This site is not maintained. Click here for the new website of Richard Dawkins.

Meme Theory, Zahavi's Handicap, and the Baldwin Effect

This is about meme theory, and why I'm not so sure about its soundness. I admit, up front, that the idea contains elements that have a lot going for it, such as the epidemiology of ideas and how they can slip past cultural barriers, rendering them blurs rather than hard lines. It also chimes with the idea that some ideas do better than others, and for identifiable reasons such as utility, its ability to appeal to our pleasures, and its compatibility with other ideas.

What I wonder about is the status of an idea as a replicator. To be a replicator, it would have to create copies of itself. To be a replicator under selection, it would have to create phenotypes which decide whether it out-competes rivals or is out-competed by them.

If memes are to have any factual basis, then they need an identifiable unit by which their status as a replicator can be judged. I think it's safe to say that nerves or neural nets in the brain currently provide our best candidates for this position, as they are the information-carriers akin to the polynucleotides of DNA and RNA. After all, the neural net representing the letter 'A' in my head is most likely to be identical to the neural net representing the letter 'A' in your head, ignoring complicating factors such as the connotations of the letter. We can agree between us that an 'A' is an 'A' is an 'A', whether I write it 'a' or 'A' on paper. This itself could be repesented by sub-nets representing '/' and '\' and '-', which combine to make 'A'. It isn't important so long as you acknowledge that there will be some universality between brains capable of learning the alphabet.

When you break it down to these basic levels, they will be universal to everyone. Sure, an Indian or a Chinese person would have a different alphabet to me, but as Chomsky et al. would point out, the fact is that we have an alphabet, and an innate understanding of grammar and language functions common to all of us. Conversely, if I try to teach you about 'postmodernism', then even if you don't agree with me on all the details, it's guaranteed that there will be common ground somewhere, even if it's only on how you spell it. Language couldn't work at all if this were not so.

The rub, though, is that these basic universal ingredients were built in by genes. By this, I don't mean that genes gave me an understanding of the letter 'A', but they certainly gave me the basic ingredients and the language instinct that meant I could learn the Western alphabet in my childhood. This is true for you too. So if I were to teach you the letter 'A', that in itself presupposes that your genes gave you the capacity to learn the letter 'A', even if confounding factors like having your own foreign language made it a little harder for you. And that foreign language's alphabet itself would be made up of those instincts and built-in presuppositions. The very malleability of the brain was built in by the genes. A polyglot can also arise who knows both Mandarin Chinese and English, or both Mandarin Chinese and Hindi, so languages don't seem to "compete" for the same loci in the brain, which raises questions about memetic competition.

Other parts of culture can be broken down in this way as well. Art, for instance, contains elements of technical and motor skill, so the basis for coordinating those actions that lead to the perfect brush stroke on canvas requires brains that are flexible enough to match the technique to the tools, the tools to the canvas, and so on, which themselves rely on rules of perception and on the brain's built-in ability to categorize different materials according to basic distinctions like colour, texture, shape, and whether it's solid or liquid.

How do memes fit into this picture? The accepted mechanism is that brains with meme hosts send out signals - say, verbal communication or a motor demonstration of some skill like carving - which reach other hosts, and manipulate their minds into rearranging themselves such that they match the original meme as a copy. These infected minds then do the same to others.

The problem with this idea is that one output signal could give rise to six or seven responses in people. At the first, people may simply not notice some detail of it, potentially jeopardizing the structure of the copy meme. At the second, people can simply misinterpret what the other is doing or saying even if they've watched and listened to the whole thing. At the third, some people can simply reject it, and maybe forget about it later. These objections aren't fatal to the idea of the meme, but they certainly suggest that it's nowhere near as refined as the gene copying mechanism. And when a replicator makes far more mistakes in its copying, its status as a replicator becomes increasingly dubious. A change from X to X to X... to X to Y is more like the replicator than a change from X to Y to Z to A, which simply looks like another causal chain.

This creates a paradox: when you try to make a meme bigger than the universal ingredients already in the brain, it becomes increasingly less replicator-like; however, when you try to make it smaller so that it hits the universal ingredients, it becomes clear that it's all the work of the genes that installed them. As in the case of group selection, we end up with an illusion of another replication process when genes are the real manipulators behind it.

Take this thought experiment, which I found quite useful in making the distinction: you read a little about something, say postmodernism, and then shut yourself in a room for a bit. You have a pen and a lot of paper, and you sit down and write out your idea of what postmodernism is. You read what you've written. You think it doesn't look right, so you edit it and write an improvement. You reread it, edit it, and so on.

At what point can we say that there are distinct replicator generations between memes? Is this just one generation (all in my head, with the assistance of paper)? In which case, look at the huge number of mutations it's accrued before it's even gone to another brain. When you have so many mutations in generation one, that suggests we are not dealing with a replicator at all. We have something closer to the X to Y to Z to A model. Or is each round of travelling from head to paper to head a generation? In which case again, the changes to the meme are clearly so different with each draft I write that it could hardly be called replication. The postmodernism meme of generation one has mutated in numerous details by the time you write generation three. And I don't think it is a replication process if it "copies" by changing itself. Such a thing falls again into the X to Y to Z to A model.

It might be argued that any other memes in my head are already confounding this process of replication. However, this still doesn't erase the fact that the idea is not replicating, and this same problem plagues any and all memes, not just this one. Replace 'postmodernism' with 'Roman Catholicism' or 'Bolshevism', and you get exactly the same problem.

Or it may be argued that I'm taking too broad a view of the meme, akin to critiquing the high mutation rate of genomes when I should be focusing on genes. But this objection hits the other end of the dichotomy: the genes supplied any and all universal nerve structures when the brain was being built. All the replication work was done long ago in genetic evolution. Ideas in brains, far from being independent forces, are the flotsam you get when you rearrange what's there. When I write about postmodernism, while the idea itself is fairly recent, the pieces that make it up date back across the history of culture and to biological processes of both every predecessor who handled them and of the current holder who favours it. Postmodernism ends up being a rearrangement of preexisting ingredients, set to a context.

The very mechanism by which my thoughts influence your thoughts seems to be a replicating mechanism, still. After all, genes are all made up of the same four-letter alphabet, and yet can be rearranged to make different genes that replicate when they meet random As, Cs, Gs, and Ts. Maybe memes are similar, simply using an intermediate (like writing or the spoken word) to get from A to B? Unfortunately, this becomes the inheritance of acquired characteristics, because the product of the meme 1 needs to be reverse-transcribed to make meme 2, and as anyone trying to communicate with a distraction knows, there's a lot of room for noise to ruin the signal. If I gave a lecture on something as simple as evolution by natural selection, it's guaranteed that the ideas created in each audience member's heads would be different to the ones found both in my head and in their immediate neighbours'. It's no coincidence that oral traditions mutate faster than written ones. And again, where there's common ground, it can be traced back to common brain features provided by our genes in the first place.

When a brain receives an input, it has preexisting machinery to deal with it. Brains don't simply absorb what's going on outside - they have ready-made equipment for dissecting it, equipment which evolved via natural selection of genes. To reconstruct anything coming in is an exercise in inverse optics, or inverse harmonics etc., and so the structures need to be there in the first place. But they were put there in the first place by genes, and so when I teach you about 'postmodernism' and all you learn is how it's spelled, is it really replication if the letters and the ability to arrange them are already present?

Certainly, the combination of these two elements is going to be the same in your head as in my head, or at least translatable, but your brain has to make an active effort to rearrange these elements. They don't have the spontaneity of replication which a replicator molecule has. And these mental replicator mechanisms - say, imitation - were installed as the flexible phenotypes of replicators (genes), so the parallels in structure from one brain to another are already accounted for by genetic factors. That suggests the mechanism of idea replication is an illusion because the so-called "replication" process is about as replicator-like as a builder drawing up a blueprint by examining the house that was built by another blueprint. Neither the house nor the blueprint is really replicating, because if left to their own devices they do nothing. They are products which brains manipulate, and brains are the robotic circuitry set up by genes.

So, quite apart from the high mutation rate that this Lamarkian evolution allows, it shows that ideas don't make copies of themselves any more than a paper makes a copy of itself - the photocopier, and in the brain's case the perceptive and motor neurons, have to do all the work. I think this is the biggest objection to memetics, and the main reason I'm sceptical of it.

As a side note, I raise this scepticism because it seems to me that a problem with memes is that they're restricted to one species, when you'd think the basic nervous systems are already in place for many other lineages. Yet, only a handful - such as chimpanzees and songbirds - actually show anything we might call culture. The one species that shows sophisticated culture (i.e. Homo sapiens sapiens) is the one species that's exploded with it and the one species with the excessively large and complicated brain. Yet, replicators like DNA and RNA are comparatively simple in their shape. This isn't a knock-down rebuttal, but it makes me even more wary about the soundness of memetics, and it certainly has a more-than-passing resemblance to the spurious logic of group selection; if it's a viable mechanism for some form of evolution, then why is the phenomenon it's invoked to explain so rare as compared to similar phenomenon?

Now, it's all very well to be sceptical about memetics, but when you knock down a theory, it's common courtesy to offer an alternative. How do we account for the massive explosion of culture that stares us in the face? This needs a special explanation, both for its rarity and for its surprising fecundity. I propose that all we really need are the tools of genetics, which gave rise to the bodies and brains that interacted to give rise to culture. Just as biology collapses into genetics which collapses into organic chemistry, so culture collapses into biology.

I justify this on the following grounds. In principle, every cultural artefact from an idea to an object owes its existence to biological logic. Our various courtship and mating rituals are nonetheless all based on the obvious biological imperative. Our social contacts are all based on the logic of interorganism cooperation and/or exploitation. Our tendency towards social hierarchies, our tendency to show off by unfakeable advertisements of our phenotypes' prowesses, our social instincts, our competitions for identified and valuable resources, our nepotism, our conflicts between exploration and safety, and our desires for status, partners, and caregivers... it's not just that there are so many parallels with known biological facts. It's that they always fulfil roughly the same purpose. Our attraction to beautiful sounds, sights, and experiences fuel our leisure activities, yet the tendency to such attraction was built-in precisely because it would lead organisms to such specific things (things that would have been salient to survival, individual success, and reproduction). Even the cut between "high art" and "low art" has something of the social hierarchy and the Zahavi-Grafen handicap principle (as Zahavi would've put it; "Look at me, rivals and potential allies and potential reproductive partners, I'm so rich and well-fed that I can afford to devote my time to pursuits completely useless to my biological imperatives, because I'm just guaranteed to win at those biological imperative games anyway").

Yet humans are culturally extremely varied and are genetically almost uniform. Surely, this must contradict my claim that culture traces back to biology? Only if you think that genes rigidly code for inflexible behaviour patterns. The problem goes away when you allow genes the ability to build structures that modify themselves based on external conditions. Neuroplasticity is one result, in which the genes build the brain from the ground up, but with materials and structures that are self-improving once released into the world. Conditional strategies can be built up in this way, and then more conditional strategies can be built on top of them. They can be cross-referenced with other strategies, further increasing the subtlety of responsiveness. It doesn't take long to see that a single body can then be highly sophisticated enough to change in an exponentially-increasing number of ways when faced with changes in the local environment. This is the first step.

The second step is to use the concept of the extended phenotype, so that a gene's effects can spread much further than its vehicle. Beaver dams, for instance, are one such consequence of this, improving with each favourable mutation as the genes build better beaver bodies to make better wooden dams. Ants' abilities to construct huge structures for their nests are another example, and can lead to increasingly elaborate nest structures with ventilation systems and chambers devoted to specific tasks. Most exciting is when this is turned on other organisms, as when ants and accommodating plants co-evolve to provide each other with safety from competitors and ready-made housing quarters respectively. The relationship need not be between species, either - the strong bond between mother and baby in mammals is likely to have been mutually reinforcing in evolutionary time.

The third step is to combine the two: to have complex conditional strategies and extended phenotypes. In humans, this might have gone something like this:

  1. The extended phenotype is our ancestors' abilities to manipulate the world around them - to use tools and crude constructions like modified caves for shelter. By modified, I mean that it could be as modest as moving some stones around to make more room, or using animal skins to soften the floor. The exact details aren't important.

  2. This sets up a selection pressure such that the mind can reverse-engineer what these tools are for. This requires conditional strategies such as "Brain, if you detect an object of unusual shape with X and Y qualities, it is a tool. Try to work out what it was used for." Genes could favour individuals that were born with some understanding of the process, in something akin to the Baldwin Effect. This gives the illusion of Lamarkian evolution, but the difference is that the phenotypes which allow for a susceptibility to a given behaviour in the face of a given scenario (such as the recurrence of specific tools from generation to generation) feed back into genetic success as usual.

  3. Following 2, the brain becomes more sophisticated at conditional strategies, and bit-by-bit the mind evolves an ability to categorize artefacts as distinct from, say, living thing or environment. This is not as far-fetched as it sounds: it could easily co-evolve with tool-using or environment-manipulating behaviour, and the likes of Dennett and Atran suggest that the human mind has an in-built design stance to deal with these things. It would be favoured in an omnivorous species with many opportunities to exploit tool use, for instance.

  4. This produces more artifacts, some with enough permanence to last beyond a single generation. Humans who simply pick up what is there and modify it, usually by improvement, prosper over those with cruder, clumsier, or less efficient tools because, say, they have more food reserves to access and to draw on. This merely requires a general conditional rule like "If you see artifact, use it. If it causes frustration, examine it and make changes." This is a conditional strategy, but it already allows for cultural lineages. The same tool is modified again and again, or another person works out what the first is for and makes another one, and then modifies that.

  5. By simple genetic recombination, people with different thinking styles will be produced. One person may make a change in one direction, another will make a change in another direction. Simple variability of the individuals manipulating the stuff allows cultural lineages to diverge even within individual generation times.

  6. Populations of people spread across larger expanses of land, becoming isolated from each other. This allows even more opportunities for deviation.

  7. Increased efficiency at tool use, modification of existing cultural ideas, and better reverse-engineering abilities of brains (note my talk about the mechanisms involved in "replication" above) allow cultures to mutate and spread far quicker than gene populations evolve.

  8. These may be facilitated by other co-adaptations, such as the evolution of language. I won't expand on this point, as I've written a fair bit already, and the number of other co-adaptations is large (I've listed a few already above). I think you've got the gist of this one.

  9. Because cultures can change within a lifetime and be taught from brain to brain, cultural changes and progressive improvements can increase exponentially, as each split of differences between people begets more splits. Every time it doubles, it creates more cultures until we reach the bewildering variety that we have today.

  10. Some cultural innovations become more complicated, and each time a new innovation allows for a revolutionary new change, it is followed by a swift multiplication of new cultures. Agriculture, domestication, and the invention of the wheel are all examples.

The overall result is that an ability to manipulate the environment, coupled with conditional strategies of what you find, can do a fair amount of work in explaining the bewildering phenomenon we call culture. Culture relies heavily on multiple biological factors, though, so my explanation is not a comprehensive picture. Let's just say that the distinction between biology and culture is simply one of how extended and complicatedly conditional a phenotype is, as one generation creates external changes that are detected and modified and left again by another generation, or even by the same generation later. I think this has a lot in common with how cultures work, and it doesn't require giving culture more powers than the evidence can justify. We should also be wary of the fact that we're trying to explain something that effects ourselves, so we must be cautious of what we take for granted regarding culture.

It's just so astonishing how many parallels with biological ideas you can find when looking at culture and dissecting it. Everything to do with culture is rooted in human psychology, which is rooted in neuroscience, somatic bodies, and the genes that built them. I think this is a far more accurate picture than, say, that drawn up by group selection or memetics. I could be persuaded otherwise, but an explanation of culture that relies on a combination of multiple factors has the advantage of explaining why it is so rare - because it requires hitting on just the right combination, and our lineage was a meeting of those combinations.

In summary, then, I suspect that memetics is not the best theory to explain cultural change, as it has problems assigning any idea with a replicator status. I also suspect that virtually every difference in culture can be traced back to biological (and therefore genetic) factors, and that this works both because aspects of culture have such astonishing parallels with biological facts, and because genes are far more flexible, and their phenotypic consequences far more complicated, than we often give them credit for. I also propose that cultures can be mostly accounted for mostly by the pulls of complex conditional strategies, the extended phenotype idea, and simple variation and geographical isolation.



Comment RSS Feed

Please sign in or register to comment