleles can depend upon which genes at other loci dominate the gene-pool, and I have already mentioned this in connection with the evolution of ‘coadapted genomes’. The statistical structure of the gene-pool sets up a climate or environment which affects the success of any one gene relative to its alleles. Against one genetic background one allele may be favoured; against another genetic background its allele may be favoured. For example, if the gene-pool is dominated by genes that make animals seek dry places, this will set up selection pressures in favour of genes for an impermeable skin. But alleles for a more permeable skin will be favoured if the gene-pool happens to be dominated by genes for seeking damp places. The point is the obvious one that selection at any one locus is not independent of selection at other loci. Once a lineage begins evolving in a particular direction, many loci will fall into step, and the resulting positive feedbacks will tend to propel the lineage in the same direction, in spite of pressures from the outside world. An important aspect of the environment which selects between alleles at any one locus will be the genes that already dominate the gene-pool at other loci.
Similarly, an important aspect of selection on any one meme will be the other memes that already happen to dominate the meme-pool (Wilson 1975). If the society is already dominated by Marxist, or Nazi memes, any new meme’s replication success will be influenced by its compatibility with this existing background. Positive feedbacks will provide a momentum which can carry meme-based evolution in directions unconnected with, or even contradictory to, the directions that would be favoured by gene-based evolution. I agree with Pulliam and Dunford (1980) that cultural evolution ‘owes its origin and its rules to genetic evolution, but it has a momentum all its own’.
There are, of course, significant differences between meme-based and gene-based selection processes (Cavalli-Sforza & Feldman 1973, 1981). Memes are not strung out along linear chromosomes, and it is not clear that they occupy and compete for discrete ‘loci’, or that they have identifiable ‘alleles’. Presumably, as in the case of genes, we can strictly only talk about phenotypic effects in terms of differences, even if we just mean the difference between the behaviour produced by a brain containing the meme and that of a brain not containing it. The copying process is probably much less precise than in the case of genes: there may be a certain ‘mutational’ element in every copying event, and this, by the way, is also true of the ‘species selection’ discussed earlier in the chapter. Memes may partially blend with each other in a way that genes do not. New ‘mutations’ may be ‘directed’ rather than random with respect to evolutionary trends. The equivalent of Weismannism is less rigid for memes than for genes: there may be ‘Lamarckian’ causal arrows leading from phenotype to replicator, as well as the other way around. These differences may prove sufficient to render the analogy with genetic natural selection worthless or even positively misleading. My own feeling is that its main value may lie not so much in helping us to understand human culture as in sharpening our perception of genetic natural selection. This is the only reason I am presumptuous enough to discuss it, for I do not known enough about the existing literature on human culture to make an authoritative contribution to it.
Whatever the claims of memes to be regarded as replicators in the same sense as genes, the first part of this chapter established that individual organisms are not replicators. Nevertheless, they are obviously functional units of great importance, and it is now necessary to establish exactly what their role is. If the organism is not a replicator, what is it? The answer is that it is a communal vehicle for replicators. A vehicle is an entity in which replicators (genes and memes) travel about, an entity whose attributes are affected by the replicators inside it, an entity which may be seen as a compound tool of replicator propagation. But individual organisms are not the only entities that might be regarded as vehicles in this sense. There is a hierarchy of entities embedded in larger entities, and in theory the concept of vehicle might be applied to any level of the hierarchy.
The concept of hierarchy is a generally important one. Chemists believe that matter is made of about a hundred different kinds of atoms, interacting with each other by means of their electrons. Atoms are gregarious, forming huge assemblages which are governed by laws at their own level. Without contradicting the laws of chemistry, therefore, we find it convenient to ignore atoms when we are thinking about large lumps of matter. When explaining the workings of a motor car we forget atoms and van der Waal’s forces as units of explanation, and prefer to talk of cylinders and sparking plugs. This lesson applies not just to the two levels of atoms and cylinder heads. There is a hierarchy, ranging from fundamental particles below the atomic level up through molecules and crystals to the macroscopic chunks which our unaided sense organs are built to appreciate.
Living matter introduces a whole new set of rungs to the ladder of complexity: macromolecules folding themselves into their tertiary forms, intracellular membranes and organelles, cells, tissues, organs, org
anisms, populations, communities and ecosystems. A similar hierarchy of units embedded in larger units epitomizes the complex artificial products of living things—semiconductor crystals, transistors, integrated circuits, computers and embedded units that can only be understood in terms of ‘software’. At every level the units interact with each other following laws appropriate to that level, laws which are not conveniently reducible to laws at lower levels.
This has all been said many times before, and is so obvious as to be almost platitudinous. But one sometimes has to repeat platitudes in order to prove that one’s heart is in the right place! Especially if one wishes to emphasize a slightly unconventional sort of hierarchy, for this may be mistaken for a ‘reductionist’ attack on the idea of hierarchy itself. Reductionism is a dirty word, and a kind of ‘holistier than thou’ self-righteousness has become fashionable. I enthusiastically follow this fashion when talking about mechanisms within individual bodies, and have advocated ‘neuro-economic’ and ‘software’ explanations of behaviour in preference to conventional neurophysiological ones (Dawkins 1976b). I would favour an analogous approach to individual development. But there are times when holistic preaching becomes an easy substitute for thought, and I believe the dispute about units of selection provides examples of this.
The neo-Weismannist view of life which this book advocates lays stress on the genetic replicator as a fundamental unit of explanation. I believe it has an atom-like role to play in functional, teleonomic explanation. If we wish to speak of adaptations as being ‘for the good of’ something, that something is the active, germ-line replicator. This is a small chunk of DNA, a single ‘gene’ according to some definitions of the word. But I am of course not suggesting that small genetic units work in isolation from each other, any more than a chemist thinks that atoms do. Like atoms, genes are highly gregarious. They are often strung together along chromosomes, chromosomes are wrapped up in groups in nuclear membranes, enveloped in cytoplasm and enclosed in cell membranes. Cells too are normally not isolated, but cloned to form the huge conglomerates we know as organisms. We are now plugged into the familiar embedded hierarchy, and need go no further. Functionally speaking, too, genes are gregarious. They have phenotypic effects on bodies, but they do not do so in isolation. I stress this over and over again in this book.