Twenty-five years ago this summer a Harvard graduate student named Sarah Hrdy went to northwestern India and met the monkeys that would make her famous. The immediate impetus for the trip was a series of lectures by Stanford ecologist Paul Ehrlich on the dangers of overpopulation. Though Ehrlich was speaking about humans, what Hrdy thought of were the Indian monkeys known as Hanuman langurs. The langurs are considered sacred by many Indians and so are regularly fed by the people with whom they come into contact. Consequently, near towns, Hanuman langurs live in extremely dense populations, and apparently this unnatural density had led to unnatural, pathologically violent behavior. There had been several reports of adult males killing infants. So there I was, listening to Ehrlich, says Hrdy, with this adolescent desire to go do something relevant with my life, and I thought, ‘I am going to go study the effects of crowding on behavior.’ Hrdy traveled to dry, deforested Mount Abu and began to get acquainted with the sandy-bodied, dark-faced Hanumans. Before long she decided the assumption that had propelled her to India had been wrong. It happened pretty fast. I was watching these very crowded animals, and here were these infants playing around, bouncing on these males like trampolines, pulling on their tails, and so forth. These guys were aloof but totally tolerant. They might show some annoyance occasionally, but there was nothing approaching pathological hostility toward offspring. The trouble seemed to be when males came into the troop from outside it. The langurs of Abu are arranged into two kinds of groups. In the first, a single male--or, rarely, two--lives with a group of females and their infants. The infant females, when they grow up, stay put; the males leave to join the other kind of group, a small all-male band. Eventually a grown male lucky enough to be in a troop of females will come under attack, either from the all-male bands or from the male of another mixed troop. Odds are that sooner or later the resident male will be chased out by a new one. Hrdy witnessed many such takeovers, and she noticed that afterward the new male would often chase after the babies in the troop, presumably all of which were offspring of the old male’s. Before long some of these infants would disappear. She didn’t actually see what happened to the infants, but townspeople around Abu told her that they had seen a male killing baby langurs. Soon after these takeovers, the new resident male would mate with the females. I realized I needed a new explanatory model, says Hrdy. Her new model would become one of modern biology’s most famous--and in some circles, notorious--hypotheses about animal behavior. There was nothing pathological about langur infanticide, she suggested. On the contrary, it actually made a chilling kind of sense: While a langur mother nurses, she cannot conceive; when she stops nursing, she can. Thus if a male langur kills her infant--one that is not related to him--she can bear the infanticidal male’s own offspring. In Hanuman langur society, in which a male’s sojourn with a harem averages a little over two years, the time saved can be critical. After all, for any offspring to survive, they should ideally be weaned before a new, potentially infanticidal male shows up. Seen in this light, infanticide could actually be an adaptive evolutionary strategy for fathering as many offspring as possible. In the quarter century since Hrdy first conceived this idea, naturalists have reported cases of infanticide among a wide range of animals. Some now argue that the threat of infanticide is such a pervasive and powerful influence that it can shape animal societies. A few theorists even claim that infanticide was an important factor in human evolution. Yet when Hrdy first published her hypothesis, she was immediately attacked, most of all by other researchers studying langurs. They contended that Hanuman langurs living in natural conditions, in remote forests, had never been observed killing infants. At Abu, they said, human feeding had crowded the langurs into conditions evolution had never prepared them for, and as a result the transfer of males into new groups became drenched with aggression. In other words, the langurs of Abu were simply not normal. Underlying this species-specific dustup, though, was a deeper conflict. Before the 1970s most researchers viewed animal societies as smooth-running systems in which each member knew its proper role and played it for the good of the many. Animal societies--and primate societies in particular--were often portrayed as utopias that we humans would do well to emulate. But then biologists such as E. O. Wilson and Robert Trivers (both mentors of Hrdy’s at Harvard) argued that such a view didn’t make sense in a world shaped by evolution. Just as a bird’s wing is the product of natural selection, so are the ways the individual bird interacts with other birds. Its social behavior, like its body, is ultimately designed for one purpose: to get its own genes duplicated as much as possible. Rather than being a peaceful group of community-minded role players, an animal society was made up of individuals trying to maximize their reproductive gain, with cooperation always a compromise between competing genetic interests. As a biological explanation for society, this school of thought came to be known as sociobiology. When Hrdy hypothesized about langur infanticide, then, she wasn’t just explaining the odd behavior of a few monkeys. She was pushing sociobiology to its logical extreme, in which a male’s drive to reproduce was so strong that it would resort to the decidedly antisocial act of killing the young of its own species. Over time, Hrdy added some nuances to her stark hypothesis. For example, she noted that female langurs did not passively sit by as invading males tried to rob them of their genetic legacy. Rather, females banded together to help fend off males bent on infanticide. Once an infanticidal male was in charge, however, they might choose a different tack. Female langurs can continue to have sex even after they conceive, and by mating with an invading male, they might trick him into thinking the infant was his own. Hrdy also began noting reports of infanticide among other animals. Her outsider male infanticide, she realized, was clearly not the only kind of adaptive strategy practiced in the animal world. A mother might resort to infanticide if she didn’t have the resources to raise all her children. Adults might also kill the infants of strangers simply for food or to eliminate the competition for limited resources. In the decade and a half since Hrdy’s work first appeared, infanticide has been reported among mice and ground squirrels, bears and deer, prairie dogs and foxes, fish and dwarf mongooses and wasps and bumblebees and dung beetles. Although the evidence from the wild has often been sketchy, most of the strategies appear to fall into one of those that Hrdy sketched out. In a few cases researchers have been able to test the hypothesis by performing natural experiments. In 1987, for example, Cornell ornithologist Stephen Emlen was studying the jacana, a Panamanian bird in which the common sex roles are switched: males sit on the eggs and raise the young alone while the females rove around their territories, mating with many males and fighting off intruding females. Essentially, if you turn Hanuman langurs into birds and switch the sex roles, you get jacanas-- and theoretically, under the right conditions, you should also see infanticide. Emlen needed to shoot some birds for dna testing, and he chose two females with male partners caring for nests of babies. I shot a female one night, and the next morning was just awesome. By first light a new female was already on the turf. I saw terrible things--pecking and picking up and throwing down chicks until they were dead. Within hours she was soliciting the male, and he was mounting her the same day. The next night I shot the other female, then came out the next morning and saw the whole thing again. Among mammals, one of the best documented killers of infants has proved to be the lion. Though actual killings have only rarely been witnessed (the total is about a dozen), massive indirect evidence for the phenomenon has been gathered by the husband-and-wife team of Craig Packer and Anne Pusey, both behavioral ecologists at the University of Minnesota. From their observations of lions in the Serengeti, they’ve found that whenever a new male comes into the pride, the death rate of nursing cubs-- and nursing cubs only--shoots up. Within six months none of the cubs are left alive. Other primates have also joined the ranks of the infanticidal. The first reports were of only a few species such as the red howlers of Venezuela, the gorillas of Rwanda, and the blue monkeys of Uganda. But in the past few years there have been more reported instances of primate infanticide, some of which demand some expansion of Hrdy’s ideas. Lemurs in Madagascar, for example, breed once a year. If a male kills another male’s nursing infant, he doesn’t hasten his own fatherhood, since he still has to wait until the breeding season to mate. Nonetheless, researchers have seen male lemurs sinking their fangs into babies. One observer, Michael Pereira of Bucknell University, offers an idea as to why it happens. Madagascar has a harsh climate, with a long dry season that keeps female lemurs on a knife-edge of survival. Pregnancy and the first few months of nursing take place in the harshest time of the year, and, says Pereira, successfully raising an infant one year may reduce the chances a mother will be able to raise the next year’s baby. Females who lose their infants are much fatter than females whose infants survive, Pereira explains. If by your killing the infant she’s more likely to be successful during your reign, then it’s to your advantage. In Sumatra, Dutch researchers have been studying a relative of the Hanumans known as the Thomas langur. Thomas langurs were essentially a mystery when the Utrecht University primatologists began to observe them in 1988. Now, after eight years of relentlessly tracking the animals through the forest and painstakingly recording their daily habits, the researchers are finding the langurs to be all too revealing. Infanticide does occur. I’ve seen the attacks, explains Romy Steenbeek, who ran the program for four years. We saw the body of a baby with canine slices in its belly. I saw a male attack a baby, and the baby disappeared. One baby received big wounds in an attack by a neighboring male, and she died after two very bad weeks. The males run a few hundred meters to the troop, silently attack, and when they leave they loud-call. Like lemurs, these langurs require yet one more variation on Hrdy’s grisly theme. Thomas langurs have the same all-male bands and one- male/many-female troops of other langurs, but what distinguishes this species is that the males don’t make hostile takeovers of groups of females--at least not directly. Instead outside males make harassing raids on a troop, chasing the infants and sometimes killing them. One by one the females abandon their male, the childless females first, the others as soon as their babies are weaned or die. Steenbeek suspects that the infanticidal male langurs are trying to discredit the harem male, demonstrating how incompetent he is by killing the troop’s infants. The females are continually judging the contest, and if they sense that their male is getting weak, they abandon him. Sometimes at the end of the tenure, a male stops protecting the troop, says Steenbeek. It’s like he’s just given up. If infanticide has long been a natural part of animal behavior, then so too, one would expect, has been the fight against it. Over evolutionary time, both currents would shape new behaviors and social organizations. Lionesses live in prides, according to Packer and Pusey, in large part to protect their young against murderous males. One result is that lionesses must tolerate cubs not their own stealing milk--something rarely seen in other carnivores. Female mice can somehow tell if a male approaching their litter is infanticidal or not; if he is, they leap into battle. And apparently even the babies have evolved a protection against infanticide: they call much more frequently in the presence of an infanticidal male. Evidence is emerging that primates may face similar pressure. Female red howler monkeys in Venezuela, for example, tend to travel in small groups--generally under five members--and are hostile to new female howlers who want to join them. What determines their group size? In some animals the availability of food is the key: if the group is too big, the competition among individuals grows too intense. Yet evidently food competition isn’t a problem for red howlers. What is a problem, it seems, is that the bigger a female group, the more likely it is to suffer an infanticidal attack by a male. The benefits for a male of taking over a big group make those groups good targets, and as a result females keep the groups small. Carel van Schaik, a primatologist with Duke University and the Wildlife Conservation Society, thinks infanticide’s effects may reach even further into the core of primate life. He first began thinking seriously about Hrdy’s ideas in the late 1980s, while studying gibbons with Robin Dunbar, who was then at University College, London. These Asian monkeys are for the most part monogamous, although it’s hard to see why. It’s not that gibbon babies need the extra parental care, because the fathers don’t give any. And calculations suggested that males might do better, from a genetic viewpoint, if they tried to impregnate as many females as possible rather than just one. Van Schaik and Dunbar concluded that male gibbons were staying close to home to guard their infants from other males, and the females were choosing good protectors as mates. That would explain why on the one hand gibbon couples make calls together--advertising to neighboring males that the infant is well guarded--but why on the other hand a nursing mother who becomes widowed falls silent. It doesn’t matter to Van Schaik that gibbon infanticide has never been reported--it’s not seen, he thinks, because the animals do such a good job of preventing it. Van Schaik now suspects that the ever-present threat of infanticide has a similar effect on all primates. Among mammals, primate males and females are far and away the most likely to form a long-term bond. That raises the issue: Why primates? says Van Schaik. The answer, he thinks, is that primate babies are particularly vulnerable to infanticide. They take a long time to mature, and compared with other young mammals, they are defenseless and exposed, more often than not clinging to their mother. Female primates also tend to stay in a given territory, thereby giving males an added incentive for disposing of unrelated infants. If you do commit infanticide, there is a good probability that you will have a future opportunity for mating, says Van Schaik. Thus the incentive and opportunity for infanticide have driven primates more than other mammals into long-term bonds, in order that males can defend their young. If Van Schaik is right, he will add considerable weight to speculations Hrdy made in the 1980s, that protection against infanticide may have had a profound impact on primates, including early humans. Nursing is a contraceptive among humans, as it is in langurs, and it can lower a woman’s fertility for up to two or three years. That could make the incentive for infanticide on the part of a new mate enormous, as would the incentives to guard against it. Such a scenario would fly in the face of the conventional view that long-term bonds between men and women evolved so that extra parental care can help their babies survive. Instead, Van Schaik suggests that infanticide may have been the prime mover behind these bonds, and only later did the added advantage of help from a father come into play. Not surprisingly, perhaps, such ideas are not easily accepted. Most anthropologists and psychologists still view humans much as biologists once viewed animals. Anthropology has a long history of believing that everything is for the good of the social group, says anthropologist Kim Hill of the University of New Mexico, and in that context Hrdy’s ideas about infanticide don’t make sense; such instances as do occur among humans can only be explained as a result of a particular cultural bias (favoring male babies over females, say) or of individual pathology. But there are a few disturbing data points. Over the past 16 years Martin Daly and Margo Wilson, both psychologists at McMaster University in Hamilton, Ontario, have collected child abuse data from governments and humane associations. One of the most startling statistics they’ve uncovered is this: a preschool American stepchild is 60 times more likely than a biological child to be the victim of infanticide. Hill himself, with his studies of the Ache people of Paraguay, has gathered some of the best infanticide data available on non-Western cultures. The Ache still go on long hunting-and-foraging expeditions, as their ancestors did for 10,000 years. When a man kills an animal, he gives it to another man, who then distributes the meat to the entire band. Congenial as this may seem, natural selection creates inevitable tension: by giving most of his food away, a man allows his efforts to be diverted from his own family. This cost is outweighed by the benefits of cooperation, but when a child’s father dies, the tension reveals itself. If a child loses a father, his chances of becoming the victim of infanticide at the hands of another man increase fourfold. It’s not uncommon for orphaned children to be thrown into their father’s grave. But Hill does not think that the pattern is purely a cultural tradition. If you ask them why they’re killing all these babies, their first answer is ‘That’s our custom,’ says Hill. And then if you push them on that, they say, ‘They don’t have parents, and we have to take care of them, and that makes us mad.’ The resistance to infanticide as a reproductive strategy is still shared by many researchers. In some cases they’ve tested some of the predictions and found them wanting. Agustin Fuentes of the University of California at Berkeley, for example, studies the Mentawai langur, which lives on the islands of the same name, off the west coast of Sumatra. Like the gibbon, the Mentawai langur is monogamous, but it doesn’t behave as Van Schaik said it should. For example, when a bonded couple are close to a solitary (and supposedly infanticidal) male, they do not become hostile or even make calls to show they are together. Deborah Overdorff of the University of Texas at Austin studies rufous lemurs, and among these primates, at least, doubts the reality of infanticide. While rufous lemurs travel in large groups, male and female pairs will often be seen staying close together. That might seem to fit the notion of males protecting their young. Not to Overdorff. I’ve found that the male is not necessarily the one the females mated with. Sometimes they turn out to be brothers. Infanticide is probably not a good explanation for pair-bonding. Others criticize the quality of the data. They complain that most reports are inferred from indirect evidence, such as the disappearance of a baby. And except for Hanuman langurs, the few witnessed infanticides have not been followed up to see how much reproductive success the killing males have had. Given the difficulty of observing monkeys in the wild, the scrappiness of the data shouldn’t be surprising, and some primatologists-- including some who think that infanticide is real--worry that the theory is getting too far ahead of the data. The biggest opposition results from the application of Hrdy’s ideas to humans. Popular accounts of the theory, Hrdy complains, are very quick to jump from the langur case to cases of strange-male-in-the- household infanticide, but the underpinnings, the groundwork for that extrapolation, aren’t there. After all, a stepfather can’t speed up his own reproduction through infanticide. Hrdy and Daly agree that this kind of abuse has more to do with resource competition--the resource being the mother. Moreover, they don’t envision a stepfather consciously trying to eliminate that competition--rather, he may simply have a lower threshold of irritation toward the child. Such a threshold is suggested by a recent study by Daly and Wilson, in which they compared the ways in which biological fathers and stepfathers killed their children. In most cases, biological fathers shoot or suffocate their offspring (and then often kill themselves), while stepfathers kill by striking--hinting that a lashing out reflex is at work. Another point of criticism is the matter of how infanticide can be carried down through the generations, and again confusion abounds. One magazine article Hrdy mentions, for example, contains a reference to an infanticide gene. She scoffs. I don’t talk about genes. While it’s true that Hrdy doesn’t dabble in oversimplified genetic determinism, some of sociobiology’s early pioneers did--and sometimes with great abandon. These days, however, a much suppler view exists. In any species, each individual keeps a Darwinian account book, and whenever it has to choose an action, it weighs the immediate and long-term costs and benefits. Selection hasn’t molded an animal that’s altruistic or infanticidal, says Emlen. It has molded an animal capable of showing a whole range of subtly different behaviors under different circumstances, but they’re all predictable. Under one set of circumstances, a female might behave by abandoning a baby, but under another set of circumstances she would care for a baby, says Hrdy. These are both maternal behaviors. In the first case, presumably selection has operated on her to postpone raising her young because there is the option that she might have a better chance of pulling a baby through at a future date. So it’s not nonreproductive, it’s not nonadaptive; it’s simply a question of an animal over the course of a lifetime gauging herself. Infanticide is thus at one extreme in a spectrum of parental care. Hrdy herself has recently been exploring the ways in which European parents have historically lowered their investment in children, such as hiring a wet nurse or leaving a child at a monastery. For animals, and to some degree ourselves, this gauging happens unconsciously. And while it might seem hard to believe that animals can make careful decisions, many experiments have revealed that they can. Few of these accounting experiments have been done on infanticide, though, and there isn’t anywhere near enough data to test in primates, let alone humans. While almost 20 percent of Ache children fall victim to infanticide before age ten, the rate is zero among many other foraging cultures. Until researchers can explain the variation, all speculation on the role of infanticide in early human evolution must be put on hold. Our long, complex social lives and our dizzying array of cultures hide the effects of evolution as the high, obscuring leaves of the Sumatran forest hide the secrets of Thomas langurs. Yet those who believe that infanticide is a Darwinian reality think that we need to keep looking through the foliage. Sure, you can deny all these results--at your own peril, says Van Schaik. What is it that makes males infanticidal, and what is it that stops males from being infanticidal? If you know these things better, you know what to do, take certain measures, counsel people. It arms us. A look back at the infanticide hypothesis on its silver anniversary makes clear how long it takes to test and flesh out the shocking ideas of sociobiology. Hrdy herself sees this as the necessary pace of any science. She often describes her job as creating imaginary worlds that other scientists can then explore to see if they can help us understand the real one. I see scientists working in different phases. Some people are better at one phase than another. Theoreticians think of other people as technicians; technicians think of theoreticians as people in outer space, not connected to the real world. But for the whole process, you need these phases, and in the initial phase, you’re selecting a project, you’re coming up with assumptions, you’re trying to model what might be true and to generate the hypotheses that you want to look at. Then you have the actual collection of data and all the methodologies that go into that. Imaginary worlds have a place in science.