Four commentaries in
Perspectives on Imitation: From Mirror Neurons to Memes.
Ed. S. Hurley and N. Chater, MIT Press 2005

Evidence for memetic drive?: Commentary on Iacoboni,  Vol 1, pp 203-5

I was thrilled when I learned of Iacoboni’s discovery that when a chimpanzee’s brain is morphed onto a human brain the areas of greatest expansion are those that are used in imitation. “Yes!” I thought “This is exactly what I predicted on the basis of memetic theory. Whoopee – memetics is right!” but then I had to pause because this is how to make the worst mistake in the book. Construct a wacky theory, derive a prediction from that theory, discover the prediction is correct and then (illegitimately) conclude that the theory must be true. So I would like to describe the prediction and consider whether these findings do have any implications for memetics or not.

Dawkins’s (1976) original idea in coining the term “meme” was to point out that when people imitate each other they not only copy information, but they must select what to copy, and their copies are not perfect. This is all that is required to apply the principles of universal Darwinism and, by definition, the information people copy is a replicator. Dawkins called that new replicator the “meme”.

One implication of the theory of memetics is that the capacity for imitation must inevitably let loose a new evolutionary process and, as Dawkins originally put it,  “Once this new evolution begins, it will in no necessary sense be subservient to the old” (Dawkins, 1976, 193-4). He criticised his colleagues because “In the last analysis they wish always to go back to ‘biological advantage’” to answer questions about  human behaviour (Dawkins, 1976, 193). But if memes are replicators then we must consider memetic advantage too. Humans must be the product of two replicators, not just one, and this should be obvious in the way they have evolved.

In exploring the implications of memetic advantage I hypothesised that the interests of the memes might force the genes to take a direction different from that which they would have taken otherwise: they would be forced to track the direction taken by memetic evolution. This is the co-evolutionary process I called memetic drive (Blackmore 1999, 2001).

Put simply the hypothesis is this. Once human ancestors could imitate, memes appeared and began competing to be copied, their success depending on the type of meme and the preferences and abilities of the people doing the copying. Given that at least some of the memes would provide survival benefits, this means an advantage to genes for the ability to copy those memes. If better imitation requires a bigger brain then this process alone will tend to increase brain size and improve imitation ability. As imitation ability increases, more memes will appear and their evolution will take off in various directions, perhaps including the creation of rituals, clothes, body decoration or music, including behaviours that are of more advantage to those memes themselves than to the genes of the people copying them. If being able to display the latest memes provides status (which is a reasonable assumption) then it will pay everyone to copy the best imitators, and to mate with them and, either way, this creates an advantage for genes for the ability to copy the latest memes. In this way genes would be expected to track the direction taken by purely memetic evolution and thus we humans have ended up with brains that are not only much larger, but are specially designed to be good at music, ritual, art and, of course, language.

This hypothesis allows for some (admittedly rather general) predictions. In particular, if brain size has been meme-driven then within groups of similar species brain size should correlate with ability to imitate. Of course there are few species capable of imitation but this prediction holds for humming birds (Jarvis et al 2000). Other aspects of the big brain hypothesis have been confirmed using simulations and mathematical modelling (Bull et al 2000, Higgs 2000). More specifically I predicted that brain scans of people either initiating or imitating actions should reveal that “imitation is the harder part—and also that the evolutionarily newer parts of the brain should be especially implicated in carrying it out.” (Blackmore 2000 p 73). This implies that the parts of the brain that differ most between chimpanzees and humans should be those involved in imitation (assuming that present-day chimpanzees are closer to our common ancestor than humans are). Finally, if memetic drive is responsible for the evolution of language then we should expect the language areas in the human brain to be derived from areas originally used for imitation.

This is what Iacoboni and his colleagues have demonstrated, so confirming these predictions. Iacoboni concludes that “From a relatively simple neural mechanism of matching observation and execution of action (mirror neurons), more complex functional properties were built and more complex behaviors were supported.” (ms p 13). The question now is why? The standard evolutionary view must be that it happened in the interests of the genes. Like Dawkins’s colleagues, most people will presumably “wish always to go back to ‘biological advantage’”. But the wider alternative remains; that when it comes to human evolution there may be more than one replicator competing for survival.

There is nothing mysterious about memetics. Memes are not mystical entities floating about in a few theorists’ minds. They are nothing more nor less than whatever it is that people copy when they imitate. So if you admit that people (imperfectly and selectively) copy each other, and you define a replicator as information that is copied with variation and selection, then you have to conclude that memes exist. All the doubt must be about whether memetics can ever prove itself useful as a science, and whether memes really have played the crucial role in human evolution that memetic theory suggests. Iacoboni’s findings fit perfectly with the predictions made, but then, as he discusses in his paper, there are many possible explanations for them. Memetics has made a start, but it has a great deal further to go if it is to prove its worth in understanding human evolution.

Blackmore, S.J. 1999 The Meme Machine Oxford, Oxford University Press.

Blackmore,S.J. 2000 The power of memes, Scientific American, 283:4, 64-73

Bull,L., Holland,O. and Blackmore,S. (2000) On meme-gene coevolution. Artificial Life, 6, 227-235

Dawkins,R. (1976) The Selfish Gene Oxford, Oxford University Press (new edition with additional material, 1989)

Higgs, P.G. (2000) The mimetic transition: a simulation study of the evolution of learning by imitation. Proceedings of the Royal Society London B., 267, 1355-1361

Jarvis, E.D., Ribeiro, S., da Silva, M.L., Ventura, D., Vielliard, J. and Mello, C.V. (2000) Behaviourally driven gene expression reveals song nuclei in hummingbird brain. Nature, 406, 628-632


A possible confusion between mimetic and memetic. Vol 2, pp 396-8

The terms “meme” and “memetics” have been used many times at this conference, and they sound very similar to Donald’s terms “mimesis” and “mimetic”. So it may be helpful to distinguish between the two.

“Meme” was coined by Dawkins (1976) to give a name to the replicator that is copied by imitation. As examples of memes, he suggested “tunes, ideas, catch-phrases, clothes fashions, ways of making pots or of building arches” (p 192) and included scientific theories, poems, chain letters and religious doctrines. He derived the term itself from the Greek word mimeme meaning “that which is imitated”, abbreviating it to a monosyllable that sounds a bit like gene. This choice itself was, as we now know, a highly successful meme. So the core meaning of “meme” is that which is imitated.

By contrast, Donald’s (1991) term “mimetic” is part of his three stage account of the evolution of human brains, culture, and cognition. These three stages are (1) the acquisition of mimetic skill, (2) lexical invention; the creation of words, spoken language and story telling, and (3) the externalization of memory, including symbolic art and the technology of writing both of which allowed humans to overcome the limitations of biological memory.

We can easily see that the terms are very different. Donald’s “mimetic” is closer to “mime” than to imitation  and he includes “gesture, body language, and mime, any of which can communicate an intention quite effectively, without words or grammars”. He also includes representing an event to oneself as a form of mimesis, and he describes mimesis as “a necessary preadaptation for the later evolution of language.” (Donald 2001 p 263). So “mimesis” includes internal representations and excludes language, story telling and writing. In contrast, memes are anything that is copied from person to person, and in the modern world the vast majority of memes are words, and combinations of words, both written and spoken.

These are the most obvious differences, but if we pursue them a bit further, we find that these two terms exemplify two fundamentally different approaches to human evolution. First, for Donald “mimesis rests on the ability to produce conscious, self-initiated, representational acts that are intentional but not linguistic.” (1991, p 168). This means that mimesis is essentially a symbolic or representational act. This is not implied in memetics. Memes are simply the actions, behaviours, or statements that people make, and the artefacts they build. If these memes can succeed in being copied they will thrive, and there is no requirement for them to be symbolic or representational, let alone conscious.

A second difference concerns the origins of variation and of creativity. Donald begins “As a purely replicative skill, imitation cannot generate much cultural variation, and plays a limited role in cultural evolution, more as a transmission device than a creative one.” (abstract). This more traditional view is completely reversed by the memetic view which might be stated like this. Imitation is the copying mechanism that made cultural evolution possible. Errors in imitation provide one source of cultural variation, recombination of old memes to make new ones provides more, and selection between the variants completes the process. Memetic evolution is a creative process in exactly the same way that biological evolution is, and depends on imitation as its mechanism of heredity. On the memetic view, imitation is the key to all of human creativity.

Finally, the two approaches differ over the question “Who benefits?” (Dennett 1995). The main point of memetics, as conceived by Dawkins, was to illustrate that memes can have replicator power as well as genes. He argued that we should get out of the habit of always appealing to biological or genetic advantage, because there may be other replicators that also drive evolution for their own advantage. Memetics is therefore the study of how the interests of memes affect human evolution and culture. By contrast, Donald sees mimesis as an adaptation; following the traditional view that the skill of copying gestures, actions and mimes is of biological advantage to the people who acquired it, rather than being of advantage to the gestures, actions, and mimes themselves.

We are far from knowing which theory, if either, is correct, but it may be helpful to realise how very different they are.

Dawkins,R. (1976) The Selfish Gene Oxford, Oxford University Press (new edition with additional material, 1989)

Dennett,D. (1995) Darwin’s Dangerous Idea, London, Penguin

Donald, M (1991) Origins of the Modern Mind: Three Stages in the Evolution of Culture and Cognition. Cambridge, Mass, Harvard University Press.

Donald, M. (2001) A Mind So Rare: The Evolution of Human Consciousness, New York, Norton



Even deeper misunderstandings of memes: Commentary on Gil-White, Vol 2, pp 406-9

Gil-White describes a memetics that I do not recognise. I will comment on just two of his main points.

First, he rejects three central concept for memetics: replication, imitation, and selfish memes. He regards replication as a red herring, but he defines replication to mean copying with 100% fidelity and a replicator as something that is copied perfectly. This is not the usual definition of a replicator, and there could be no memetics if it were. In fact the term replicator was coined by Dawkins (1976) in terms of the very evolutionary process that Gil-White describes; Universal Darwinism. Dennett refers to this general process as the evolutionary algorithm: if you have variation, selection and heredity you must get evolution. The information that is copied is called the replicator. Dawkins then invented the term “meme” to provide an example of a replicator other than the gene; one that is copied by the process of imitation. Gil-White claims that Dawkins insists on replication on the grounds that perfect copying is the only thing that can set Darwinian processes in motion. Not only is Gil-White wrong about Dawkins, but if replication is perfect there is no variation and hence there can be no evolution – whether genetic, memetic or any other kind.

Note that I prefer to use the original term “variation” here, because variation is general to all evolutionary processes whereas  “mutation” is a term specific to genes. I fear that Gil-White misleads us by talking about mutation in memes.

This highlights a major problem with Gil-White’s analysis which is that he relies heavily on analogies between memes and genes when these may not be warranted. He claims that insistence on a close genetic analogy reflects a misunderstanding of evolutionary genetics and regards Dawkins, Dennett and myself as pushing the analogy too far. But it is he who insists on this analogy, not memeticists.

I want to make this point as clear as possible. Memetics starts not from an analogy between memes and genes but from the recognition that both are replicators (information copied with variation and selection). Beyond that the two replicators work in very different ways. Genes have been evolving for several billion years and are now copied with extremely high fidelity. Memes have been evolving for only two or three million years at most, and are copied by all the messy, inaccurate and variable processes of imitation that we have been learning about at this conference. This difference is crucial to understanding how memes work, and it does not undermine the point that both memes and genes are replicators. Understanding replication by imitation is the heart of memetics. Replication cannot be a red herring for there can be no memetics without it.

Gil-White’s second red herring is imitation. Again he defines imitating as making exact copies. From everything we have learned at this conference it is clear that imitation rarely, if ever, produces exact copies of behaviours, and memetics must be based on this reality. Since memes are defined as the information that is copied by imitation, imitation cannot be a red herring for memetics.

Finally Gil-White rejects the notion of “selfish memes”.  It might be helpful to point out that the concept of selfish replicators falls straight out of the idea of the evolutionary algorithm. Once you have a replicator being copied with variation and selection, then that information will get copied whenever it can, regardless of the consequences. So memes are selfish in exactly the same sense as genes are. This does not mean that they have selfish plans or intentions, but merely that they will spread whenever and however they can. This notion of selfishness is intrinsic to the idea of the meme as a replicator, and is the basis on which memetics can be used to make predictions different from those made on the basis of adaptations that favour genes. If this is a red herring then there is nothing left of memetics.

I will skip now to the end of Gil White’s paper. Here he claims (quite rightly) that meme content is not everything. The problem is that he accuses me of saying that meme selection occurs solely as the result of meme content, which I do not.  Moreover, in arguing against what he takes to be my views on memetic drive he quotes a section from my book which is not actually about memetic drive.

Let me be clear about memetic drive. I think that most memeticists would agree with what I have said so far, but they may not agree that memetic drive occurs or that it can, as I have suggested, account for the rapid expansion of the human brain or our capacity for language. My hypothesis, put very briefly, is as follows. Once our ancestors were able to imitate, a new evolutionary process was let loose in which memes competed to be copied. The successful memes then changed the environment in which genes were selected, giving an advantage to genes for the ability to copy the currently successful memes. By this process, brains became successively better at copying the memes that had been successful in the memetic competition. In other words genes were forced, to some extent, to track the direction of memetic evolution.

There are perfectly legitimate theoretical questions about whether this process could work, and empirical questions about whether it actually happened. The process has been successfully modelled, and also compared with better-known versions of gene-culture co-evolution (Higgs 2000, Kendal & Laland 2000). However, what is important here is that memetic drive depends on heuristics such as copying, or mating with, the best imitators, which are related to those described by Gil-White. Without such bias memetic drive cannot work. The memetic drive hypothesis may be false, but it is not, and cannot be, based on the idea that only content determines which memes are selected.

I hope these very brief comments may help to clear up some fundamental misunderstandings about memetics.

Dawkins,R. (1976) The Selfish Gene Oxford, Oxford University Press (new edition with additional material, 1989)

Higgs, P.G. (2000) The mimetic transition: a simulation study of the evolution of learning by imitation. Proceedings of the Royal Society London B., 267, 1355-1361

Kendal, J.R. and Laland, K.N. (2000) Mathematical models for memetics.  Journal of Memetics 4(1)


Can memes meet the challenge?: Commentary on Chater and Greenberg. Vol 2, pp 409-11

Chater and Greenberg, in their different ways, describe the challenge faced by any memetic theory of cultural change. Chater points out how counter-intuitive the idea is. Indeed it is. As he puts it, meme theory views cultural complexity as generated by the blind watchmaker of non-purposive selectional forces rather than by the “sighted watchmaker” of conscious humans and their goals. Greenberg sets this up as “goals versus memes”, or as “humans’ deliberate pursuit of their conscious goals” versus competition between memes.

The challenge to meme theory, then, is to show that at least in some instances, Darwinian forces with memes as the replicator can better explain cultural complexity than can intelligent, purposeful design by humans.

It is important to stress at once that no meme theorist is likely to reject human goals as irrelevant to memetic evolution. The interesting question is what role they play. Are human goals the ultimate design force for culture (goal theory) or are they just one of many factors in a Darwinian design process acting on memes (meme theory)?

Analogies between genes and memes must be treated with caution, but we may usefully consider the role of animals’ goals in biological evolution. Imagine a bird of prey whose goals include staying alive, guarding its territory and feeding its chicks. These goals do not, on their own, design anything. Yet their pursuit has many design consequences, including the evolution of defences or camouflage in the prey, improvements in the predator’s hunting skills, and many unintended effects on genes in other organisms, such as the moss used for lining the predator’s nest, the parasites that take up residence there, the food eaten by the prey, and so on. We can best understand the overall design of this bird by thinking of the ultimate driving force as a Darwinian process in which genes compete to be copied. In this process, the goals of the bird are relevant but they are not the ultimate designer.

Consider now a favourite example of meme critics, the design of an aircraft. Surely, say the critics, this is completely different from the case of the bird, for the aircraft was designed by human intelligence and foresight in order to fulfil their conscious goals. My response is that the goals are relevant, but that we get a better overall understanding of aircraft design by taking the memes-eye view. On this view memes such as wing shapes, engine types, seating plans, window construction, food storage and safety features all compete to be copied. The numerous human designers are meme machines who, through their intelligence and education, are able to recombine memes to make new designs. Their many designs then compete to be brought to fruition, with many being dropped. The few that are finally built then compete in world markets and even fewer succeed for reasons that include fluctuations in the cost of oil, unexpected terrorist attacks that increase people’s fear of flying, birds that accidentally get sucked into jet engines, the evolution of the bacteria that thrive in warm food, and the patriotism of people who want their national flag painted on the tail fins. In this process, the goals of the human designer are essential but they cannot, alone, explain the final design of the plane.

The analogy is not close because genes and memes are copied and selected by such very different mechanisms, but I hope it is close enough to make this simple point – that on meme theory human goals are not treated as irrelevant but are given the same status that the goals of plants and animals are given in biological evolution. That is, they are just one of many factors contributing to design by selection.

If we accept this role for human goals, can meme theory meet Greenberg’s challenge and “show that an appeal to an analogue of evolution by natural selection is the best explanation of cultural evolution”? In the ambitious sense of providing a complete theory of cultural evolution, meme theory is very far from achieving this goal, but in limited domains it has already done so. A good example is the evolution of religions or, as Dawkins (1989) calls them, “viruses of the mind”.

In their adherence to religion, people invest precious resources in building temples and mosques. They spend time reciting scriptures and singing hymns. They burn offerings and pour away oils. They mutilate their children’s genitals causing them pain and disease. They wage wars and even kill themselves. And what is the role of “humans’ deliberate pursuit of their conscious goals” in all this?

If asked, religious people might say that they wish to serve God, to enter into communion with spirit, to be a better Muslim, Jew or Christian, or to attain life after death, but these goals are themselves derived from the religions and cannot explain the design of those religions. Meme theory takes the viewpoint of the religious memes themselves and investigates how and why they have succeeded in getting into people’s behavioural repertoire and being passed on to others. The major religions are seen as memeplexes built around instructions to pass on those memes (teach your children, pass on the good news). Obedience is obtained by a mixture of immediate benefits (a sense of belonging, or aesthetic pleasure) and protected by untestable threats and promises (hell, heaven and eternal damnation). These religions therefore out-compete other memes and so persist. Human goals were both exploited by those memes, and redesigned by them. No “sighted watchmaker” was needed.

A more modern example is the design of the Internet, with its high fidelity copying and massive storage capacity. From a memetic perspective the whole system is a rapidly evolving product of coevolution between memes and their new copying machinery, with human goals being just one factor in selection.

Meme theory is a long way from fully meeting the challenges that Greenberg and Chater raise, but it has made a start. The implication is that human goals are part of the world in which memes compete and evolve, while the ultimate designer of culture is the Darwinian process of memetic selection.

Buy this book from or