Oliver Goldsmith (1776)

In the last chapter we saw that during the previous twenty-five years our understanding of biological evolution has been deepened by several new or newly clarified concepts. Specifically, the following four stand out in importance for the evolution of behavior. (First) is the realization that in the vast majority of cases natural selection takes place at the level of individuals and not groups.' Thus, arguments that such-and-such structure, process, or behavior occurs "for the good of the species" are generally incorrect. This is a point that was clearly understood by Charles Darwin but for many biologists was only brought into sharp focus by George Williams's Adaptation and Natural Selection, published in 1966.

Second is the concept of inclusive fitness, which recognizes that copies of many of an animal's genes reside in close relatives and that in certain circumstances fitness can be enhanced by seemingly altruistic acts extended to others.2 Third the theory of parental investment, which has far-reaching implications for mating behavior and the relationship between parents and young.3 And fourth is the concept of evolutionarily stable strategies, which arises from an application of the theory of games to the evolution of behavior.4 We are now going to examine in more detail some arguments that are based on these ideas. We shall use for our prime example a matter that has very broad and general importance in the animal kingdom, that is central to reproduction and fitness, and yet whose application to the human condition is frequently misunderstood or

First, however, let's consider a fundamental question that we alluded to earlier: Why does sex exist at all? When we speak of sex in ordinary speech we usually mean male or female gender or the physical act of mating. Sex also refers to something more basic: the mixing of parental genes to produce a new genotype.

The very existence of sex presents a biological conundrum. What makes it a conundrum, however, starts from the conclusion, independently reached, that individual organisms and not larger groups or species are the principal focus of natural selection. The paradox is as follows: If the fitness of the parent has been well tuned by the history of selection of the parent's ancestors, the shuffling of genes (by independent assortment and recombination) that takes place in the formation of eggs and sperm together with the new diploid combinations created when the egg is fertilized should break up successful assemblages of genes and generate genotypes with substantially reduced fitness. Producing poorly fit genotypes is not only a waste of reproductive effort, it is also a loss that would not be incurred if the parent were to reproduce asexually. Furthermore, in species with negligible parental investment by the male, females would maximize their reproductive output if they relied upon asexual reproduction and produced only daughters.

What, therefore, is to be gained by sex? Sexual reproduction might provide a way of probing for even better combinations of genes. Which mode of reproduction is then best? The answer seems to depend on the circumstances, but the problem is not thoroughly understood. Ordinarily, if natural selection has been effective in a lineage, better combinations of genes for the existing environment are relatively improbable. But if a deleterious mutation should arise in the parent, sexual reproduction provides an array of genotypes among the offspring that natural selection can then edit. In contrast, when a mutation arises in an asexually reproducing line, unless it is removed by another mutation it will be found in all offspring and in their descendants. If individuals bearing the mutant gene encounter strong negative selection, the gene can drag into oblivion all of those individuals, along with the rest of their other genes.

With asexual reproduction, selection will occur between clones, and under certain conditions the rate of evolution will be substantially slower than is possible in the presence of recombination. Sexual reproduction can lead to faster evolutionary change if the breeding population is large enough to bring together favorable new combinations of genes at rates that exceed the successive accumulation of favorable mutations in asexual lineages.

Sexual reproduction will be favored where there is strong competition between individuals. This condition may be met if the organisms arc at high density and have limited capacity to disperse. Similarly, sexual reproduction may increase the number of offspring able to occupy variant niches in the habitat, also a more likely necessity at high population densities.

A further potential advantage to producing a variety of genotypes arises when the environmental future is variable and uncertain. The organism does not, in fact cannot, plan for future generations, and like a poker player whose luck is not limited to the quality of the first five cards he is dealt, sexual reproduction gives the organism at least a chance of putting forward an assemblage of genes better suited to the selective competition that will be encountered. But for sexual reproduction to prevail, this benefit must exceed the costs of sexual reproduction that were enumerated above. The "Red Queen" hypothesis (named for the character in Lewis Caffoll's Through the Looking-Glass who explains to Alice that in her world " . . . it takes all the running you can do, to keep in the same place") is a variant of this idea whereby it is suggested that the ever-changing nature of environments puts a premium on the genetic diversity of offspring just to keep the lineage going.5

Fluctuating environments will assert a maximum premium on sexual reproduction when environmental changes actually oppose the current adaptational state and do so relatively frequently. This condition is most readily imposed not by the physical environment, where fluctuations are more likely to be random, but by the biotic environments Predator and prey, host and parasite are in states of adaptive warfare in which an improvement in one member issues a counterchallenge to the other. The occurrence of asexually reproducing organisms tends to support the paramount role of other organisms in generating the environmental instability responsible for sexual selection. Sexual reproduction is more common in parasites than in closely related but free-living forms. Asexual reproduction is more common in fluctuating, disturbed, and ephemeral habitats, at high latitudes relative to the tropics, and in fresh water rather than marine environments. These general correlations at first seems surprising, but they suggest that the physical environment is not in fact the primary causal agent in determining whether reproduction is sexual or asexual. Asexual reproduction correlates better with conditions in which there is a premium on a high rate of reproduction and where there are minimal interactions between individuals or between species.

The fact that some organisms reproduce sexually, others asexually, while still others switch modes indicates that there are advantages and disadvantages associated with both systems. The existence of sexually and asexually reproducing organisms is but one of numerous examples we shall encounter of natural selection's traveling multiple paths. The origin of sex is also a realm of evolutionary theory in which there is considerable intellectual uncertainty and ferment.7

In parts of what now follows I have quite consciously paraphrased Richard Dawkins,g an Oxford biologist whose vivid and inspired writing on evolution has successfully reached a general audience. I have done this, however, not in criticism -of his lucid style, but to illustrate a point about the use of language in discussions of ultimate cause. Many readers may find portions of this description troublesome, because a number of the common English words and phrases will seem to be invested with too much meaning. I have flagged some of the more likely examples with quotation marks, and I shall return later in the argument to a further discussion of the language.

Sex can exist in bacteria and fungi in the form of different mating types, yet without the morphological differences we recognize as male and female. How do we decide which sex is female

and which is male? We recognize that in general the two sexes Vhave adopted different "strategies" for the manufacture and dissemination of their germ cells. One sex makes relatively fewer cells (eggs) and invests more of itself in each. This is what Dawkins has termed the "honest" strategy, and the individuals who practice it we call females. The other sex makes more but smaller and more mobile germ cells (sperm) whose task it is to find the eggs. Unlike the egg, each sperm has little to contribute to the zygote but its genetic material, its DNA. This is the "sneaky" strategy of Dawkins, and its practitioners we call males. To quote Dawkins again: "Female exploitation begins here." Let us explore what this means.

For many species, parental investment only starts with the formation of germ cells, and there is a great deal of species difference in the disparity of investment made by the two sexes. In fish (where fertilization is external), both sexes cast their seed into the water and the differential investment in eggs and sperm is probably not large. In some species the young receive no parental care at all; the eggs are deposited in the water, fertilized by an attending male, and the fry fend for themselves on hatching. In other cases, occurring in ponds or coral reefs (where the habitat is structured and demarcated), males defend territories and may frequently protect their young, keeping them together, sometimes by sequestering them in the males' mouths.9

Among birds and mammals the situation is different. Because fertilization is internal, the female is stuck with the zygote in a way the male is not. Female birds put a lot into their eggs, not just figuratively but in fact. Although the eggs of female mammals are much smaller than those of birds, the embryo is nurtured internally and the young are fed after birth by milk secreted by the mother. All of these features place an energetic burden on the female that the male shares only indirectly, if at all. When the female's parental investment is much more extensive than the male's, the female is potentially capable of contributing genes to fewer offspring than the male is. Under these conditions, females are, for males, a resource," and selection tends to favor patterns of reproduction in which males attempt to inseminate many females. For females, on the other hand, there is usually not an equivalent advantage to be gained by multiple matings. Moreover, because for the males the females are a valuable resource, they frequently become the object of competition between males. This competition expresses itself in various ways, often in aggressive interactions between males, by trends toward sexual dimorphism-with the males becoming larger and more aggressive than females, and a tendency for males to try to coerce females into mating and to prevent them from mating with other males. As mentioned previously, the ultimate causes of sexual dimorphism were recognized by Darwin as sexual selection. Sexual selection is a result of differential parental investment.

Successful genes are those that are propagated into subsequent generations. Because the genes of males and females are packaged in different bodies, they must use somewhat different strategies to achieve this end, and there is therefore some conflict inherent in the interests of the two sexes. Each parent makes essentially the same contribution of genetic material to the zygote, and there would be some advantage to each if the other could finish the job of rearing the young, freeing the first to start another round of reproduction. But as the total initial investment of the female (particularly a bird or mammal) is greater, we would expect to find males much more likely than females to desert once fertilization has occurred, leaving the female to complete the task. On the other hand, desertion is an evolutionarily acceptable strategy for a male only if there is a reasonable likelihood that the female will be successful in rearing the young without his help. Moreover, the female's genetic interests will be served by processes that decrease the chances of desertion and increase the parental investment of the male.

If the male's reproductive interests are fostered by multiple inseminations, a tactic for the female is to defer her choice of partner until after a courtship period in which males compete in demonstrating their capacity for parenthood. In short, the females become "coy." For example, in many species of birds the males stake out territory sufficient to support the young, start building nests, and announce their presence with songs and bright feathers before mating. This process also serves the female by increasing the male's parental investment, a tactic that is effective if it involves him until it is too late in the breeding season to seek another mate. Moreover, it allows the female to choose a mate "that is in the best interests of her genes."

Because the reproductive interests of males and females are never identical, there are inevitable conflicts. Even when the parental investment of the male is large, for example in guarding territory or in feeding young, there is still some potential gain to be had by 'philandering," even short of deserting. It is in the interests of the female, however, to minimize the time and energy her mate expends on creating offspring who do not share her genes, and as we have described, the female can look to her interests by seeing that the male spends time and energy wooing and by being selective in her choice of mate. Seen from the perspective of the male's genes, though, if he is to make a substantial parental investment, it pays him to maximize the odds that he is the father of any offspring. Whenever fertilization is internal, as in birds and mammals, the male can never be certain of paternity unless he pays great attention to his mate's activities. Consequently, it becomes particularly important for males with a large parental investment to guard their mates to prevent cuckoldry. Cuckoldry can be in the interests of the female, however, if a "better" male should be transiently available.

There is another reason why females might be expected to be more 'fussy" than males in selecting mates. As they have more invested, they have more to lose should the offspring be genetically inferior. Assuming incestuous relationships are initiated by the older partner, mother-son incest should therefore be rarer than father-daughter incest,'O a prediction that seems to be fulfilled in human populations.

Another pattern of courtship involves displays by males, such as the bowers of bower birds or the struts and accessory plumages of grouse and peacocks. These male behaviors, which appear to our eye irrelevant to the practical business of rearing young, seem to convey to the females some general information about the vigor and general genetic worth of potential mates. It is in the interests of the femalc's genes if she can produce sons who themselves are successful in siring offspring. "Sex appeal" therefore becomes a desirable end in itself.

As females are generally the reproductive resource, competition for mates should be more frequent between males than between females. Competition between females is indeed less obvious, and it tends to be of a different sort. In a mating system where a few males are inseminating most of the females and there is little further paternal investment, not only is there no reason for females to compete for mates but it is actually in the reproductive interests of every female to mate with the most fit males. Under these conditions, therefore, polygyny (where one male mates with several females; see below) can actually work to the genetic advantage of the females. Food and a safe habitat present a different kind of challenge to the females, however, and active competition for and defense of these resources by females is common in many species. In some primates there is a dominance hierarchy among females, and the offspring of dominant mothers enjoy the results of the mother's social status, for example in access to food.11,12

In those animal societies (e.g., langurs, lions) where exclusive breeding opportunities are a temporary privilege of any male, usurpation may be followed by the new M were sired by his predecessors 'Mis behavior has struck some observers as an aberration on the mistaken premise that it is maladaptive, but this interpretation assumes that adaptive behavior must be for the good of the species. In these instances nothing could be more in the interests of the new male and his genes than to get the females ready for him to inseminate as quickly as possible.

Dawkins has presented a nonmathematical description of the consequences of various reproductive behaviors for an evolutionarily stable strategy.14 In brief, if all females were "coy" and required a demonstration that males were "faithful" before mating with them, the system would be unstable because it would be open to, females who cheated by being "fast" and not enforcing a period of courtship. The presence of "fast" females would open the way for " philandering" males, who would otherwise be excluded. The presence of 'philandering" males would in turn drive up the frequency of at the expense of females. What results from this ki@d of analysis is a mix of behaviors in the population (and over time) that is stable in the sense that it is immune to perturbation by changes in the frequency of any of the behavioral forms. What the actual proportion of "coy," "fast," 'faithful," and philanderer" are depends on the assumptions one makes about the payoffs and penalties to reproductive success that are associated with each behavior. It is important to realize, however, that this kind of modeling does not assume that each behavior is rigidly fixed; an individual female could be "coy" part of the time and 'fast" the remainder, and the conditions of the model would be fulfilled. We shall see subsequently that behavioral plasticity of an equivalent degree is readily observed in nature.

The preceding several pages convey some of the flavor of evoluy arguments about the ultimate causes of behavior. The r who is unaccustomed to thinking along these lines may well be disconcerted by some of the language I have used: 'successful" genes, evolutionary "strategies," an animal's "confidence" in paternity, "philandering" male animals, and creatures who "choose mates that are in the best interest of their genes." These kinds of descriptions have been criticized because they suggest moral judgments by animals and conscious foresight by molecules. The criticism is understandable, because language is supposed to be able to convey meaning unambiguously. Biologists, for their part, bring their own background and understanding to the problem, and frequently find such interpretations so silly as not to require comment. The result is often a serious failure of communication.

We must establish a common understanding of this matter before we proceed further. Nothing in these arguments about the adaptive significance of various patterns of reproduction is intended to address the proximate mechanisms that determine mating behavior. For example, calling behavior "coy" is simply a shorthand way of saying that the female does not copulate until time has passed and more information is available to her. It implies nothing about the cognitive processes by which that information is assimilated, her emotional state, or any other property of her nervous system. Such things are important, but they are not directly relevant to statements about the overall effects of "coy" as opposed to 'fast" behavior on the transmission of the female's genes to the next generation. The argument obviously assumes that there are some mechanisms by which "coy" or "fast" behavior can be expressed, but the details of those proximate causes are generally not of central concern when one is thinking in terms of ultimate cause and evolution. Similarly, to say that the male has "confidence in paternity" simply means that the animal is involved in a mating system in which the probability is low that his mate will be inseminated by another. Whether he frets about the problem is another matter. Clearly males of at least one species do.

The alternatives to a breezy style of description of animal behavior can be cumbersome or involve the creation of a special jargon. To date, evolutionary biologists have had relatively little trouble communicating with one another without a great deal of the latter, but they sometimes have a serious problem in reaching beyond their circle. Bridging this communications gap will require more sophistication in distinguishing between proximate and ultimate cause than our educational system now provides most students-or is likely to provide in the near future. Greater efforts on the part of biologists will be needed to convey the power, importance, and beauty of evolution.

Finally, although evolutionary biologists are often not particularly concerned with proximate cause, social scientists and behavioral physiologists clearly are. Much of the remainder of this book has to do with the manner in which explanations of proximate and ultimate cause can illuminate each other. But for the moment this is getting too far ahead in the story.

The concept of differential parental investment holds the key to understanding various mating systems: monogamous, polygynous, and more rarely, polyandrous. What has been described to this point as the basis for sexual selection is a polygynous system, in which one male mates with a number of females. Characteristically, in polygynous systems of birds and mammals, most of the females of reproductive age become inseminated and attempt to reproduce. Their reproductive success is determined more by the resources available to them-a suitable breeding site relatively safe from predators and with access to sufficient food to feed the young-than it is by the availability of males. The reproductive success of males, on the other hand, is more likely to be limited by competitive encounters with other males or with other dangers, for example, predators, to which the male is exposed during conspicuous and persistent courtship behavior. As a result, the variability (again, the variance) of the reproductive success of males tends to be greater than that of females. As with elephant seals, one or a few males may achieve most of the matings in a strongly polygynous system in which there has been vigorous sexual selection, and some males may be cut out of the picture entirely.

The Canada goose makes long-term pair-bonds. In summer the geese travel to the far north to breed, returning south in the winter to find open water and a supply of food. Their honking vee formation is a nostalgic reminder of the changing seasons, and at least in our culture, belief in their monogamous mating system seems to stir human emotions with a similar power. In fact, most birds are monogamous, at least in the limited sense of forming pairs within a single breeding season, whereas most mammals are to some extent polygynous. Only about four percent or fewer of mammalian species approach monogamy. What circumstances tilt the mating system toward monogamy? Simply put, the less disparity there is in parental investment, the more likely the mating system will tend toward monogamy. It has been argued that the difference between the monogamy of most birds and the polygyny of most mammals follows from the fact that female birds deposit their developing embryos outside of their bodies in large, shell-covered, yolk-containing eggs, whereas female mammals gestate the embryos internally, nourish them before birth through a special structure called the placenta, and following birth feed them milk, also produced by the mother. 15 The mammalian form of matemal investment is enormous, and for many species the paternal role is correspondingly reduced. Following insemination of the female, to the extent that young will develop successfully to weaning without any further paternal input, the reproductive interests of the male are best fulfilled by seeking other females to inseminate. For birds, on the other hand, particularly those whose young hatch from the egg in a relatively naked and helpless state, sustained parental investment by the male may be an absolute necessity in order just to feed the gaping mouths. As long as the bonds remain stable enough to rear young successfully, strict monogamy may nevertheless be compromised by sneak fertilizations. These are always to the advantage of males, and can be equally advantageous for females if a genetically superior male should present himself. Strict monogamy, defined in terms of sexual encounters, is probably more a human ideal than it is a common biological reality.

The nature and degree of polygyny vary greatly in their expressions, even among related species.16 Frequently they appear to be tuned to other aspects of life history. Among the hoofed mammals that graze and browse for sustenance, there are many different examples. The dik-dik is a diminutive African antelope that lives in brushy country and hides from predators. Dik-diks are found in seemingly monogamous male-female pairs. At the other extreme are the largest grazers like the eland and buffalo that congregate in herds where individual males fight to control the copulations of as many females as possible. In between is the impala: modest in size but considerably larger than the dik-dik. Impalas move in herds and flee from predators. When breeding, adult males establish territories and wait for a group of females and immature males to pass through. The male's life then becomes a frantic rush to copulate with as many females as possible while chasing young males. The male has to scramble for his reproduction rather than having to fight for it, and anyone who witnesses this performance is likely to feel pity for his plight as well as wonder how he accomplishes anything.

The basic social unit of the hamadryas baboon of Ethiopia is a single male and a harem of several females. These units may congregate in larger assemblages, forming bands and troops. The males jealously guard their females, biting them on the neck if they stray, and keeping other males from mingling with them. Other species of baboons from farther south in Africa have a somewhat different social structure consisting of troops of two or three dozen individuals of both sexes, adults and juveniles. Females remain in the troops of their birth, whereas there is some exchange of males between troops. The males have a dominance hierarchy that is reasonably stable over time but subject to rearrangement as individuals come, go, mature, and age. The frequency of copulations that males achieve corresponds with their social rank; higher ranking males are more successful in mating females. The system is not as exclusive as in the closely related hamadryas baboon, because a given female is not monopolized by a single male.

The primates as a group show great variation in social and mating systems. The orangutan (an ape) and a number of nocturnal prosimians (tree shrews, lemurs, loriscs, galagos, and relatives) are solitary. Both the Old World and New World families of monkeys contain examples of species that congregate in singlc-male, multifemale bands or multi-male, multi-female bands, along the lines described for baboons.

Our closest primate relatives, the chimpanzees, usually live in multi-male, multi-female groups, but some all-male groups also exist. When individuals transfer between groups, they are usually females, an unusual pattern for mammals but one also shared with gorillas and a few species of monkeys. The human social structure based on male kinship employing exchanges of women may therefore have its antecedents in the social organizations of the great apes. At one time it was thought that chimpanzees were indiscriminate philanderers, and it was hypothesized that male-male competition occurred primarily at the level of sperm productional (A chimpanzee produces more than twice as many sperm per ejaculation than a man, and twelve times more than a gorilla.) More recent field observations, however, suggest that copulatory successes of male chimpanzees are not uniforms Frequently at the time of estrus a female effectively pairs off for a few days with a single male. Higher ranking males are more successful in monopolizing females in this manner.

The theory of parental investment is supported by cases in which the roles of males and females are to some extent reversed.19 In a family of fish that includes the little sea horse, the female transfers her eggs to a pouch on the male, who then gestates and nourishes the developing young. Sexual selection has been at work on these fish, for the females are not only the more brightly colored but also the active wooers in courtship.

Phalaropes, a small group of shorebirds, show a similar reversal in the usual pattern of sexual selection. The females are larger, brighter, arrive on the breeding grounds first, and guard the males from other females. The males, for their part, take sole responsibility for brooding the eggs and newly hatched chicks. Why have male phalaropes become the limiting reproductive resource, with the greater parental investment? A plausible explanation for this phenomenon draws on the breeding ecology of shorebirds. These creatures nest on the ground in open country, usually in the far north. The breeding season is short, and the nests are subject to predation. Since the young hatch able to feed themselves (like ducks and geese and unlike familiar songbirds), both parents are not required once the eggs are laid. If the female can defect and find another mate she does so. In fact polyandrous mating (one female, several males) is practiced by other species of shorebirds that experience similar environmental problems and share the same precocial pattern of development. Polyandry is relatively uncommon among vertebrates but it is an exception that supports the rule of differential parental investment. We shall describe in a later chapter the special and equivalently rare case of polyandry in human societies.

The mating systems monogamy, polygyny, and polyandry represent but one measure of reproductive diversity. Some animals (and plants) concentrate their reproductive energies in a single but frequently massive effort. The butterfly characteristically lays her eggs and dies; the migratory salmon returns to the site of its birth only to terminate its life after an arduous upstream swim as part of a crescendo of reproductive exertion. Other species , however, reproduce repeatedly. The queen honeybee does little but lay eggs for several years, and birds and mammals generally reproduce seasonally, often more, sometimes less frequently. In animals, the capacity for repeated reproduction appears to be a prerequisite for the evolution of social organizations.

Life histories can also be characterized on a related dimension. At one extreme are species that produce many offspring, develop rapidly, exploit habitats quickly, and frequently have means for dispersal to distant sites. They frequently are found in unstable environments such as puddles or forest clearings, and there is a premium on at least some of the progeny finding suitable new places to live and reproduce. At the other end of the spectrum are longer-lived species that occupy more stable habitats, generally have larger bodies, develop more slowly, and reproduce repeatedly but have relatively few offspring. The former depend for reproductive success on a large number of offspring and often on the ability to colonize new areas. The latter depend more on their ability to exploit the habitat in which they live. Viewed another way, they also represent two extremes of parental investment, the first in fecundity, the second in nurture.

The two extremes were originally called, respectively, r-selected and K-selected species, a lifeless terminology that refers to parameters in the equations for growth of populations. The original concept was that some species have evolved with less emphasis on fecundity and more on competitive efficiency in the extraction of energy from their environment. As a consequence, in these latter species population sizes would tend to be more stable over time, but the densities might approach more nearly the carrying capacity of the environment. At the risk of oversimplification, this theoretical basis for r- and K-selection has been criticized on the grounds that r (the intrinsic rate of increase of the population) might reasonably be the object of selection, but K (the carrying capacity of the environment) is not a parameter of the same quality.20 Regardless of how one formulates causal evolutionary hypotheses, however, animals do display this variety of life histories, and it is useful to think of this diversity as different designs for fitness, or more specifically, as different strategies of parental investment. We see once again that there are multiple solutions to the evolutionary challenge of reproduction.

Long life, slow development, and extensive parental care all require attention to gamering food resources and a greater need for the individual to ensure its own safety over extended periods of time. So we might expect to find in such species evolutionary changes that have increased the effectiveness of sensory detection and efficiency of exploitation of energy resources, including behavioral corollaries such as territoriality and sociality. With this form of life history, prolonged survival becomes necessary for successful reproduction, and with fewer offspring and a long period of development, mechanisms to forestall premature demise are also necessary. Behavior comes to exercise greater control over the rates of both births and deaths. In later chapters we shall elaborate on the idea that much of what is referred to as human nature is understandable as a consequence of the evolution of a long-lived, slowly developing, resource-requiring, mildly polygynous social primate that also happens to be highly intelligent.

The origins of the relationships between the sexes in human societies have received much attention in recent years. Evolutionary hypotheses about male domination have been advanced, based on observations of contemporary hunter-gatherer cultures and on supposed divisions of responsibility in hunter-gatherer societies during the early evolution of humankind. To some extent these hypotheses recognize biological realities such as lactation, but they nevertheless remain narrowly based. As there is very little direct evidence as to how such societies were organized tens of thousands of years ago, it becomes possible to assert that they were truly egalitarian and that the domination of women is a cultural phenomenon that came later." This is both fashionable and facile, and as others have sensibly argued, the very common asymmetry between the sexes of animals means that arguments about the causes of sexual selection should not be based on the characteristics of any one species.

The results of sexual selection in humans are evident to those who wish to see. The mild sexual dimorphism of body size and strength is accompanied by competition between young males for women, a greater variance in reproductive success among men than women, extensive efforts of men to control the reproductive destinies of women, a greater tendency of men to seek multiple partners, rape as a coercion of women by men and not vice versa, and prostitution as a female profession. Human males are capable of nurturing their offspring to an extent unparalleled among mammals, but that has not neutralized the force of sexual selection. Women still make the greater parental investment, and the evolved reproductive strategies of men and women are not congruent.

Among the primates, about eighteen percent of the species are seemingly monogamous. Where do humans fit in this picture? Some regularized pairing of men and women akin to marriage is so widely distributed across cultures as to be a characteristic of the human species. Features of Homo sapiens that contribute to this practice include slow development of the young with correspondingly great parental investment, long lifetimes, and elaborate social structures that are made possible by the cognitive capacity of the human brain. But what do we see in human mating practices that fits into the larger comparative and evolutionary picture that we have been painting?

By one estimate, about eighty percent of human societies are at least mildly polygynous and twenty percent monogamous.22 These numbers alone, however, are misleading. Monogamous unions are in fact the most prevalent, even in societies where polygyny is acceptable. Polygyny is made practical when one man is able to accumulate sufficient resources to support more than one wife. Among the !Kung San people of Africa, whose traditional way of life is probably the clearest reflection still available of a preagricultural, hunter-gatherer culture, about ninety-five percent of the pairings are monogamous.21 And in Western Europe and North America, where monogamy is decreed by law, divorce and remarriage effectively relax the formal imposition of monogamy.

The idea that polygyny is associated with accumulations of wealth and power is documented by an interesting analysis performed by Laura Betzig.24 She compiled data on 104 autonomous societies from different places in the world and from different times in history with an eye to comparing information on polygyny and the means by which conflicts were resolved. She focused on a subset of twelve societies that exhibited the most complex political structure, with four hierarchical levels of organization. All were characterized by a despotic political organization, with vast power vested in a chief or king who settles disputes, frequently to his personal advantage. In all twelve, punishment of the aristocracy was less harsh than that of commoners. Interestingly, the concentration of power and despotic behavior correlates with extensive polygynous control of women, sometimes in such vast numbers as to preclude the potentate's sexual activity with all of them. In Incan society, access to wives correlated throughout the political structure. Depending on where a man stood in the formal hierarchy of principal persons, governors, administrators, and petty chiefs the system might provide "for their service and multiplying people in the kingdom" women in the numbers of 50, 30, 20, 15, 12, 8, 7, 5 or 3. Clearly a substantial number of men at the bottom of the hierarchy must have been without wives. This is a condition that ordinarily can be expected to generate considerable unrest and requires a despotic political structure to keep it from erupting.

Does polygyny increase male fitness? It certainly does for those males who hold resources and power. In all of the societies in Betzig's study, polygyny is associated with "a high degree of differential reproduction." This indicates why, in evolutionary terms, resources and power are important ends in themselves, a matter to which we shall return in later chapters. The promiscuous tendencies of men would seem to be obvious, and in Western cultures they have also been documented by attitudinal surveys.25 Eight or nine times as many married men as women-in total, nearly half of the men-would like to engage in extramarital sex. The same individuals also held the view that men have a greater sexual desire than women.

The sexual interest of male mammals can frequently be aroused by novel females. Are human males an exception? This phenomenon is called the Coolidge effect,26 supposedly in honor of a Presidential visit to a poultry farm. According to the story, Mrs. Coolidge, out of earshot of her husband, inquired whether a rooster could copulate more than once a day. On being told 'dozens of times a day' she requested that this information be conveyed to the President. When this exchange was reported to Mr. Coolidge, he asked, "Same hen?" On hearing "no, a different one each time" he is alleged to have said, "Tell that to Mrs. Coolidge."

Monogamy is a mating practice that increases the parental investment of the male, which is to the advantage of the female. Although any offspring will have the female's genes, we have seen that the male cannot have the same certainty as the female about the presence of his own. The female's genetic interests are fulfilled if the male sustains his parental investment in her children, but the male's interests are furthered only if he is their father. The male may therefore gain in fitness by casual liaisons with other females, but he loses heavily (in terms of Darwinian fitness) if he invests in another man's offspring at the expense of what might have been his own. On the other hand, female infidelity has a genetic cost only if she loses the support of her mate, and under some circumstances infidelity could have a genetic benefit. Monogamy increases the male's confidence in paternity, but it does not make it certain. It is therefore no accident that through history human males have expended much effort in trying to control the reproductive activities of females.

There are numerous manifestations of this asymmetry, both cultural and emotional.27 Female chastity is a virtual commodity in many societies. Unmarried women are chaperoned, veiled, hobbled, or otherwise protected or mutilated so that their value as future wives will not be compromised, and a transgression on their part can irrevocably soil the family honor. Violence induced by male sexual jealousy is a familiar theme in literature and life, and infidelity is the principal cause fof the killing of wives by husbands. Until very recent times, most cultures have had a double standard regarding infidelity in that the crime is defined in terms of the marital status of the woman. The cuckolded husband is the aggrieved individual and is frequently excused of violence if he takes the law into his own hands. Margaret Mead's myths to the contrary notwithstanding, male sexual jealousy is a significant source of violence everywhere in the world.28

Martin Daly and Margo Wilson,29 in a fascinating analysis of homicide statistics, have provided a great deal of insight into matters usually left to sociologists. Contemporary urban violence in the United States frequently seems to us now to be the result of drug wars, which to a large extent are also conflicts over resources. They are also largely conflicts between young men with otherwise limited economic prospects. In another generation, as indeed seems 'to have been the case for centuries in Europe and America (for which historical studies exist), a substantial fraction of all homicides involved young men fighting over utter trivia like a spilled glass of beer, a minor gambling debt, or a perceived insult to pride or honor. This is but one manifestation of what appears to be a need to demonstrate one's manhood, and it seems particularly prone to flare into violence where the participants have strained economic circumstances with little likelihood of improvement. But why should men and not women behave in this statistically predictable and dangerous way? What is it that drives the psyche of the young male to take such seemingly foolish risks? As the phenomenon is widespread and recurrent and the stakes frequently so high, the explanation of ultimate cause must involve something quite important to males. The most likely answer is that through time, access to resources and elevated social status have been very important in determining the reproductive success of males. Respect of peers is a major determinant of social status, and considering the ever-present hidden agenda that evolutionary history has provided, it is not at all ironic that the proximate goal of the participants in these altercations is to demonstrate that they "have balls."

Rape is a special form of violence that men inflict on women. It is not unusual for women in our present society to see rape as misogyny, an expression of hatred of women. A victim of rape who recently chose to make her story public was incredulous that as her attacker left he told her how pretty she was. The columnist who reported this story, also a woman, was equally shocked at the rapist's remark. Their bewilderment followed from their beliefs about the causes of rape.

Much sociological and psychiatric effort has gone into the study of rape, and the effort has illuminated some aspects of proximate cause. A smaller number of analyses have tried to put rape into an evolutionary perspective. Women react very negatively to rape for sound evolutionary reasons; the act not only subjects them to physical harm, it deprives them of their natural role of choosing with whom and when they will mate. Quite simply, rape is not in the interests of either women or their genes. As a further complication, rape can also compromise a woman's relationship with her husband; witness the tendency of men to lay blame on the victims of rape.

An evolutionary perspective suggests that, as in animals (e.g., ducks), rape has something to do with improving the rapist's Darwinian fitness.30 In our mildly polygynous species we might predict that rape is most frequently performed by men whose educational level is low and whose economic (and thus reproductive) prospects are poor. This is in fact the case, not just in the racially troubled cities of the United States, but in the ethnic and cultural homogeneity of Denmark, and in cultures where the marriage prospects of men are constrained by the need to purchase wives, the custom of 'bride-price." Economics is only one ingredient, for the typical rapist is not only a young man but one that has low self esteem and little sense of purpose in life. Whatever the proximate cause of his psychological profile, his prospects for increasing his fitness through orthodox means are poor, and he usually knows it.

The victims of rape are characteristically young women early in their reproductive years. If rape were an expression of hate of women, this would not necessarily be so. In fact, the age distribution of rape victims is skewed to younger years than the distribution of female victims of murder.

This overview of rape should not be taken to mean that men high on the socioeconomic ladder never try to coerce women into granting sexual favors. The way many men behave may be determined by their power over women and by their perceptions of the risks they are running. Groups of young men in fraternities and athletic teams can behave in insensitive, exploitative, and even criminal ways toward young women. And through history women have routinely suffered rape from conquering armies. None of this, however, detracts from the force of an evolutionary perspective.

The incidence of infanticide among mammals was mentioned previously as an example of how its explanation was muddled by the confusion of individual and group selection. Infanticide is not unknown in human societies either. Consider the advice of Moses to his people following their victory over the Midianites, in which

Numbers, 31

What could be a more explicit set of instructions both for eliminating reproductive competition, present and future, as well as for assuring paternity among the appropriated females?

Infanticide does not require the conditions of war. It has been encountered by anthropologists in a variety of cultures, where it is almost always an expedient to address one of several situations.31 One is the birth of a deformed infant. A second is the wrong father. A third, and numerically the most important, occurs when the mother, usually quite young, does not have the social support and resources to care for the offspring. Whatever moral judgment one cares to make from the comfort and security of a modem Western culture, these are outcomes that are evolutionarily in the interests of the mother's ultimate fitness.

Sometimes there is a selective killing (or neglect) of daughters, and there is some evidence that this is most prevalent in highly stratified societies. A likely explanation in evolutionary terms is that in such societies the reproductive potential of sons of the upper classes can be enormous, much greater than that of daughters, and amplified if the society is polygynous.

Infanticide also occurs in modem Western cultures, although the mechanism may frequently be neglect rather than a single decisive act. The frequency is low-less than three dozen per million children per age class per year-but the circumstances are interesting. Infanticide at the hands of the mother usually results from the same situations as in "primitive" cultures, and young mothers are more likely to be involved. Children who are at risk from parents are at much higher risk during the first year of life and are at higher risk from stepparents than from natural parents. The fables of Cinderella and of Hansel and Grctel are rooted in fact.

The relationships between men and women are like an intricately cut diamond whose appearance changes when viewed from different directions. In this chapter we have been peering at our reflections from just a few of its many facets, and we have seen that there is more to human nature than can be understood without biology. The social sciences can describe, but in their present state they are unable to explain, some of the deepest questions that are posed by the behavior of their subjects.

Let me now head off one reaction that is probably inevitable in today's social climate. This is not a political essay. In invoking evolutionary biology and the concept of ultimate causation I trust I will not be saddled with the view that because something is "biological" it is necessary or appropriate or right for human society or that I am defending any social or economic status quo. Quite the contrary, there are a number of aspects of human behavior-regardless of what their origins may be-that may be maladaptive or culturally inappropriate in the technologically complicated world in which we now live. Unarguably, homicide, rape, and a host of other forms of violence and exploitation are deplorable, yet despite both moral and legal sanctions they remain disturbingly ubiquitous. I have argued above, and I shall argue repeatedly again, that in order to address biological and social problems we must accept the inherent complexity of what is meant by the word "cause." In short, where there is a problem, there is much to be said for trying to understand it before attempting to solve it.

On August 23, 1989, a sixteen-year-old youth named Yusuf Hawkins went with two friends to the Bensonhurst section of Brooklyn in New York City to answer an ad for a used blue Pontiac automobile that one of the boys had seen. Yusuf and his friends were black, and unbeknownst to them they were entering an etlmic Italian neighborhood where some of the young men had worked themselves into a dangerous state of mind. The outsiders were met by a mob that has been estimated to have been as large as forty, and before he knew what was happening, young Yusuf lay dead of gunshot wounds.

This tragedy is easily seen as simply one more example of wanton urban violence and racism. It is all of that, but it is also more. The residents who set upon the three black youths were young and male. They were from the local community, and their victims were not. Furthermore, being black, Yusuf and his friends were easily marked as outsiders. Why had the mob assembled, looking for trouble from other young men entering the neighborhood?

The answer to this question is fascinating. Equally interesting, the answer has been widely reported with absolutely no discussion of its deeper meaning. One of the young Bensonhurst men had had an altercation with a young woman in the neighborhood, a onetime girlfriend who had black and Puerto Rican friends from outside the immediate community. The young man took exception to this display of independence, and on the day of the murder the two of them had exchanged unpleasantries. According to newspaper accounts of the testimony at his trial, he insulted her, and she taunted him with the threat of reprisal at the hands of her friends. Apparently he took her seriously, and he seemingly had no trouble in persuading a sizable number of compatriots to defend the local honor, turf, and male prerogatives from outside invasion. When Yusuf and his two friends appeared later in the day they were promptly set upon, and Yusuf was killed.

The inability of young men to control the behavior of a young woman of shared ethnicity when she showed an interest in men from outside the group was clearly one of the ingredients of this tragedy. Lest the reader dismiss this as gratuitous hypothesizing, know also that in the subsequent criminal trial, one of the defense attorneys did his best to blame the entire incident on the young woman, a tactic he obviously expected might resonate with some of the jurors. Moreover, some residents of Bensonhurst offered the opinion that the woman was responsible for the killing, and she reported having been threatened. At a number of levels in our culture today, there is a deep acceptance of men's efforts to control the behavior-rcad reproductive lives-of women.

The story continues, for the responses of members of the community to the murder and subsequent trials are equally revealing. There was considerable group solidarity in the aftermath of the killing. Despite the number of residents who were on the scene, the prosecution had great trouble finding anyone willing to testify as a witness, partly due to fear of retaliation by other residents. 'Bensonhurst amnesia' it was called. Bensonhurst nepotism would have been equally apt. There was concern for the image of the group. Following a conviction of one of the young men, a resident was quoted as saying, 'The system condemned Bensonhurst yesterday, and today it vindicated Bensonhurst.- Another asked, -why should everyone suffer because of the action of a few?" Residents also found a material basis for concern, complaining that their property values and businesses were suffering from the publicity. The issue was cast in a manner calculated to downplay its significance: " - - - it was a freak thing. They were young, and they didn't know what they were doing." Or it was described in terms that were supposed to sound understandable and thus morally acceptable, in the same vein as the legal defense that had been based on the suggested culpability of the woman: "Residents insisted that the killing was a mistake provoked not by racism but by a desire to defend home turf." Community expression also took on a taunting, derisive tone with displays of watermelons when the residents witnessed demonstrations by black citizens from other parts of the city, responses subsequently attributed to "outsiders."

There is much of interest in this ugly story, and it gives us another invitation to contemplate causes, immediate and remote, proximate and ultimate. Some of its essential elements sound familiar bells. The Trojan War? The hatred of Montague for Capuict? These themes recur in literature because they recur in life. Yet there was nothing even faintly romantic about the loathsome behavior of the young men of Bensonhurst. How do we explain what happened? Some blame individuals, others society; but why should we assume that these are alternatives? What do we really mean by human nature? This question brings us without further ado to the interplay of proximate and ultimate cause.

It is often said that there is no more sterile exercise than attempting to attribute particular behaviors to either heredity or environment. The effort is indeed without point or merit for most of the behavior of vertebrates. What is astonishing, however, is the frequency with which the presumption of which 'either nature or nurture" recurs and the forms it takes. Virtually every account of sociobiology in 11 science" sections of the popular press is reduced to this simplistic notion, but one does not have to search far to find serious scholars who have unwittingly garroted themselves on the same clothesline. Thus we find an eminent anthropologist writing (in criticism of sociobiology and E.O. Wilson):

What then of genes and more "open" programs in which all of the steps in a behavioral sequence are not prescribed?

non-gcnetically determined alternative action that sociobiological

The writer would have us understand that the existence of alternative behavioral choices, to be exercised with the benefit of learning, places the behavioral program beyond the reach of natural selection and therefore outside the concern of evolutionary biologists. That is what he seems to mean, and yet he also acknowledges a role for evolution when he writes:

Before we proceed further, we had better examine this la ment closely, because it contains a serious confusion. Wha relationships of "closed" and "open" behaviors to the genes? The writer, I submit, has it precisely backward. In explaining why, I am hard pressed to improve on Richard Dawkins's brief but eloquent description:

The execution of a behavior thus involves the operation o neural circuits that were laid down in development. The attempt to escape, an example of a relatively closed prog the blink of an eye or the jerk of a knee, does not require struction of new circuits. It needs only their operation, whi place on a time frame of small fractions of a single secon therefore does 'not directly involve the genetic code." On tt. 4 hand, if the open programs of behavior involve learning, the time frame changes, and there is much reason to believe that the processes that occur in the nervous system include the synthesis of new proteins. But I am getting ahead of myself. The immediate point is that by contrast with simple reflexes and other relatively invariant or closed programs, in its operation an open program of behavior is very likely to involve the participation of the genetic code.

At little risk of oversimplification, this kind of thinking can be summarized as follows. "Genetic" and "biological determinism" have come to be code words for forms of behavior that unfold xed path and cannot be significantly altered by environontingencies. Any deflection of behavior that seems to inoice (for human behavior, read "conscious choice") must be something else. As this "something else" is, by definition, not genetic, it should, according to this fallacious argument, lie beyond the scope of evolutionary biology.

The roots of this dichotomous view of behavior run deep, for "biological determinism" is just instinct warmed over. A number of years ago the psychologist Frank Beach' pointed out that from the Middle Ages until the nineteenth century, instinct was a theological rather than a scientific concept and referred to the gamut of apparently purposeful (adaptive in post-Darwinian language) behaviors exhibited by animals. It stood in contrast to human behavior, which was motivated by reason. Instinct was a logically necessary construct, because the exercise of reason was the path to the soul's salvation, and as only humans had souls, presumably only humans could reason. In the nineteenth century the same sort of binary classification was extended to scientific usage, with instinct coming to mean the alternative to learned behavior. As Beach recognized, there is no theoretical justification for supposing that behavior must be either genetically programmed or acquired entirely by experience. And, in fact, such a view of behavior is quite simply wrong. Moreover, as Beach pointed out, a classification in which instinct is defined as something it is not is operationally unsatisfactory. Logically no behavior should be classified as an instinct unless it is first shown by observation and experiment to appear without any contributions from learning, a consideration that is frequently ignored and is extremely difficult to deal with experimentally.

Today it is recognized that behavioral phenotypes are the result of the interplay between internal (genetic) and external (experiential or environmental) factors. Consequently the concept of .Species-specific behavior" I find more useful than the over-burdened term "instinct." Specics-specific behavior is exhibited by most members of the same species of the same sex and age and under equivalent circumstances. It is generally adaptive. It is not synonymous with the common understanding of instinct.

There are many examples of species-specific behavior that can be drawn from the literature of ethology, but I shall mention here only two, selected because they illustrate the interplay of internal and external factors in the development of vertebrate behavior. A few hours after hatching, goslings start to follow the mother goose when she moves away from the nest. This behavior is also characteristic of a number of birds that nest on the ground, such as ducks and many shorebirds. Moreover, it is adaptive in that it keeps the large clutch of down-covered, mobile hatchlings close to the protection of a parent. If the adult bird is not present, the goslings will follow any moving object, even a human being, and subsequently behave toward that object as though it were the mother. This seemingly unreinforced, single-trial learning is called imprinting.

Note that the genetic program for this following behavior is not complete, an important detail being supplied by the environment after hatching. In other words, the genetic program does not equip a gosling to recognize a mother goose from all other objects in the world; rather it creates a short time in the developmental processa critical period-in which the gosling's central nervous system is primed to receive and store some rather specific sensory information about mother's identity. Moreover, the information received at the critical period is important in determining the subsequent behavior of the young bird. The process works with high reliability in nature because it is rare that the mother bird is not the object of imprinting. The intervention of an ethologist is not an event that has influenced the evolutionary history of geese and ducks, and under normal circumstances their natural history virtually assures a normal outcome.

Another example that illustrates both the interaction between genetic and environmental factors and the notion of critical or sensitive periods in the development of behavior comes from studies of the ontogeny of bird song. The work on chaffinches and white crowned sparrows has been particularly instructive.6 If these songbirds are reared in isolation, their adult vocalizations are not completely normal. If the birds are deafened at hatching, so that they can hear neither themselves nor other birds, their adult songs are even more distorted. The songs of deafened birds, however, are not completely random. It is as though the birds possess some kind of internal representation of their species' song against which they compare and refine their own vocalizations. Deafened birds, unable to make the comparison, nevertheless produce a song that has a number of the appropriate elements present, and a match to the template is improved if the birds are able to hear the songs of other individuals during a sensitive period during the first months of life.

For these species there are severe limits to the degree to which the adult song can be made to differ from the species-characteristic form. Some variation can be introduced by exposing the birds to different sounds during the early sensitive period, but the internal template limits the amount of novelty that can be learned. The amount of flexibility is sufficiently great, however, that different populations of wild birds occupying different geographical areas may sing recognizably different "dialects."

In summary, these two examples of the development of speciesspecific behavior illustrate several features of general importance. First, the development of behavior involves an interplay of genetic and environmental factors. There is something in the genetic code that makes a bird a chaffinch and not a chicken, and part of being a chaffinch is singing a chaffinch's song. Likewise, part of being a gosling is tagging along after mom. But in the process of development, the bird's song comes into full being partly as a result of listening and practicing, and for the gosling, knowing which object is mother requires that the nervous system be primed with sensory input. Second, the interplay of genetic and environmental influences is not random in time, for there exist prescribed intervals during which the developing animal is particularly susceptible to specific external influences, and it is at these times that the developmental process can be tricked. Finally, the environmental influences are not random in character, for the developing animal may be much more susceptible to some external events than others. This last point is another way of saying that not everything can be learned with equal facility. This is an additional concept that we shall deal with in due course; but first, a few more words about development.

Behavior-what animals do-depends on their nervous systems. More precisely, behavior is determined by the microarchitecture of the nervous system-by an enormous number of specific functional connections (called synapses) between nerve cells (neurons) that have different shapes and that communicate with one another using different chemical messengers (neurotransmitters). The development of behavior therefore becomes, at one level of analysis, a component of one of the most central, difficult, and elusive problems in all of experimental biology: How does a fertilized egg, a single cell, give rise to an entire functioning organism with many different kinds of cells each one of which performs a different task? The code for this process clearly resides in the genome: Moss genes produce mosses and mouse genes produce mice. But just what information is explicitly stated by the DNA? Does the genome specify, in detail, all of the connections a developing nervous system makes within itself? A simple calculation shows that this is not possible. The human brain is estimated to contain about 1012 neurons and roughly 1015 synapses, but human chromosomes contain about 105 genes. Even if these estimates are off by one or two orders of magnitude, one can see that the instructions for wiring together the brain must be quite general in character. There is simply not enough information in the genetic code to specify in advance every synaptic connection, let alone the finer details of neuron geometry.

Biologists have recognized for decades that the process of development involves an interplay between information coded in the genome (genetic factors) and a continuum of external signals influencing how that information is expressed (epigenetic processes). In response to these signals, local populations of growing and differentiating cells change their character, frequently irreversibly. A central feature of this interplay is that each stage in the process creates the conditions necessary for subsequent steps to occur. Although this feature of development has been recognized by experimental embryologists throughout the century, until recently few developmental biologists have been interested in the ontogeny of behavior or of human mental function, and relatively few psychologists have immersed themselves in developmental biology. Consequently, with a few notable exceptions, the conceptual framework that I present in this section has been slow to enter psychology, as described by Ronald Oppenheim7 in an interesting essay on the history of epigenesis and preformation (the alternative and now discarded idea that the fertilized egg contains a miniature analog of every adult structure) as guiding ideas in the ontogeny of behavior.

The principle of epigenesis is at work in the very earliest stages of differentiation. For example, by the third week after fertilization, the embryo begins to form a nervous system. A group of outermost (ectodermal) cells, along the dorsal midline are induced to become the future nervous system through the influence of a chemical (a peptide) that is produced by underlying (mesodermal) cells. 'Induced" is a splendid descriptive word for what happens, and "induction" is actually the technical term that developmental biologists call the process. The presence of the chemical inducer sets the overlying ectodermal cells on a course of differentiation that is different from cells in neighboring regions of the embryo. Individually, the cells start to become either neurons or the associated supporting cells called glia. Collectively they form a hollow tube that in time changes further to become the spinal cord and brain. Moreover, the fates of these neuroectoderm cells become specified sequentially along the anterior-posterior axis of the developing embryo, presumably by a gradient of the inducing factor. The molecular details of how all of this occurs are not yet clear, but the process is believed to involve the local expression of different genes whose function is to regulate the expression of still other genes.8 Thus among the first of the external signals influencing the developing nervous system is a chemical substance, produced elsewhere in the embryo, whose qualitative effects are in part determined by the spatial and temporal relations that the target cells have to one another in the embryo. The process is very complicated, because the embryonic cells are themselves moving, individually migrating and specializing, and collectively molding the shape of the embryo.

Subsequent differentiation of the nervous system continues to involve cell divisions and the creation of identifiable classes of cells, migrations of various classes of cells to new locations, growth of the long extensions of nerve cells called axons and shorter processes called dendrites, and the formation of specific synaptic connections between axons and dlendrites from different cells. Furthermore, the various functional classes of cells appear at different times in development in response to specific local conditions. Differentiation continues to involve the activation of select subsets of genes. For example, the fertilized egg contains the information to direct the synthesis of all of the synaptic transmitters found in the adult; therefore, when a class of nerve cells commits itself to the use of, say, serotonin, genes for the appropriate synthetic enzymes are selectively activated and genes for other transmitters lie quiescent. Molecular signals may influence many cells or potential cell types, as in the example of neural induction, or they may have more restricted targets, as for example the influence of "nerve growth factor" on the induction of enzymes needed by neurons that use the transmitter epinephrine.

The identification of the external signals that control or modulate the development of the nervous system is in a rudimentary state. What seems clear, however, is that there is probably a variety of influences of rather different kinds. We have mentioned diffusible molecules, which are one example. The growth of the long cell processes called axons that make up the fibers in nerves is frequently directed along a scaffolding of supporting glial cells, or even other axons, indicating the importance of surface contacts. In other words, growing axons seem to "feel' their way along, detecting molecular cues as they go. Moreover, some of these contacts involve very specific recognition mechanisms, evident as cells grow into distant regions and select appropriate partners with which to synapse.

The process of development also includes elements of chance. For example, many more neurons of the type that control muscles (motor nerves) are formed than are ultimately employed, and those that fail to make the right connections in timely fashion subsequently die. There is thus a programmed redundancy in the pool of cells whose developmental fate is to be a motor neuron, and if a cell is not needed when all of the necessary synapses are formed, it is not kept.

Furthermore, the influences of one cell upon another can have elements of reciprocity. The formation of contacts between the terminals of a growing motor nerve and the muscle it is destined to activate makes the muscle less receptive to other nerve fibers and causes receptor sites for the synaptic transmitter on the muscle membrane to congregate in the region of the synaptic junction. Moreover, the speed of contraction of the adult muscle-the chemical character of the contractile machinery-is influenced both by the frequency at which the motor nerve delivers nerve impulses as well as by an unidentified trophic molecule secreted from the nerve terminal.

To summarize, although the molecular details of the process are far from clear, the nervous system develops through a complex series of genetic and epigenetic events in which the results of each step set the stage for the next. Along the way, cells, in effect, make a series of developmental choices, for the most part irreversible, about which genes to activate, in which directions to grow, and with which neighbors to settle. The outcome of each choice is determined by both the immediate local environment and the past history of the cell. The time at which each external influence plays its role on a given cell is thus crucial; the process of differentiation is a stream of critic trends. 9

Let us look at several more specific examples that extend to times later in development, for they enlarge the concept of external influences. Although the sex of mammals is genetically determined, the process by which this determination occurs is such that the genome can be fooled. 10 Genetically, males differ from females in one of the pairs of chromosomes. In females these two sex chromosomes (called X chromosomes) are equal in size. Males, however, have one X chromosome and a small Y chromosome. A mammalian egg always has a single X chromosome, whereas sperm have either an X or a Y chromosome. The sex of offspring is therefore determined by the sperm, a particular irony considering the frequency with which, throughout history, wives have been blamed for the inability to produce sons. (In birds, the roles are reversed, and it is the genotype of the egg that determines the sex of the offspring.)

Maleness of mammals results from the production of testosterone during critical periods during development and the action of tesrone (or one of its derivatives) on the developing embryo. The earliest developmental fork in the road to maleness or femaleness toocsctuers when a gene on the Y chromosome becomes activated and sets the fetal gonad to becoming testes instead of ovaries. At this writing the product for which this gene codes has not been identified, but it is thought to be a protein that binds to DNA and regulatcs the activation (the transcription) of other coding regions in the DNA. The developing testes begin to produce the steroid hormone testosteronc, which, as we shall soon see, has, in its turn, pronounced effects on the subsequent development of the rest of the reproductive system and on the brain.

In humans the early activation of the sex-determining region on the Y chromosome is thought to occur somewhere between the forty-third and forty-ninth day of development, and herein lies a digressive tale. Shortly after the discovery of this gene was reported, it received some publicity in the popular press, and under "Religion Notes" The New York Times gleefully proclaimed "Talmudic wisdom confirmed.... The Talmudists cited page 60a in the tractate B'rachot in which the ancient rabbis question whether it is worthwhile for the husband of a pregnant woman to pray for a son. The rabbis determine that such prayers are worthwhile up to the 40th day of pregnancy, but after that point the supplication is a I vain prayer' since the sex is already determined."' We can speculate that ancient rabbis were not operating totally in the dark; they had probably taken a close look at six- to eight-week-old fetuses and had taken their cue from what they saw. The religion editor of The New York Times and his Talmudist sources, however, have missed the point. In the course of normal development, the sex of the offspring is of course placed beyond the reach of prayer at conception, not when the sex-detcrmining gene on the Y chromosome is first activated.

As fetal development progresses, testosterone synthesized by the developing testes becomes a major player in organizing the rest of the reproductive system. An early effect of testosterone is to promote the continued development of the Wolfflan ducts into the male sex organs, a process that is accompanied by degeneration of the Miillerian ducts, which is in turn stimulated by another molecule (a protein rather than a steroid) secreted by the developing testes. Without this organizing influence, the Miillcrian ducts continue developing to become the female reproductive organs, and it is the Wolfflan ducts that degenerate. In humans there is a genetic mutation that blocks the molecular target for the androgen (a generic name for male steroid hormones), and such individuals, although genetically XY, have the cxtemal,genitalia and sexual interests of females.12 They are not reproductively functional females, however, because they do not develop the internal female reproductive organs. They have testes, which do not descend and do not produce spermatozoa as in a normal male. Conversely, genetic females that receive an early exposure to androgens develop the external genitalia of males. This can result from a genetic defect in which the adrenal glands produce the wrong steroid hormone, one with the activity of testosterone. From such observations as these it is apparent that in mammals the female condition is in a sense the base condition, and maleness results from a supplemental occurrence during development. This makes for an engaging contrast with the biblical story of Eve's creation from Adam's rib.

The steroid hormones also influence the developing brain. In rats there is a critical period during the first five days after birth, which fact has allowed considerable experimental manipulation. Genetically male rats that are castrated on the day after birth do not receive this critical exposure to androgen, and if they are subsequently given estrogen as adults, they exhibit female behavior. Exposure of the developing rat to androgen during the critical period not only determines aspects of adult behavior (e.g., the receptive posture called lordosis in females, more aggression in males), but produces a demonstrable sexual dimorphism in other features of the brain. In terms of neuroanatomy, differences have been described in the preoptic area of the hypothalamus at the level of cell numberl3 and in the distribution of synapses on dendritic shafts relative to spines." These small differences in architecture in the regions of cells near synaptic junctions with other cells can be seen with the electron microscope, but their precise functional significance is not yet known. They very likely correlate with functional differences, however, because all neural activity has a structural basis, even if it is only expressed at the level of molecules. At a biochemical level, hypothalamic cells (a part of the brain) of adult males show lower binding of estrogens to intracellular receptor proteins, and the male brain does not exhibit a cyclic pattern of secretion of luteinizing hormone from the anterior pituitary, whereas the female does.

Sex differences in behavior are not limited to the act of reproduction, and the sexually dimorphic changes in neuronal architecture that result from early differences in the hormonal environment extend to the cerebral cortex. Male rhesus monkeys show an earlier specialization of the prefrontal cortex in certain spatial leaming tasks, and prenatal exposure of female fetuses to androgens will abolish this differences In humans the left-right specialization of the hemispheres develops later in girls than in boys," although there is no direct experimental evidence that this is the result of differences in the prenatal environment.

The example of sex determination is interesting for two reasons. First, it shows that external influences on the genome of developing cells need not originate from adjacent structures. Hormones released by remote tissues can reach throughout the body of even an adult, and during development they can affect the transcription of specific portions of the DNA in select target cells. (The next example, however, will expand even further the concept of external influences.) Second, it provides a clear instance of measurable differences in the structure and biochemistry of brain cells that correlate with differences in behavior. There is really no alternative to the premise that structural and biochemical properties of neurons and groups of neurons lie at the basis of behavior, but there are relatively few natural examples in which variations in mammalian behavior can be correlated with structural properties of individual kinds of nerve cells. A principal reason is that the tools currently available are, in general, not adequate to the task.

Although there are clearly sex-related differences in the cellular structure of mammalian brains, the full implications of these findings for human behavior are far from clear. Humans retain an enormous behavioral plasticity, and the early social environment can doubtless influence the way men and women relate to one another. -But at this juncture in the argument we have seen that Homo sapiens not only exhibits sexual dimorphism, we have also witnessed how some of this comes into being during development, how and why it is widely shared with other species, as well as how it is widely manifest in human behavior around the world. What we do with this information in order to produce a more equitable social structure is a different matter, and not one that is intractable. The challenge is not made any easier, however, by ignoring the reality that evolution has made us what we are.

Studies of the visual system have revealed still another dimension to the interaction between intrinsic and extrinsic factors during developments Individual cells in the visual cortex of mammals respond to precisely oriented edges, or moving edges, in restricted parts of the visual field. Moreover, many of these cells receive convergent messages from corresponding parts of both retinas. These cells are properly "wired" at birth or shortly thereafter; that is, during development all of the neurons between the receptors in the retina, the lateral geniculate nucleus (a relay station in the thalamus), and the visual cortex have found the correct neighbors with which to synapse, and visual information is processed along this pathway much as it is in the adult. But something is still missing in order for the pathway to be consolidated: namely, sensory experience during a critical period shortly after birth. If mammals are prevented from seeing for weeks or months postpartum, they appear unable to resolve images and behave, as adults, as though functionally blind. If only one eye is covered, even with a translucent screen that allows light to pass, but no images, subsequent examination of the visual cortex reveals many fewer cells that respond to stimuli from both eyes. The synapses from cells driven by the covered eye fail to become validated and they decrease in number, their places having been taken by connections from the functional eye. A similar outcome occurs if, instead of covering one eye, the muscles that move the eyeball are cut so that the operated eye cannot be made to look at the same object as the unoperated eye. When the two eyes are unable to view the same point in space simultaneously during the critical period, they are subsequently unable to work together in analyzing the world. Again, this is because of a failure of nerve fibers with information from the two eyes to form synapses on the same cells in the part of the cerebral cortex to which they report. Thus we see that sensory experience-or the lack thereof-can cause irreversible structural changes in the nervous system and that this can have profound consequences for behavior.

The larger value of this discovery is that it gives us a perspective on other aspects of the ontogeny of behavior by extending the concept of critical or sensitive periods. The Harlows'18 well-known study of the development of social behavior in monkeys fits this picture well. Young monkeys deprived of physical contact with both their mother and their peers grow up with severe behavioral deficiencies and invite the characterization of being neurotic. Somewhat less severe effects are produced if the infant has contact with its mother but not with other youngsters. The effects of deprivation are, furthermore, difficult to reverse. Similar stunting of the emotional growth of human children results from early lack of social contactl9 and the acquisition of first language seems to occur during a sensitive period.20

We should therefore view the processes of development as extending well after birth and involving periods of time in which certain broad classes of external events are necessary for fine-tuning the synaptic structure of the nervous system. For example, this is the role of play, so prominent in social species from primates to carnivores, where functioning as an adult involves complex interactions with other members of the species-interactions in which aggressive and cooperative impulses need to be appropriately balanced.

Many of the later stages in the maturation of the mammalian nervous system have been studied as part of psychology. They have therefore been described as leaming or socialization, and generally in language that does not do much to stimulate thought about the underlying cellular processes. The work that has just been discussed, however, suggests that a more biological orientation would be useful, at the very least in providing a conceptual framework more likely to provide deep understanding in the future. There is, in short, good reason to see critical and sensitive periods in the development of behavior as part of a long developmental process in which the information contained in the genetic code can only be expressed through an intricate series of interactions of the partially completed product with a variety of extrinsic events. The final tuning of the nervous system-the construction, consolidation, and validation of all of the synaptic connections required for the full array of social behaviors-requires a variety of sensory and motor experiences extending, in our species, for years after birth.

Development clearly involves changes in structure, and what is ordinarily called leaming must also involve structural changes in the nervous system, too. The immunity of long-term memory to changes in metabolic rate due to lowered body temperature or to anesthesia imply some structural base. Cooling or disruption of function with anesthetics do not erase memories. Memory is thereforc not simply the result of a pattern of metabolic activity of neurons to be lost like the letter you were composing on your computer when the power failed. That memory survives when the brain is even partly shut down means that it must have a physical representation in the architecture of the brain. This structural basis is believed to reside in the organization of neurons and the synapses that connect them. Alterations of synaptic structures or efficacy22 as a result of sensory experience or leaming have been described. The key to understanding leaming and memory will likely be found by further study of structural changes at both the cellular and molecular levels.

As an aside, the tradition of viewing learned behavior as somehow nonbiological has its counterpart in medicine. Psychiatrists traditionally classified mental disorders as organic-meaning some relatively gross structural abnormality appears in postmortem examination-or functional-meaning no structural changes are evident. It is now more commonly recognized, however, that all mental disorders have a biological basis, whether there be gross tissue damage due to injury, genetically prone, environmentally provoked imbalances in transmitter biochemistry, or subtle morphological and biochemical abnormalities caused by inadequate sensory experience (social environment) at an early sensitive period.23 Traditional diagnosis compounds the nominal fallacythe illusion that by naming a phenomenon one has gained understanding. As with the supposed distinction between nature and nurture, the classification of mental disease as organic or functional makes understanding more difficult.

As much of leaming occurs as a natural part of development and shares with the development of gross morphological change alterations in cellular and molecular structure, is it sensible to view all of leaming as part of the process of development? The aphorism about the difficulties of teaching old dogs new tricks notwithstanding, mentally active people like to feel, with considerable justification, that they continue to learn throughout their adult life. But development, with its critical periods, would seem to be a series of opportunities, which, if once missed, are gone forever. Surely this is not a description of leaming as it is generally experienced. Failure to optimize one's income tax return never prevents one from leaming how to do the job better the next year. A useful way to view the relation between development and leaming is through the common influence of the genes. Without question, the brains of adult mamnials in general, and humans in particular, are endowed with a plasticity that enables them to continually adjust their behavior with experience. The developmental process does not tie down every conceivable synapse in a rigid and unalterable form, but leaves considerable scope for ongoing readjustment in the adult. At the same time it seems equally true that certain kinds of compctence-perceptual, linguistic, social-do need to develop on schedule, or the deleterious consequences are reversed with difficulty, if at all. This is because the capacity for leaming, like the development of body form, is subject to some genetic constraints.

The anthropologist Gregory Bateson, in a serious effort to place mind and biological evolution in a common framework, has contrasted ontogeny with natural selections Development is conservative and predictable; to use his terminology, it is a convergent sequence. Evolution, on the other hand, feeds on randomness. The next event is not precisely determined by the last and is therefore not predictable. The sequence is divergent. The formation of mind, and Bateson defines mind in sufficiently general terms that it is not necessarily a uniquely human attribute, is a stochastic affair. As with natural selection, the essence is exploration and change.

I believe this view of how brains work seriously underestimates the steering that evolutionary history imposes on the behavior of animals, including humans. In contrast, both proximate and ultimate aspects of cause are addressed if Icaming and development are seen in their proper relation to the genes. (Bateson in fact wavers and is not always consistent. Cultural transmission becomes a hybrid, and 'there is, surely, always, a genetic contribution to all somatic events.") We will return to the evidence that leaming is not a totally open process in a following section, but first a few final words about cellular development and change in the nervous

system.

Species differ in the amount of external programming of behavior that is necessary during development. A caterpillar needs no practice to spin a cocoon, whereas a child needs considerable experience to master a language. Indeed, one might place animals along a scale according to how much external programming is required during development, and doubtless our species would be at one end of this array. Such a onc-dimensional image of behavioral diversity would be misleading, however, because it speaks only to very general interspecific comparisons and does not recognize that in any one animal behaviors have diverse and interlocking ontogenies. Recall the example of social development in young primates that was discussed above. The process involves a mixture of behavioral elements developing at different rates with varying degrees of autonomy. Consider further the importance of nonverbal communication, and in particular, the smiling of humans. Smiling conveys information about emotions, and as it implies a state of pleasure it is frequently employed as a signal of reassurance in encounters between strangers. The fact that certain individuals can learn to use smiling deceitfully only underscores its general importance. But smiling is not something human infants must learn as a result of visual or auditory cues from other people; smiling first appears in congenitally deaf and blind children at the age of several months under conditions when they are apparently happy.25 It does not take much imagination to envision the social difficulties an otherwise normal child would experience should he or she be afflicted with an inability to smile at all, or a compulsion to smile exclusively at inappropriate moments. This one defect, in a behavior that appears without any identified leaming, would have manifold consequences cascading through the process of socialization. It would change the nature of the feedback loops that operate during the acquisition of social experience, and although one can anticipate compensatory behavior on the part of close friends and relatives, one can also imagine severe episodes of ridicule and ostracism and ultimately a warping of personality. The point of this hypothetical example, however, is not to generate disagreement about the details of the outcome, but to emphasize, from one more perspective, the essential inseparability of intrinsic and extrinsic factors in the development of mammalian behavior.

To summarize where we have come so far in this chapter, a detailed understanding of the proximate causes of behavior leads us, inevitably, backward along the paths of development. Some of these paths we find are much straighter that others, as though the genes had provided the zygote with an unambiguous road map. In other instances, however, we see that the developing animal received additional instructions along the way. Moreover, for complex behavior we find that different elements may have traveled aiong separate paths for part of the developmental journey. These paths may branch or merge, generating a tangled maze of social behaviors in which the origins of any path are hard to discern. For some of the straighter paths, instinct may remain a useful word. But for much of the behavior of higher vertebrates, attributing behavior to either leaming or instinct is about as fruitful as arguing whether it is the sugar or the flour that makes the cake.

Lumsden and Wilson 2l have summarized the evidence that a prominent feature of the way the human brain deals with a complicated array of possibly conflicting information is to try to reduce the options to a dichotomous choice. This certainly is evident in the nature-nurture controversy, and understanding is not furthered by trying to allocate a percentage of human behavior that can be accounted for by the genes, with the balance to be attributed to culture. The biologist John Bonner has written, " . . . it is not clear that one will ever be able to determine to what extent any human action is genetically or culturally determined."27 I would go farther and assert that the proposition has no meaning.

To close this discussion of development, let's contemplate briefly the termination of the life cycle. Development from fertilized egg to adult is just one end of a series of biological changes that all organisms experience. For more than a century, development has represented a central challenge to experimental biologists, and it is currently a major focus of research activity. The inevitability of aging and death haunts the human consciousness and would seem to pose an equivalent intellectual challenge, yet we seem to have little in the way of explanation for why this happens to us. We plant annuals and perennials in the garden and we stand in awe of trees that have lived for centuries. We know that many insects may live for only days or weeks, a dog or cat for little more than a decade, while we and elephants and sea turtles may hope for several score years. But whether an individual is claimed by accident or infirmity, the end is inevitable. Why?

Sometimes the answer to this question is souyht in terms of proximate cause. Perhaps our parts wear out, and if we just knew more about repair we could greatly extend our lives. And so we put a modest amount of money into research on aging, or gerontology, as the field is called. Interestingly, medical advances have increased the average life expectancy, but this has been accomplished solely by preventing death during younger years. Nothing in medical science has extended significantly the maximum age of about a century that individuals can reach; we simply increase the number of individuals who live to approach that maximum.

As we get along in years, a number of unpleasant things happen to us, including susceptibility to a host of ailments such as coronary disease, cancer, dementia, broken bones that do not heal, and so forth, and it becomes just a matter of time until one or another of these afflictions carries us away. We may not like to think about it this way, but we are programmed to age and die.

If that is the correct way to view the matter, then perhaps we can find some meaning, or at least explanation, if we wonder about ultimate cause. But to arrive at a sensible answer I think we have to recognize again the great variety of life-history strategies that living forms have adopted. Different forms of life cycles represent alternative ways genetic information has of propagating itself through time. It is not that one strategy is better than another; all that now exist have simply proved to be adequate to the challenges that they have met in their evolutionary history. Nothing will be gained by looking for the advantage of one compared to another, although we can make some sensible statements about ecological conditions that may either favor or compromise success of one or another strategy. To find the answer to our riddle, we should instead look for arguments that can be applied more generally.

It matters little where we start, so let's begin with ourselves. In our life history, our young take time to develop strength and experience, and it would be pointless for reproductive capacity to mature so early that novice parenting would not be successful. Within this constraint, imposed by the particular life history of our species, there is nevertheless reason for individuals to start reproducing as early as is consistent with a successful outcome. One reason why this is so is that the longer an individual postpones reproduction, the greater is the probability that he or she will either succumb to an accident before being able to reproduce or that there will be unwanted mutations in the germ line. If for any reason the probability of successful reproduction declines with age, regardless of physical condition, then those individuals that make an effort to reproduce early will, on average, leave more and healthier off-spring than those that defer. In our species the risks of pregnancy are in fact greater for older women. Furthermore, earlier reproduction hastens the contribution of the bearer's genes into succeeding generations. Everything else being equal, selection will favor early reproduction, but not so early as to jeopardize success. In itself, this is a compromise.

Two other codsi-derations bear on the argume tri@ act of reproduction carries with it sufficient cost-for example, either direct risk to life or indirect risk through weakening or added exposure to danger-there may be an effective limit to an individual's capacity for reproduction. Perhaps it is for this reason that human females produce but several hundred eggs in their lifetime. If there is a limitation on total reproductive effort, and selection is eni couraging early reproduction, the reproductive potential of individuals will decline after a certain age. As it diminishes, however, various forms of selection for somatic survival will be diminished. There will be decreasing need, in an evolutionary sense, to keep the body in the pristine shape it enjoyed when reproductive potential was high, and defenses against the ravages of tumors may relax. These changes may proceed slowly, as in humans, where an individual's contributions to inclusive fitness may continue through behavioral means long after direct capacity for reproduction has ended. Alternatively, changes leading to death may be cataclysmic, such as with salmon, where reproductive effort is compressed into a single act with no subsequent parental investment.

This argument is based on the observation that most genes have multiple phenotypic effects and the inference that genes may be beneficial at one stage of the life cycle and neutral or deleterious at another.28 The process of development in fact involves the differential expression of genes, so in general genes can and do have different effects at different times. It is more speculative whether some genes and gene products that play a positive role in developing or in reproductively competent individuals may have deleterious effects later in the life cycle. Such genes would be selected for, but would contribute to the process of senescence. The uncertainty about the effects of estrogen supplements for postmenopausal women, however, provides an example that may make the concept seem more than plausible.

temperament characteristics are selected for in females by males.

16. What strategies are useful for females in selecting a good male mate