copyright ©2000 by William H. Calvin and Derek Bickerton
The nonvirtual book is available from amazon.com or direct from MIT Press.
The nonvirtual book is available from amazon.com or direct from MIT Press.
Webbed Reprint Collection
William H. Calvin
University of Washington
Email || Home || Publications
Reciprocal Altruism as the
Predecessor of Argument Structure
I=m about to do something that at first sight may seem paradoxical and perverse. I spent a chapter claiming that social intelligence had nothing at all to do with the emergence of language. Now I=m going to claim that syntax, which I and others see as the distinguishing feature of human language, derives from one of the most important components of social intelligence. What=s going on?
The only reason my moves might seem paradoxical is because people don=t always distinguish clearly enough between the emergence of protolanguage and the emergence of syntax. These are two entirely different things, even if one did eventually lead to the other. If I=m right, the events weren=t even close in time. It=s not only possible, it would have been pretty easy for a species only just ahead of the other apes to get along with words (or signs B makes no difference) and nothing else for a million years or two. Evolution is a conservative thing. It doesn=t need constant novelty. Good enough is good enough. At least, until something better comes down the pike.
So there=s no inconsistency in proposing that, although social intelligence had little to do with the birth of protolanguage, it had a lot to do with the birth of syntax. But I might as well warn you, the story of how syntax came to be is not a simple one, not the kind you can boil down into a thirty-second soundbite. To follow it, we=ll have to follow a long and tortuous trail, through social intelligence, the birth of altruism, cheating, episodic memory, and then on into some aspects of how the brain works that definitely don=t form part of Brain 101. So we might as well start with sex.
Sex gets people=s attention. And of course it=s central to evolution. It makes us procreate, and procreation leads to variation, and variation is necessary for natural selection; otherwise there=d be nothing to select from. As if that weren=t enough, differential interests in sex (males want to spread their seed as widely as possible, females B who do the real work B want to limit their offspring to the best and brightest) drive a great deal of animal behavior. Female preference for particular types of mates determines many creatures= physical form and behavior (peacocks= tails, bowerbirds= bowers, battles between rutting stags, and on and on).
We modern humans tend to forget, when we pursue our romantic inclinations with little restraint beyond the willingness or otherwise of our prospective lovers to cooperate, how lucky we are in comparison with a number of other mammalian species. Take bull elephant seals, for example. In each troop of these masses of blubber that can be seen decorating the Californian coast, the largest and most aggressive male (the alpha male, as ethologists would call him) virtually monopolizes the females in the group. Any male seal who wants to break this monopoly has to choose a moment when the alpha male=s attention is otherwise engaged. If not, his first move in an amorous direction will be met with extreme physical violence. No matter how receptive a female bull elephant may be to the advances of another suitor, a head-on charge from an alpha male discourages all but the most reckless. It has been estimated that up to 85 percent of all copulations in a group are performed by the alpha male. As there are several other mature males in each group, this means that large numbers of male elephant seals mate extremely rarely, and indeed some of them may go to their graves as virgins.
This arrangement may be advantageous to elephant seals as a whole, as it ensures that only the toughest and brawniest pass on their genes to succeeding generations. But then again it may not. Brawn=s fine, but what about brain? Does the sexual dominance of alpha males help make elephant seals smarter? Probably not. Certainly elephant seal brains are tiny relative to their immense body size, and they show no sign of doing much beyond lying on rocks and catching fish.
Smarter species are unlikely to put up with forced celibacy if they can find a way to escape it. And the easiest way, perhaps the only way of dealing with a member of your group who is stronger than you, is to form an alliance against him. Two heads are better than one, and so are two bodies and two sets of teeth. A pair of medium-tough monkeys can take on any alpha male.
Unfortunately, there=s still no direct evidence of a correlation between forming alliances against alpha males and reproductive success. There is, however, a correlation between neocortex size and both frequency of tactical deception and the social skills involved in competition between males. That=s to say, in species with bigger brains, the sexual tyranny of alpha males is typically circumvented. Because those species are precisely the ones in which male-male alliances are formed, this does provide some indirect evidence that the motive for alliance formation is primarily the desire to spread one=s genes.
Two animals who form an alliance against the alpha animal are not necessarily kin. Yet recent research does seem to indicate that while an animal may do things that help to perpetuate its own genes (and hence may help close kin, who carry a proportion of those genes), it doesn=t, normally or naturally, do things that will benefit some set of competing genes. (Nothing mysterious about this B if there had been animals that gave the genes of others preferential treatment to their own, those animals would surely be extinct by now). So how could altruism have developed? Only if helping someone else indirectly helps you B if, when I scratch your back, you scratch mine.
And that=s how reciprocal altruism, father of the more selfless kinds, was born.
The term was coined by biologist Robert Trivers. For a long time people had puzzled over the existence of altruism among humans. Why were some of us willing to sacrifice our own interests, occasionally even our lives, for others? This question became even more pressing with the rise of Darwinism, the decline of belief in the supernatural, and a growing acceptance that all living organisms are of their nature irredeemably selfish.
As Trivers showed, and as many subsequent ethological studies confirmed, reciprocal altruism was the answer. You may wonder how selfish behavior could spawn selfless behavior. But even the cynic must admit that, on occasion, humans do sacrifice themselves for others, even for members of other species, without hope of reward. That kind of altruism has a slightly different story, a story mediated by language and featuring duty, responsibility, and ideal forms of behavior. (It would take us too far from our path to tell that story here B we should merely note that this broader type of altruism could never have come about if the more self-oriented types had not preceded it.)
But wait a minute, you may say. What do you mean by saying Atake us too far from the path@? Aren=t we a long way from the path already? What on earth have elephant seals and sex alliances and reciprocal altruism to do with language?
As a matter of fact, reciprocal altruism contains the roots of many of the things we hold most dear B morality, democracy, and yes, even language (or at least syntax). For the morality and democracy bits, you should read Frans de Waal=s delightful (and extremely important) book, Good-natured. Nothing we could say about them would improve on that.
That anything as abstract as syntax should have come from reciprocal altruism may well seem surprising. For what I=m proposing here is that the practice of reciprocal altruism created the set of abstract categories and structures that, once they were joined to a structureless protolanguage, yielded the kind of syntax that all modern human languages exhibit.
Consider what primate social life was and is like. Primates characteristically live in small groups, seldom exceeding a hundred individuals or falling below a dozen or so. In other words, these groups are small enough for each individual to know every other individual quite intimately. Primate social life, as many excellent studies vividly show, is an intense and continuous experience. Individuals are competitive, flexible, and opportunistic; they won=t succeed if they don=t keep on their toes. Alliances based on reciprocal altruism play a vital part in helping individuals to succeed. Describing a troop of baboons, Shirley C. Strum observes that
As this extract shows, the building of such alliances takes time, and exploiting them to the full takes more time. It places a heavy load on the memory, too, if you have to remember over weeks and months who you owe and who owes you.
Alliances will not last without constant work to maintain them. You cannot use someone to gain your own ends and then just ignore them. You=d feel used if you were treated like that. And the more you learn about primates, the more you appreciate that their emotions do not differ substantially from ours.
One vital factor in group cohesion and bonding between individuals is mutual grooming. Primates of many species spend hours examining one another=s hair, far more time than would be justified if the practice was no more than a means for removing parasites. Although any group member may groom any other, members of alliances groom one another much more frequently than non-allied animals. But in any alliance, the question must be: does A groom B much more frequently than B grooms A?
All of us have been, at one time or another, tempted to cheat B to give in return less than we have received. Cheaters may be gross or subtle. Gross cheaters accept your favors and do nothing in return B they=re easy to spot. Subtle cheaters are another matter:
How can you recognize a subtle cheater? Cheaters, subtle or otherwise, must be unmasked, for if cheaters can get away with it, who will fail to cheat? And if everyone cheats, then the system of reciprocal altruism breaks down into a war of all against all.
So it must be possible for any member of an alliance to detect whether his partner is cheating or not. If neither partner is cheating, the alliance can be maintained indefinitely. If one partner is cheating, then the other is wasting valuable time and effort that could be better spent on a different partner. Animals with partners who don=t cheat will do better (suffer less stress, win more fights, gain more sexual access, and thus probably leave more progeny behind) than animals whose partners cheat and get away with it. If animals originally fell into two classes (good cheater detectors and not so good ones), then over long periods of time, the good detectors would gradually squeeze out the poorer ones. Good cheater detection would then form a fixed part of that species= genetic recipe.
So it should be possible, in any species that typically practices reciprocal altruism, for animal B to know whether he grooms his partner A more or less often than A grooms him. And grooming, although it=s among the most important primate activities, is by no means the only activity in which reciprocal altruism comes into play. Take meat-eating among chimpanzees, for example. Chimpanzees, like many human hunter-gatherers, like to hunt young monkeys as a valuable source of protein; this practice probably goes back to the common ancestors of chimps and humans, if not earlier. In general, chimpanzees do not share food, but when one or two of them kill a monkey, they share the meat with most chimps that beg hard enough for a piece.
There is a good reason for this. Successful hunts, unlike finds of fruit or nuts, do not happen every day. Because they are such rare occurrences, even a successful chimp would go a very long time without meat if meat wasn=t shared. Chimpanzees share food for the same reason they carry out any other act of reciprocal altruism B if they didn=t, they wouldn=t get the benefits they otherwise could.
And so it goes. In addition to remembering who groomed who and how often, who gave meat to who and how often, chimpanzees and other primates have to keep track of how often partners stood by them in fights, how often they ran away, and doubtless other types of behavior too. In other words, to detect cheaters and protect their own interests, they are obliged to develop a social calculus that weighs their own acts against the acts of others.
What was necessary for such a calculus? It would require at least the following ingredients: (1) an ability to distinguish individuals of the social group, (2) an ability to distinguish different types of action, and (3) some kind of abstract representation of the roles of participants in actions.
The first is essential because you have to know who=s who before you know who to form alliances with, and that=s a prerequisite for any kind of reciprocal altruism B even for the one-off AI-just-did-something-for-you-so-now-you-owe-me@ kind, let alone for the stable alliances that characterize so much primate social life.
The second B distinguishing different types of action B is essential because if you can=t distinguish, you can=t keep track of whether you are doing favors for your partner more often than your partner is doing them for you. But wouldn=t it be simpler if you just had a single category called AActions requiring reciprocity@? After all, the nature of the actions isn=t relevant, all that matters is making sure the balance stays in your favor. Well, just try to imagine how you could possibly set up such a category in the brain. It=s too vague, too abstract, too general. On the other hand, many primate species B not just apes, but even some monkeys, such as macaques B have assemblies of neurons that respond to particular actions, like some other primate grabbing for them. If individual actions can be represented in the brain, and if some are already represented, it shouldn=t be too hard to set up the mechanisms for telling whether you did X to someone oftener than someone did X to you.
The third feature B distinguishing roles of the participants B is essential because reciprocal altruism isn=t like a kinship system, where a given individual always has the same relationship to others. Sometimes you are grooming, sometimes you=re being groomed. Thus the categories have to be abstract enough to cover a variety of individuals, each of whom will perform the role concerned at different times. Roles like agent (the performer of an action) and theme (who- or whatever undergoes the action) will then serve like tags, to be attached to any individual whose role, on a particular occasion, happens to fit.
Now consider episodic memory. The exact status of episodic memory remains somewhat controversial. But although the relationship between episodic and semantic memory may be uncertain, nobody doubts that humans have the capacity to remember events in the order in which they occurred, and remember too, for most events, who or what performed the action and who or what it was performed upon. Some believe that episodic memory is shared by many species. If it is shared by other primates, then we can suppose that every time primates record an event in memory, they tag each participant in that event with the role of agent, theme or goal (whoever the action was directed towards). Thus they too, when they remember the event, will remember exactly who did what to whom.
With the apparatus I=ve described, you can keep track of individuals, remember how often they=ve done things for you and how often you=ve done things for them, see whether there=s any imbalance there, and determine whether that imbalance goes against you. (I can=t imagine monkeys lying awake worrying over whether the imbalance went the other way, whether they had done enough for others, but a more conscience-stricken species could use the calculus for this too.) In other words, with that apparatus you have a mechanism powerful enough to detect cheaters and freeloaders and make reciprocal altruism really work as the lynchpin of your social life.
WHC: You=ve persuaded me that what reciprocal altruism (RA) needs, besides identification of individuals, is some mental bookkeeping for debts: AI supported Alpha the last time he was challenged by Beta, so maybe I can get away with taking this morsel of food, after all.@ From such beginnings, a concept of debt could have later developed: AAlpha owes me.@
But when do debts need to be communicated, in ways beyond the body language that signifies reluctance to re-engage in sharing or supportive behavior? A vocalization for ABut you owe me!@ would be an interesting public way of labeling an individual as a freeloader. (I don=t like the term Acheater@ B it implies an implicit promise to reciprocate and that=s a bit fancier a concept than we need here, like using adultery when promiscuity would suffice.) If such a cry reduced the tendency of other individuals to cooperate with the individual so labeled, it would have some force in evolutionary terms.
RA, after all, is not going to get started with an explicit set of promises of reciprocation. It=s more likely to be a matter of mutual grooming or food sharing with individuals until a significant imbalance arises, then reproaching the other or withdrawing from further cooperation. From a sharing foundation, coalitions might develop that would subvert the usual dominance hierarchy.
Getting started down the RA path requires individual identification, an ability to categorize advantageous acts (and who they advantaged), a memory for the balance of debts, some level of an indiscriminate tendency to share (to get things started), and a sufficiently rich environment so that the occasional losses are not serious compared to the rewards of cooperation (grooming, food, sex, dominance). There might be a hierarchy of cooperation: mutual grooming is cheap, food sharing less so, and support in conflicts likely to develop only upon a successful foundation of other cooperation because of the risks of being bitten or losing dominance rank while actively assisting another in a chase or fight.
Altruism has always been mixed up with the whole issue of group selection. About thirty years ago, many people turned their backs on group selection because it looked, mathematically, as if freeloaders could always swamp the boat B that even if you had, by chance, a subpopulation form up with a high percentage of individuals having a genetic tendency to share, or to perform other altruistic acts (say, breaking up fights between unrelated individuals, risking injury to do so), the situation wasn=t stable: it would always deteriorate as the freeloaders received more than they gave back, diluting the altruistic genes because their reproduction was enhanced. It wasn=t an evolutionarily stable strategy.
Trying to see the Abig picture@ can mislead you. In this case, there were several errors: aggregating all subpopulations together to talk of a grand average, and focusing on what=s stable in the long run without taking into account what shorter-term fluctuations could generate via pumping.
First, suppose that inbreeding subgroups with altruism out-reproduce subgroups that lack it. Yes, even though those altruistic groups are going to backslide in the long run, they can still B in the short run B raise the overall percentages of altruism when such a group expands faster than the other groups. And it=s not just a mathematical abstraction: allow the subgroups to merge into one big group, and the altruistic genes are more than before the prior split into separated subgroups. So the overall short-term picture can look very different than the long-term one, if only populations split occasionally into subpopulations and later merge.
The notion that group selection Acan=t work@ always reminds me of what intellectuals said about the second law of thermodynamics a century ago. Literary types from Swinburne to Henry Adams knew all too well what the second law said and reasoned that if heat flows invariably from hot to cold, that the stock of useful energy in the world is always running down, and that disorder (so-called Aentropy@) is always on the rise, therefore the universe is moving toward Aheat death.@
Of course, a closed system like the universe may exhibit rising levels of disorder while, at the same time, local pockets of order are arising. Open systems like the Earth, with enormous throughputs of energy from the Sun, have a tendency toward self-organization, the same as you can see in a pan of cooking porridge if you fail to stir it (the surface becomes furrowed into cells, even hexagonal mosaics on occasion). So too, populations made up of many inbreeding subpopulations can exhibit local pockets of successful altruism.
Second, the long term may never arrive, because the climate changes in the meantime and pumps up the fading percentage with another round of selection favoring groups with more altruism. Beware of the combination of aggregation and of long-term thinking, or you may miss seeing the interesting stuff, all those good stories.
|On to the NEXT CHAPTER
Notes and References for this chapter
Copyright ©2000 by
The handheld traditionally comfortable version of
this book is available from