This question forces us into a difficult situation. As several authors have pointed out (Deacon 1997; Dunbar 1996; Pinker 1994) it appears that either we must understand what selective advantage language gave early hominids, or we must abandon hopes of a Darwinian explanation. This is not a happy choice – if indeed it is a choice.
CHAPTER 8
Meme–gene coevolution
The mystery of language origins has apparently presented us with an unpleasant choice – abandon hopes of a Darwinian explanation or find a function for language. But this is only a forced choice if the function has to be for the genes. If there is a second replicator this is no longer the only option. I shall argue that once imitation evolved and memes appeared, the memes changed the environment in which genes were selected and so forced them to provide better and better meme–spreading apparatus. In other words, the human language capacity has been meme–driven, and the function of language is to spread memes.
What is language for?
If we want to understand the evolution of language, a Darwinian explanation is the obvious starting point. However, it has been argued that language shows no genetic variation, could not exist in intermediate forms, and would require more evolutionary time, and more space in the genome, than could possibly have been available – quite aside from the fact that its selective advantage is not obvious (Pinker and Bloom 1990). All of these arguments have been forcefully opposed. Nevertheless, they keep reappearing in various guises.
Oddly enough, the two major opponents of a traditional Darwinian approach to language origins are one of the world’s most famous evolutionary theorists, Stephen Jay Gould, and the world’s best–known linguist, Noam Chomsky.
In the 1950s, the prevailing behaviourist paradigm treated language as just another aspect of human beings’ general ability to learn. It denied any innate restrictions on what could be learned or any universal properties of language structure. Chomsky went right against this view. He pointed out that the logical structure of languages is far more complex than anyone had thought before, even though it is easily picked up by children without explicit training, and that vastly different languages actually share a common ‘deep structure’. He proposed the now familiar idea of an innate Universal Grammar. However, he has subsequently argued that
natural selection cannot explain the origin of this Universal Grammar, nor the evolution of language (see Pinker and Bloom 1990). According to Chomsky, we do have innate language structures, but they have not got there by natural selection. They must have got there purely by accident, as a by–product of something else, such as the general increase in intelligence or brain size, or by some other process that we do not yet understand. On this view there were no selection pressures for language itself.
Gould has long argued against the power of selection and adaptation in evolution in general (Gould and Lewontin 1979). He argues, instead, that many biological features evolved as by–products of something else or as consequences of natural physical processes and constraints on structure and form. In the case of language, he says, it must have come about as a by–product of other evolutionary changes – such as the overall increase in brain size (though, as we have seen, that is also unexplained), or because of some as yet unidentified physical constraints.
I do not think that such an approach can work. There is no doubt that simple physical processes can produce intricate designs, such as snow–flakes, interference patterns, or ripples on the sand of a beach. There is no doubt that physical constraints are important; the properties of air constrain the shapes of wings and tails, and gravity puts limits on height and size. By–products inevitably occur as designs change, and some of these by–products turn out to be useful and are then exploited by evolution. But these processes alone cannot account for evolutionary progress (though remember that Gould does not believe in progress either) nor for complex functional design. The only process that can produce new designs that build on and develop the old is the evolutionary algorithm (p. 11). With heredity, variation, and selection you can explain the gradual appearance of incredibly improbable things like eyes, ears, fins and tails. Language is an incredibly improbable thing, showing obvious signs of intricate design. It is no explanation at all to say that it came about as a by–product of something else or entirely because of physical constraints.
The non–selectionist arguments of Chomsky, Gould and others have been roundly criticised by Pinker and Bloom and many other contributors to a lively debate in the peer review journal
Brain and Behavioral Sciences
(1990). Pinker and Bloom argue that language shows signs of complex design for some function, and that the only explanation for the origin of organs with complex design is the process of natural selection. They conclude, therefore, that ‘specialization for grammar evolved by a conventional neo–Darwinian process’ (Pinker and Bloom 1990, p. 707).
But what is the function? A ‘conventional neo–Darwinian’ explanation assumes a selective advantage of having language. My question about
why
we acquired language now becomes ‘what was the selective advantage of having language?’ Without an answer to that question the existence of human language remains a mystery.
Pinker and Bloom’s answer is that language is designed ‘for the communication of propositional structures over a serial channel’ (1990, p. 712). But what, then, was the selective advantage of ‘communication of propositional structures over a serial channel’? Language would have allowed our ancestors to acquire information and pass it on far faster than biological evolution could achieve, giving them a decisive advantage in competition with other species, they say. But to complete this argument we need to know what biologically relevant information was to be passed on and why the use of propositional structures would have helped. This they do not explain.
There have been many answers before Pinker and Bloom’s but none is universally accepted. Some of the earliest theories revolved around hunting. Primitive man was seen as a great hunter who needed to communicate plans for herding prey or trapping them in particular places. In other words, we needed to speak in order to hunt better. A more modern version comes from palaeontologists Walker and Shipman (1996, p. 231) who suggest the function of language was to communicate ‘places to hunt; new sorts of traps; locations of water, good caves … techniques for making tools … or ways to make and keep fire’. Other theories emphasised foraging – perhaps early humans needed to communicate about the locations, nutritional value or safety of available foodstuffs. What is not quite clear from any of these theories is why humans, and humans alone, should have developed such a complex and neurologically expensive solution to the problems of hunting or foraging. For example, wolves and lions achieve clever pack–hunting strategies without grammatical language, and bees communicate the whereabouts and value of food sources with a specialised dance. Vervet monkeys have different warning cries for at least five different predators, including leopard, eagles and snakes (Cheney and Seyfarth 1990) but use no grammar or propositional structures. Presumably, our innate Universal Grammar provides advantages over these simpler systems, but there remains the question why the advantage is so great that we can communicate who did what to whom, why you couldn’t make it to the party, and the advantages of the Big Bang theory over a Steady State cosmology.
Perhaps the answer (as in theories relating brain size to Machiavellian
Intelligence, p. 74) lies in the complexity of our social lives. Our hominid ancestors were presumably social animals like their early primate predecessors and, like modern monkeys, we may assume that they could recognise and compare different social relationships and respond appropriately without having verbal labels such as ‘friend’ or ‘sister’ (Cheney and Seyfarth 1990). Social primates need to understand matters such as alliances, familial relationships, dominance hierarchies and the trustworthiness of individual members of the group. They also need to communicate. If you are maintaining a complex dominance hierarchy then you need to be able to show (or hide, or pretend to show) fear and aggression, submission and pleasure, desire to be groomed and a willingness to have sex. But emotions are notoriously difficult to talk about. Modern primates get on very well at these complex tasks by means of facial expressions, calls, gestures and other behaviours, and our language does not seem to have been designed to do that job especially well.
The function of language is gossip, says British psychologist Robin Dunbar (1996) – and gossip is a substitute for grooming. He asks the same question I have asked – only rather more poetically ‘Why on earth is so much time devoted by so many to the discussion of so little?’ In many studies he and his colleagues at Liverpool University have shown that most of our talking is gossip. We discuss each other, who is having what relationship with whom and why; we approve and disapprove, take sides, and generally chat about the social world we live in. Why?
The real function of both grooming and gossip, says Dunbar, is to keep social groups together, and this gets harder and harder as the groups get larger. Many other primates live in social groups and much of their time is taken up with maintaining them. It matters very much who is in an alliance with whom. You chase away your enemies and groom your friends. You share food with your allies and hope they will help you if you are in trouble. You come to your friends’ aid – or not, and risk their letting you down next time. Social interactions of this kind demand big brains because so much has to be remembered. You need to remember who did what to whom, when, and how strong or shaky every alliance is at the moment. You will not want to try to steal food from even a low–ranking male if he is in alliance with a stronger one. And you will not risk sex with a receptive female if another stronger male has priority. Also, as group size increases, freeloaders and cheats can more easily escape detection.
How are these complex relationships maintained? For many primates the answer is grooming, but there is a natural limit. As groups get larger
the requirement for grooming becomes impossibly high until there are simply not enough hours in the day. Baboons and chimpanzees live in groups of about fifty to fifty–five and spend up to a fifth of their time grooming, but humans live in even bigger groups. We may be able to recognise up to a couple of thousand people, but the more important group size, argues Dunbar, whether in social life, the armed forces or industry, is about one hundred and fifty. Extrapolating from monkeys and apes suggests that we would have to spend an impossible 40 per cent of our time grooming each other to maintain such large groups.
That, says Dunbar, is why we need language. It acts as ‘a cheap and ultra–efficient form of grooming’ (1996, p. 79). We can talk to more than one person at once, pass on information about cheaters and scoundrels, or tell stories about who makes a reliable friend. So, Dunbar rejects ideas about language being primarily a male–dominated function used for hunting or fighting strategies and suggests instead that it is all about cementing and maintaining our human relationships.
But the obvious question now is why there was selection pressure for larger groups. Dunbar’s answer is that our ancestors faced increasing predation as they moved out of the African forests and into the grasslands; safety in numbers would have been a valuable strategy for survival and they had already reached group sizes too large for any more grooming. But many other species have managed other ways of living in the grasslands, some in large herds and others in smaller groups. So could this pressure for larger groups really explain all the drastic and expensive changes required? Dunbar’s theory hinges on this point.
Other theories emphasise the evolution of using symbols (e.g. Deacon 1997; Donald 1991). The Harvard neuroscientist Terrence Deacon proclaims humans ‘the symbolic species’. He argues that symbolic reference provided the only conceivable selection pressure for the evolution of hominid brains – and by symbolic reference he means the use of arbitrary symbols to stand for something else. Among the advantages of symbolic communication are mother–infant communication, passing on foraging tricks, manipulating competitors, collective warfare and defence, passing on toolmaking skills, and sharing past experiences – ‘there are too many compelling options to choose from’ he says (p. 377) – but, he argues, these could only have come into play once the ‘symbolic threshold’ was already crossed. Once true symbolic communication was possible simpler languages (now extinct) would have created a selection pressure for bigger and better brains able to understand them and extend them, leading ultimately to our modern kind of language. But we had to cross ‘the symbolic threshold’ in the first place.
So how and why did this happen? For marriage, he says. According to Deacon, early hominids could only take advantage of a hunting–provisioning subsistence strategy if they could regulate their reproductive relationships by symbolic means. ‘Symbolic culture was a response to a reproductive problem that only symbols could solve: the imperative of representing a social contract’ (Deacon 1997, p. 401). On this theory, then, symbolic communication began because it was needed to regulate marriage, and then was gradually improved because of the myriad advantages it provided for other forms of communication.
If I have understood him correctly, Deacon sometimes comes close to a memetic theory. For example, he notes that language is its own prime mover and language evolution a kind of bootstrapping. He even likens one’s own language to a personal symbiotic organism. But he does not consider the possibility of a second replicator. For him ‘the transmission of genes is the bottom line’ (p. 380). Thus, he is stuck with finding the selective advantage
for the genes
of using symbols.
The Canadian psychologist, Merlin Donald, also puts symbolic representation at the heart of his theory (1991, 1993). He suggests that human brains, culture, and cognition all coevolved, passing through three major transitions: mimetic skill, lexical invention (i.e. the creation of words, spoken language and story telling), and finally the externalisation of memory (symbolic art and the technology of writing allowed humans to overcome the limitations of biological memory). His first transition – the development of mimetic skill – sounds as though it might be similar to memetics, but it is not (it is perhaps closer to ‘mime’ than to ‘meme’). Donald clearly distinguishes mimesis from imitation, stressing that mimesis includes representing an event to oneself and is not tied to external communication. He explains: ‘mimesis rests on the ability to produce conscious, self–initiated, representational acts that are intentional but not linguistic’ (1991, p. 168).