All in the Family
And the man knew his wife Eve; and she conceived and bore Cain...and again she bore his brother Abel....And in the process of time it came to pass, that Cain brought of the fruit of the ground an offering to the Lord. And Abel, he also brought of the firstling of his flock and of the fat thereof. And the Lord had respect unto Abel and his offering, but unto Cain He had no respect. And Cain was very wroth, and his countenance fell. And the Lord said unto Cain: "Why art thou wroth? and why is thy countenance fallen? If thou doest well, shall it not be lifted?"
According to the Old Testament, the first human siblings were not particularly fond of one another. Fair enough -- we all know of cases of sibling rivalry, not to mention the animosity that we might hold for some of our more distant relatives who pop in for the occasional holiday dinner. But, in general, blood really is thicker than water despite tales of murderous brothers. Later in the story of Cain and Abel we encounter a poignant question: "And the Lord said unto Cain: where is Abel thy brother? And he said 'I know not; am I my brother's keeper?'" The answer, at least in terms of how we behave, is a qualified yes -- you are indeed your brother's keeper.
If your brother and a stranger were drowning, who would you save first? The reply most of us would expect -- "my brother, of course" -- suggests, but certainly does not demonstrate, the importance of the role of kinship in structuring human cooperative acts. Is it possible, however, that the reason we might choose our brother has nothing to do with the fact that he is a blood relative per se? Is it merely the fact that we have spent so much time with our siblings that drives our actions? A simple thought experiment might help us to understand whether this true. Imagine it is not a stranger with your brother there in the water, but your closest friend. How would you feel in that case? How would most people react? If kinship was the overarching theme, most should still reply "my brother, of course." This train of thought has led behavioral ecologists to appreciate just how important kinship is in the human social dynamic.
Family Accounting Schemes
The scenario in which you save your brother or someone else functions as a nice illustration of how kinship affects human cooperation, but it is not something one envisions as a starting place for significant scientific breakthroughs. Or is it? Evolutionary biologists' first introduction to the notion that blood relations affect social behavior actually came in the 1930s in a form similar to this example. It was then that J. B. S. Haldane, a founder of modern evolutionary theory, suggested that he would risk his life to save two (but not one) of his brothers and eight (but not seven) of his cousins. Haldane, quite versed in mathematics, made this rather bold statement by counting copies of a gene that might code for cooperative behavior. Such a gene-counting approach to kinship and the evolution of cooperation has been extended by theoreticians but in its most elementary form is the heart and soul of kinship theory. Let us see how this idea works and how it has been formalized into what is known as kin selection or inclusive fitness theory.
The evolutionary biologist's definition of relatedness and kinship may strike many as surprising, if not odd. In such a definition, relatedness centers on the probability that individuals share genes that they have inherited from some common ancestor (parents, grandparents, etc.). A jargon phrase summing up this approach in behavioral ecology is "identity by descent." For example, you and your sister are kin because you share some (in this case many) of the same genes and these have been inherited from common ancestors, mom and dad. Similarly, you and your cousins are kin, because you share genes in common (not as many as siblings) and common ancestors, your grandparents. Common ancestors are the most recent individuals through which two (or more) individuals can trace genes that they share in common.
Once we know how to find the common ancestry of two or more individuals, we can calculate their relatedness, which simply amounts to the probability that they share genes that are identical by descent -- genes that have been inherited from a common ancestor. In the literature on kinship, this probability is often labeled r (for "relatedness"). For example, you and your brother are related to one another by an r value of 1/2.
From a "gene's-eye" perspective, calculating relatedness is the first critical step in understanding how kinship can favor cooperative behavior among individuals. Genes' survival depends on the number of copies of themselves that they get into the next generation. This is often thought of in terms of what effect a given gene has on the individual in which it resides, but relatedness suggests that this is a myopic view. If relatives have a high probability of sharing a given gene, then that gene can potentially increase its chances of getting more copies of itself into the next generation by coding for some behavior that helps relatives. Again taking a gene's-eye view, relatives are just vehicles who are likely to have copies of you (the gene in question) inside them as well. But, and this is a big "but," relatives only have some probability (r) of having a copy of, for example, a gene for cooperation. A gene in sibling 1 "knows" that a copy of itself may reside in sibling 2, but only with a 50 percent probability. The more distant the relative, the less likely a copy of the gene resides in them as well. So, phrased in the cold language of natural selection, relatives are worth helping in direct proportion to their relatedness. This is because relatedness is a measure of genetic similarity, and genes are the currency of natural selection.
Behavioral ecologists are not so foolish as to assume that animals are able to calculate relatedness in the manner described above. We only assume that natural selection favors individuals who act in ways that make it appear as though they are able to make such calculations. How animals determine who is kin and who isn't is a matter of some debate these days. For example, one theory suggests that animals determine relatedness by matching a suite of traits (a template) that they possess against the same suite of traits in another individual. Depending on the degree to which traits match up, individuals are treated as full siblings (if many matches occur), half siblings (if fewer matches occur), cousins, and so on, down to the category "unrelated individual" (if, for example, no matches occur). Such "matching games" have their flaws; mistakes can be made in determining the level of overlap, and some relatives may erroneously be treated as nonrelatives, while some nonrelatives may be viewed as relatives, Often, however, rather than a suite of traits, a single characteristic is used to determine whether another individual is kin and if so, what type of kin. In many insect species, for example, kinship is assessed by odor. Individuals who smell like you (or your nest) are relatives, and how closely related they are is determined by how similar their odors are to yours.
While most behavioral ecologists accept that such matching (either of many cues or a single cue) is important, they believe that there is another, simpler explanation for how animals determine who qualifies as kin, an explanation that I'll refer to as the "no place like home" hypothesis. Under this hypothesis, animals simply treat all others that grew up in their nest (territory, burrow, etc.) as relatives. This very simple rule is often quite powerful. With the exception of some species that try to trick other species into raising their offspring, the odds are quite strong that those who grew up in your nest are in fact your siblings and parents.
The details of how animals evaluate relatedness are fascinating, but all we really need to know to examine kin-selected cooperation is that many animals do in fact behave in ways that allow them to distinguish between kin and nonkin and even to distinguish between different degrees of relatedness. Once we have calculated relatedness, we are very close to reaching a general rule for when cooperation among relatives should be favored and when it should not. We need only consider two more factors: the cost of the action to the individual cooperating and the benefit to the recipient of such a cooperative act. Let us call the cost of a cooperative act to the donor c, and the benefit to the recipient b. In 1964, W. D. Hamilton (now at Oxford University) showed that cooperation among relatives should evolve when the following holds true: r X b is greater than or equal to c.
In other words, cooperation among relatives is favored if, and only if, the benefit of the act multiplied by the relatedness of the actors is greater than or equal to the costs. This equation, r X b is greater than or equal to c, has become known as Hamilton's Rule. Essentially, Hamilton's Rule says the following: There is some cost (c) that "must be made up for" if the gene for cooperation is to evolve, as cooperating with others is often a risky business. One way to make up for this cost is through the benefits (b) a relative receives, because relatives may carry the gene for cooperation as well. But, relatives have only some probability of carrying the cooperation gene and so the benefits received must be devalued by that probability. If I pay a cost for undertaking an action, but there is only a probability that I will receive indirect benefits (in this case through my relatives), I need to factor that into my equation and that is just what r does.
We can illustrate the use of relatedness to predict cooperation among kin with a simple chart. Consider an action that you take that reduces your chances of survival by 50 percent (a very serious cost) but increases the probability of survival of the relative(s) you are trying to save by 50 percent each (a considerable benefit). Such extreme costs and benefits might, for example, mimic a situation in which you scream out when a gang of armed thugs is approaching. This serves to announce the presence of thugs to the relatives around you, but at the same time the scream draws the marauders' attention your way, a dangerous action indeed. Based solely on kin selection theory, the table shown here outlines the number of relatives that need to hear your scream before natural selection alone would favor such dangerous behavior on your part.
The table illustrates the fundemental point of inclusive fitness theory: the greater the degree of relatedness between individuals, the more likely that kin-selected cooperation is selected. There need only be two (or more) siblings around for you to make that scream, but you'd need eight or more cousins present (a much less likely event), if they were the only relatives in the vicinity! How exactly, though, do we use Hamilton's Rule to come up with the correct number of relatives in the table? Consider the case for siblings. If a single sibling hears an alarm call, then r = 1/2 and b and c are still each 1/2. In that case r multiplied by b is not greater than c, Hamilton's Rule is not met, and cooperation via kinship is not favored by natural selection.
Suppose, however, that three siblings hear the alarm call. Now b is tripled (three recipients), but c is the same (the alarm call still draws the predator's attention), so r X b = 1/2 X (1/2 X 3) for a total of 3/4, which is greater than c, and Hamilton's Rule is satisfied. The same logic can be applied to any relative in the table (or for that matter, any relative not in this table). Take note, as well, that the relatedness of an individual to his/her spouse is 0 (with the exception of marriages among relatives). Although one's spouse is kin in the everyday usage of the term, we don't generally share genes inherited from a common ancestor with our spouses and hence this category of relative is in effect removed from kin selection theory.
Of the four paths to cooperation that we will focus on, kinship is the best understood, most accepted, and least controversial. It is in every legitimate textbook on evolution and is cited in more papers in the field than any other set of theories. There is even a belief among some evolutionary and behavioral biologists that Hamilton's work in this area marks the start of the modern discipline of behavioral ecology. But even kin selection theory is not without its controversies.
One area of contention with respect to kinship and cooperation centers on whether it really matters where the genes we are counting are located. Kin selectionists correctly argue that blood relatives are more likely to carry the same gene than are individuals drawn at random from a population. But what if some other mechanism besides kinship could create groups in which individuals were all likely to carry one or more genes coding for cooperation? Does it really matter that such individuals don't share other genes, like kin do? After all, we are interested in the gene(s) coding for cooperation, and everything else is in some respects background material for that gene. Who cares whether individuals carry the same genes because of kinship or for some other reason -- shouldn't the process by which cooperation is selected for work just as well in both cases? The answer to this question, as Hamilton himself noted, is yes, the process works the same; whether individuals share the gene(s) for cooperation because of relatedness or some other factors is irrelevant.
Yet kin selection advocates are not so fast to roll over. Sure, mathematically speaking, you are right, they say, but in practice the distinction we are arguing about is still real and important. Give us, they say, a good example of how individuals sharing a gene are brought together, if relatedness (which automatically brings them together) is not in force. The answer typically given by kin selection critics is that individuals that share a gene for cooperation may gather together specifically to be near other cooperators, because cooperators do particularly well when around others like themselves and so should choose this option, when it becomes available. "Be specific," say kin selectionists, "give us a real example." And this is where the kin selectionists start looking a bit better than they did after losing the mathematics argument, because behavioral ecologists are usually stopped in their tracks when it comes to finding a good animal example to answer this question.
Although such examples may be hard to uncover in animals, those interested in enhancing human cooperation argue that the evidence for cooperators choosing other cooperators as partners in our own species is anything but scarce -- even when kinship is not in play. How others will act is one primary means by which we choose with whom we will interact. So, for humans then, while kinship is an extremely important force selecting for cooperation, there are many other ways cooperators may cluster together aside from kinship. We need to recognize this in our behavioral studies and our conjectures about human cooperation.
The above controversy is admittedly a semantic one in part, but semantic arguments can be quite illuminating. Hamilton's Rule -- which in words roughly translates to "all else equal, cooperation should be most common among close relatives" -- is as close as behavioral ecologists get to a "law of nature." It is an underpinning of all modern evolutionary approaches to social behavior and is, in many ways, as much an approach to behavioral biology as it is a theory. The data gathered to date certainly support the claim that Hamilton's Rule is extremely powerful. It is not a "law" in the sense that gravity is, but it is about as near to one as behavioral biologists can hope to come, given the astonishing complexity and variability that is an inherent part of the subject matter they tackle.
From the standpoint of reputation, Hamilton's Rule was quite good for the field of behavioral ecology, at least in one sense. While solid mathematical theory has been part of evolution since the seminal work of J. B. S. Haldane, Ronald Fisher, and Sewall Wright in the 1930s, it was not truly a centerpiece of evolutionary approaches to behavior until Hamilton's Rule. For many in the field of behavior, there was an unspoken envy of the hard sciences (physics, chemistry, even other parts of biology) that had steadfast "rules" that could be written out for skeptics (not to mention funding agencies). Hamilton's Rule provided such ammunition to behavioral ecologists. Let's take a look at some examples of why this is so, with a few cases from the animal kingdom, before moving on to how such scenarios can help us foster human sociality.
The Insect Police
The so-called social insects have been a godsend for advocates of kin-selected cooperation. The reason lies, at least in part, with the bizarre genetics of social insects such as bees, wasps, and ants (collectively known as hymenopteran insects). Humans (and most other animals) are diploid organisms, which means that we have two copies of each of our chromosomes. Our forty-six chromosomes are twenty-three matched pairs. The only stages of human life that are not diploid are sperm and egg, as they have only a single copy of each of our twenty-three distinctive chromosomes. Sperm and egg then are called haploid rather than diploid. Of course, sperm and egg later fuse to form diploid animals.
Much of life on earth, such as bacteria and viruses, is always in the haploid phase. Why some life on earth is diploid and some haploid is a fascinating question, but not one critical to the issues we are examining. What makes bees, wasps, and ants so bizarre is that females are diploid and males are haploid -- a genetic system known as haplodiploidy. What this means is that when a male fertilizes a female, only daughters are produced because the sperm and egg fuse to produce a diploid creature, and in most social insect species diploids are female. Females, however, produce sons from unfertilized eggs (eggs that have not fused with sperm) -- which means that sons never have fathers!
Haplodiploidy creates some very strange scenarios. In diploid and haploid creatures, relatedness between two individuals is symmetric; that is, if a father is related to his daughter by an r of 1/2, then a daughter is related to her father by the same value. Not true for the social insects. To see why, focus your attention on the father/daughter relationship. Fathers are haploid and give a copy of each chromosome they have to their daughters. Hence fathers are related to daughters by a value of 1. Daughters, however, are diploid, in that they get one copy of each chromosome from each parent, both mom and dad; so a daughter's relatedness to her father is 1/2 (half her chromosomes come from dad) -- fully half of her father's relatedness to her.
The most relevant effect of the strange genetics of the social insects is its impact on average relatedness within insect colonies. Before seeing this in detail, keep in mind that in many social insect colonies a single queen produces all the offspring for a group. This means that the vast majority of individuals in such colonies are sisters and brothers. Haplodiploidy has the twofold effect of making sisters "super-relatives" and making the relatedness between brothers and sisters only half of what it is in diploid brothers and sisters. Sisters end up with a relatedness value of 3/4 (50 percent greater than the same relationship in diploid species), and sisters are related to brothers by a value of 1/4 (half the value found in diploid creatures). So, a clear prediction from kin selection theory is that since females are much more related to their fellow colony members than are males, when colonies have more females than males (as in most social insects), cooperation should occur predominantly in this sex. And of course it does, as "workers" in insect colonies are almost always female! It is females that sacrifice their lives by stinging folks and ruining an otherwise pleasant summer day. It is also females that undertake virtually all of the everyday activities that keep a colony functioning -- food gathering, care for the young, and so on. One particularly interesting and unique behavior found among female social insects is "policing" behavior.
Bee colonies are rightly thought of as models of both efficiency and harmony. It is mind-boggling what a colony of tiny insects can accomplish in a short period of time: regulating the temperature of a hive, caring for young, defending against many predators, finding food, recruiting others to join in bringing back the booty, and a myriad of other activities. Some of this efficiency (and harmony) has been attributed to a single queen often producing all of the eggs for a colony, thus allowing worker females to spend their time on other hive-related necessities. Queens accomplish this enviable task by using a barrage of chemicals to inhibit other females -- the workers -- from reproducing. Yet, as with any chemical inhibition system, it is inevitable that some workers will escape these anti-aphrodisiacs and thus will have a much greater chance of reproducing than their subdued sisters. Once this fascinating new door is opened, we can ask whether kinship theory can guide us with respect to a rather nasty question: should the eggs laid by workers that ignore the queen's chemical castration cues be left alone by their sisters or vigorously attacked? The answer to this question is rather personal, if you happen to be the queen, as it depends on how many males you opt to mate with.
The relatedness between individuals in an insect colony depends on how many males inseminate the queen. The more males the female mates with, the more different lineages there are in a colony -- each line's ancestry going through the queen and a given male. Once again, however, the strange genetics of such insects creates a novel situation. Rather than showing a family tree more complicated than that of the British monarchy, it can be shown that if the queen of a colony mates with a single male, then female workers in the colony turn out to be more related to nephews than to brothers. If the queen mates with numerous males, however, that situation reverses itself and female workers in a hive are now more related to brothers than to nephews. We shall focus on the second scenario because of the fascinating kin-selected cooperation emerging from it.
Whether females in a social insect colony are more related to brothers than to nephews can have quite serious implications about when and whether we should see kin-selected cooperation, and if it exists, what form such cooperation should take. To see this, first recall that brothers are those individuals produced by the queen, while nephews are those produced by sisters that have somehow managed to avoid the queen's chemical anti-reproduction agent. A conflict of interest then arises between sisters that have managed to escape and those that have not.
Aside from the queen, females who can reproduce (i.e., those that do not fall victim to the queen's attempt to monopolize reproduction) are always selected to do so. When females reproduce they always produce males, since such females are almost never inseminated. This creates a problem, however, for those females who can't reproduce, as they are more closely related to the queen's offspring (their brothers) than to their sisters' children (their nephews). Kin-selected cooperation on the part of those nonreproducing female workers then favors any action that increases the odds of the queen's offspring surviving at the cost of nephews.
There is little a female can do to stop one of her sisters from reproducing, if her sister has avoided the queen's attempt to do so already. But there are options available. Once a worker has laid eggs behind the queen's back, her sisters could, for example, refuse to care for and help nephews. Or they could take more drastic action -- they could eat eggs destined to be their nephews! Francis Ratnieks and Paul Visscher examined this possibility in honeybees, where females mate with ten to twenty different males. Their results were astonishing. Those honeybee females who did not produce offspring "policed" the reproductive actions of their sisters. If their sisters produced eggs on the sly, policing females destroyed the eggs. Ratnieks and Visscher found that honeybee workers showed remarkable acumen in discriminating between sisters' eggs and the queen's eggs. In a controlled laboratory setting, after twenty-four hours, only 2 percent of the sister-laid eggs remained intact, while 61 percent of the queen-laid eggs remained unharmed! But, given that the actual act of egg laying is rarely observed, how could honeybees know which eggs were laid by sisters and which by the queen? The answer appears to be that eggs are chemically "marked," such that queen-laid eggs smell different from worker-laid eggs. Why eggs should be marked so is still unclear, but one tantalizing possibility is that the queen marks her eggs to encourage workers to police the activities of their sisters.
Kin-selected policing is qualitatively different from the other types of cooperation so often found in animals. Rather than having individuals form a cooperative unit to accomplish some task, cooperation in honeybee police work takes the form of stopping others from cheating -- a more subtle and complex action. At a more fundamental level, policing is powerful, because it provides a direct deterrent to cheating, whereas in many other cases, we simply rely on cooperation being somehow more profitable than cheating, and this holy grail is often difficult to obtain.
There are many other cases of cooperation in highly related social insects. I'll mention one other curious example: honeypot ants. In one species of these ants, the largest individuals actually hang from the top of a colony and act as living storage tanks for water and sugar. These "honeypot" individuals have soft and elastic abdomens, and if you watch long enough you will see other individuals come up and "turn on the faucet" to drink the resources stored there. For significant periods of time, honeypot individuals do nothing but hang from the rafters and supply this service.
It is fascinating to find policewomen and living storage bins in the insect world, but how much of the cooperation we see is strictly due to the bizarre haplodiploid genetics of social insects? Can we expect anything so dramatic among mammals?
"Eureka!" Naked Mole-Rats
Physics is not the only discipline in science that has "Eureka!" stories. Just as physicists can recount the bathtub adventures of Archimedes and his famous exclamation when coming up with his theory of buoyancy (specific gravity), so too can the ardent student of behavioral ecology recite the story of Richard Alexander and Jenny Jarvis's discovery of extraordinary cooperation in a bizarre creature, the naked mole-rat. Alexander, a professor of biology at the University of Michigan, traveled to various universities in the 1970s, giving lectures on the evolution of social behavior, particularly cooperative and altruistic behavior. One of his themes was why, despite significant effort, extreme sociality (like that seen in insects) had not been uncovered in mammals. Alexander described the characteristics he believed a mammalian system would need for insect-like ultrasociality to exist. He outlined a hypothetical creature that would undertake altruistic acts for relatives who lived in a safe environment with lots of food. He went so far as to give details: the species would eat large tubers (potato-like foods) and live in burrows in a tropical spot that had clay soil.
One day in May 1976, Alexander presented these ideas to some folks at Northern Arizona University. Afterwards, he was approached by someone in the audience who told him he had given a perfect description of the naked mole-rat of Africa. On the advice of this fellow, Alexander contacted Jennifer Jarvis (at the University of CapeTown), who knew more about naked mole-rats than anyone in the world. After much back and forth, which included trips by Alexander and his colleague Paul Sherman to Africa to actually see the creatures, Jarvis and Alexander realized that they indeed had found the first eusocial (ultrasocial) mammal.
After all the attention this animal has attracted from both scientists and the media, it is almost disappointing to see how bland the native habitat of the naked mole-rat actually is and how ugly these creatures are, even by rodent standards! Naked mole-rats are hairless and blind, with crinkled skin and two large incisor teeth sticking out from their mouths. And those are the adults; the babies are even harder to look at for very long. First collected in Ethiopia in 1842, naked mole-rats (whose scientific name is Heterocephalus glaber) live within groups averaging about seventy individuals (but ranging up to almost three hundred) in underground burrows, from which they rarely, if ever, emerge. Such burrows average about two miles in length. Naked mole-rats have been studied primarily in Kenya and are often found in arid areas covered with dust and brush. Typically found near dirt roads, colonies can be located by molehills that pock the landscape. But what naked mole-rats lack in beauty and scenic living conditions, they make up for in fantastic behaviors.
One female alone (among many in the colony) is responsible for all the reproduction in a naked mole-rat group (three or so males in the group are responsible for the male side of mating). No other mammal that we know of, except another species of naked mole-rats discovered later on, has a single "queen," and this finding sent shock waves through the behavioral biology community. Kin selection theory suggests that such extreme cooperation, wherein most individuals give up the opportunity to reproduce, should be limited to species in which individuals are somehow extremely related to each other, yet naked mole-rats are mammals and don't have the bizarre genetics that allow for the "super-relatives" we saw in the bees and ants. So how could such a bizarre system have evolved here? Before answering this question, let's get a more comprehensive sense of just how much cooperation goes on among these creatures.
The queen and the handful of males she mates with have a twofold advantage over others in naked mole-rat colonies: not only do they monopolize all colony reproduction, but they also live longer than their nonreproducing colony-mates. Yet in the relatively short time that nonreproductive males and females are around, they get a lot accomplished, and without their cooperation naked mole-rat colonies would surely come to a screeching halt. In fact, those individuals not specialized in reproducing take on virtually all of the everyday cooperative actions that are the very lifeblood of colony existence. They excavate new tunnels (an absolutely critical aspect of colony survival), sweep debris, groom one another as well as the queen, and take on the unenviable and dangerous task of defense against predators.
Why such dramatic examples of cooperation in a single species? What s
The Nature of Cooperation in Animals and Humans
Cheating Monkeys and Citizen Bees
The Nature of Cooperation in Animals and Humans
In Cheating Monkeys and Citizen Bees, evolution and animal behavior expert Professor Lee Dugatkin, known throughout the academic community for his ingenious animal behavior experiments, reports, from the cutting edge of scientific research on the startling evolutionary truth about cooperation and how it works. He explains the four paths to cooperation that we share with animals and provides the experimentally verified definitions of a behavior no one thought science could ever explain. The first path is through our families and demonstrates that blood really is thicker than water; the second shows why it makes biological sense to do unto others as they do unto you; the third reveals the dynamics of a kind of selfish teamwork; and the last and grandest path is to complete altruism Dugatkin illustrates his argument with marvelous behaviour in the natural world: baby-sitting mongooses and squirrels that willingly martyr themselves to save relatives; fish that switch sexes in order to share reproductive duties; and vampire bats that regurgitate blood for their hungry mates. With these colorful insights into the natural world, Dugatkin shows that what comes naturally to animals can teach us about the instincts that underlie the complex web of human social networks. We can use our understanding of these instincts to encourage purposeful human cooperation, even in situations where animals would not naturally band together.
Those readers with an interest in ecology, evolutionary biology, psychology, even anthropology will find Cheating Monkeys and Citizen Bees an essential handbook of the dynamics of cooperation. And everyone will find it to be a lucid introduction to the surprising evolutionary history of how we came to behave in the ways that we do, of how nature came to be less brutal than we tend to think.
- Free Press |
- 224 pages |
- ISBN 9781439136515 |
- February 1999