The Meme Machine (12 page)

By contrast, the skill of generalised imitation means that humans can
invent new behaviours of almost unlimited kinds and copy them on to each other. If we define memes as transmitted by imitation then whatever is passed on by this copying process is a meme. Memes fulfil the role of replicator because they exhibit all three of the necessary conditions, that is, heredity (the form and details of the behaviour are copied), variation (they are copied with errors, embellishments or other variations), and selection (only some behaviours are successfully copied). This is a true evolutionary process.

•••••

We have now established that imitation is rare and special, but just what is entailed in an act of imitation? There is considerable research on imitation in infants and children (Meltzoff and Moore 1977; Whiten
et al.
1996; Yando
et al.
1978), and some on sport, on social conformity, and on questions like whether violent television causes copycat violence (Bandura and Walters 1963) and whether suicide, vehicle accidents, and even murder can spread by imitation (Marsden 1998
b
; Phillips 1980). However, there is little on the mechanisms underlying imitation, and so I will have to speculate a little.

We might liken the process to ‘reverse engineering’, a common way of stealing ideas in modern industry. If an unscrupulous manufacturer wants to make a cheap version of the latest high–tech compact disk player then specially trained engineers tear the real thing apart bit by bit, trying to work out what all the parts do and how they can be made. With luck they can then build their own version to perform the same way -without paying royalties. But it is not easy.

Now imagine you are going to copy a simple action. Suppose I put my hands to my mouth in a trumpet–like shape, point them upwards and hum ‘de–tum-de–tum’. I would bet that, unless you were physically unable, you would have little trouble in copying me – and that people watching would agree on whether you managed a good performance or not. What is so difficult about that?

Everything. First, you (or rather some unconscious brain mechanisms) have to decide which aspects of the action are to be copied – does the angle of your leg matter? or the position of your feet? Is it more important that your hands look something like a trumpet or that their exact position is as close as possible to my version of a trumpet? Must your humming be in the same key, or only follow the same melody? I am sure you can make up your own questions. Having decided on the important aspects to be copied, a very difficult set of transformations has to be effected. You watched me, let’s say, from the side. Nothing you saw of my actions will
correspond to the way the actions will look from your perspective when you carry them out yourself. You will see only your hands from the near end of the ‘trumpet’. Somehow, your brain has to create a transformation of the action I did that will enable it to instruct your muscles to do whatever they have to do to get your action to look like mine to someone else. Now it begins to sound complicated.

It sounds complicated because it is. Imitation necessarily involves: (a) decisions about what to imitate, or what counts as ‘the same’ or ‘similar’, (b) complex transformations from one point of view to another, and (c) the production of matching bodily actions.

Once you realise how difficult this natural–seeming kind of act is, it is tempting to think we cannot possibly do it – although obviously we
do
do it. Or that a science of memetics cannot be based on something so peculiar. I am reassured by simply reminding myself that human life really is like this. We
do
copy each other all the time and we underestimate what is involved because imitation comes so easily to us. When we copy each other, something, however intangible, is passed on. That something is the meme. And taking a meme’s eye view is the foundation of memetics.

CHAPTER 5

Three problems with memes

Is Beethoven’s
Fifth Symphony
a meme, or only the first four notes?

This raises a real question for memetics and one that is worth exploring – but I do not think it is a problem. There are several such objections to memetics that are frequently raised and worth trying to resolve. I am going to consider three and will argue that all are either soluble or irrelevant.

We cannot specify the unit of a meme

Whether by coincidence or by memetic transmission, Beethoven is the favourite example for illustrating this problem. Brodie (1996) uses Beethoven’s
Fifth Symphony,
Dawkins (1976) uses the
Ninth,
and Dennett (1995) uses both the
Fifth
and the
Seventh.
Dennett adds that the first four notes of Beethoven’s
Fifth
are a tremendously successful meme, replicating all by themselves in contexts in which Beethoven’s works are quite unknown. So are they the meme, or the whole symphony?

If we cannot answer this question we cannot identify the unit of the meme, and some people clearly think this is a problem for memetics. For example, many years ago Jacob Bronowski wondered why we do not have a better understanding of social change and blamed our not being able to pin down the relevant units (Hull 1982). I have heard people dismiss the whole idea of memetics on the grounds that ‘you can’t even say what the unit of a meme is’. Well that is true, I cannot. And I do not think it is necessary. A replicator does not have to come neatly parcelled up in ready–labelled units. Since genes are our most familiar example we should look at the same issue for them.

Defining a gene is not easy and in fact the term is used quite differently by breeders, geneticists and molecular biologists because they are interested in different things. At the molecular level, genes consist of sequences of nucleotides along a molecule of DNA. Names are given to different lengths of DNA, such as a codon, which is a sequence of three
nucleotides, or a cistron, which is a sufficiently long sequence of nucleotides to provide the instructions for building one protein – with a start symbol and a stop symbol. Neither of these is necessarily passed on intact in sexual reproduction and neither corresponds with what we think of as the gene ‘for’ something. DNA provides instructions for protein synthesis and it is a long way from there to having blue or brown eyes, finding men more sexy than women, or having a flair for music. Yet it is these effects of genes that natural selection gets to work on. So what is the unit of the gene?

Perhaps there is no final answer. One useful suggestion is that a gene is hereditary information that lasts long enough to be subject to the relevant selection pressures. A sequence of DNA that is too short is meaningless -it lasts almost indefinitely, being passed on identically from generation to generation but taking part in countless different kinds of protein synthesis and countless different phenotypic effects. A sequence that is too long does not survive through enough generations to be selected for or against. So some intermediate length has to be chosen, and even this varies with the strength of the selection pressure (see Dawkins 1976; Williams 1966).

This intrinsic uncertainty about just what to count as a gene has not impeded progress in genetics and biology. It has not made people say, ‘We cannot decide what the unit of the gene is so let’s abandon genetics, biology and evolution.’ These sciences all work by using whatever unit they find most helpful for what they are doing at the time.

The same logic applies in memetics. Dennett (1995) defines the units of memes as ‘the smallest elements that replicate themselves with reliability and fecundity’ (p. 344). A blob of pink paint is too small a unit for memetic selection pressures to apply – to be enjoyed or disliked, photographed or thrown away. A whole gallery of paintings is too large. The single painting is the natural unit for most of us and that is why we remember Van Gogh’s
Sunflowers
or buy postcards of Edvard Munch’s
The Scream.
Styles of painting, such as impressionism or cubism, can also be copied and therefore count as memes, but can hardly be divided up into units. A single word is too short to copyright and an entire library too long, but we can and do copyright anything from a clever advertising jingle to a 100000-word book. Any of these can count as memes – there is no right answer to the question – ‘What
really
is the unit of the meme’.

I might have argued that four notes is too short to be a meme but everyone’s favourite example shows I am wrong. If a musical genius picks on just the right four notes, starts a wonderful symphony with them and
has the luck to have his work survive into an age of mass communication – then his four notes can be heard and remembered by literally billions of people. I am sorry if you are one of them and now are not able to get these four notes out of your mind.

This problem – why can’t I get that tune out of my mind – provides a good example of memetics at work, and I shall use it to show that the size of the unit makes no difference.

Why do tunes sometimes just go round and round in my head and will not go away? Why do we have the sort of brains that do that? What possible use is it for me to spend all day singing ‘Coke Refreshes you Best’, or the theme tune from
Neighbours?
The answer from memetics is that it is no use at all to me – but it is of use to the memes.

Memes are replicators and if they can get themselves copied they will. The imitating machinery of the brain is an excellent environment for copying tunes. So if a tune is memorable enough to get lodged in your brain and then passed on again then it will – and if it is really memorable, or singable, or playable, it will get into a lot of brains. If it turns out to be just what some TV producer needs to start her latest soap opera then it will get into even more brains, and every time you start humming it there is a chance that someone else will hear you and you will set them off. Meanwhile, plenty of other tunes are never heard again. The consequence of all this is that the successful ones increase in the meme pool at the expense of the others. We all get infected with them and they are stored in our memories, ever ready to be activated and passed on to anyone who has not got them yet. All this singing is not for our benefit, nor for our genes’ benefit. Being haunted by horrible tunes is just an inevitable consequence of having brains that can imitate tunes.

Note that this argument works regardless of all the specific reasons why one piece of music may be singable or likeable and another not. Those reasons might include, for example, innate preferences for certain sounds, the pleasure to be found in predictability and unpredictability of sounds, or overall complexity. Gatherer (1997) has explored the development of jazz in terms of the adaptiveness of its component parts, looking at complexity, memorability and the effect of the available technology at different times. Simple melodies are easy to remember but may not be interesting enough for people to pass on. Complex improvised music can evolve but may only survive in a community of trained musicians and listeners, while even more complex music may be simply too difficult to remember and so fail to be replicated, even if it can be enjoyed. Memetics in the future may discover what makes for successful replication in music. It may find out how different kinds of music fill different niches, such as
specialised replication among minority groups, or short–lived mass popularity. But note that none of this matters for the simple argument I am making here. That is, that any catchy tune that gets you to rehearse it in your head will get passed on, and so we will all come across such tunes and be in danger of ‘catching’ them.

Memetics thus provides a simple and obvious explanation for those irritating tunes that go round and round in our heads – as it did for why we cannot stop thinking in general. The tunes are like weeds and just tend to grow. Does it matter what we count as the unit of a meme in either of these cases? I say no. The competition to grab any spare brain power will go on regardless of the way we might decide to divide the competing instructions. The meme is ‘whatever it is that is passed on by imitation’. If your irritating humming at work passes on both verses of Blake’s Jerusalem to the rest of the office then the whole inspirational song is the meme. If you infected them only with ‘Da, da, da, dum’, then those good old four notes are the meme.

We do not know the mechanism for copying and
storing memes

No we do not. The fact that we now know so much about how DNA works can easily lead us to imagine that we need that level of understanding for memetics – right away. I do not think this is so. Don’t forget how far evolutionary theory got before DNA was even heard of. Darwin’s
Origin of Species
was published in 1859. It was not until the 1930s that genetics and natural selection were brought together (Fisher 1930); not until the 1940s that other areas of science were brought into what is sometimes known as the modern synthesis, leading to neo–Darwinian theory; and not until the 1950s that the structure of DNA was finally discovered (Watson 1968). In the first century of Darwinism an enormous amount was achieved in the understanding of evolution without anyone having any idea about chemical replication, the control of protein synthesis or what on earth DNA was doing.

The memetics we build at the very end of the twentieth century will doubtless appear inept in another century’s time, but that is no reason not to begin. We may get a long way with the general principles of memetic selection without understanding the brain mechanisms it relies on. We can also make some educated guesses about those mechanisms based on what little we do know.

First, we may assume that, at least at some phase in their replication, memes have to be physically stored in brains. As far as storage is concerned neuroscience is making great strides in working out the biological basis of memory. Artificial neural networks have demonstrated that many of the features of human memory can be simulated in computers. Work on synaptic transmission, long–term potentiation and neurotransmitters is finding out whether real brains do anything similar. If they do, we may guess that human memory probably works something like this (for example, see Churchland and Sejnowski 1992).