Joshua D. Duntley
Although there are more victims of aggression than perpetrators, violent humans have received the majority of interest from researchers and the public. This chapter explores the evolutionary logic and evidence of adaptations that defend against dangers from humans.
What constitutes a danger from other humans depends on the unit of analysis. Intuitively, people tend to focus on the individual person as the unit of analysis when considering survival dangers. It is the individual who can be punched, stabbed, raped, or killed. From this view, dangers from other humans are limited to behaviors that threaten a person's survival and physical health.
Using the gene as the unit of analysis requires the consideration of a greater range of dangers from humans. The survival of genes is not limited to a single lifespan or a single individual. Genes can simultaneously exist in multiple, different individuals and continue to exist across thousands of generations. For genes, contributions to genetic fitness are the relevant indicators of success. Any behaviors from other humans that put genes at a replicative disadvantage can be considered a danger. A physical attack on one's child, being cheated on by a spouse, or being humiliated by a rival are all behaviors that endanger the replicative success of genes. But none are dangers to the focal individual's survival or physical health. That human psychology reacts to these and other dangers to genetic fitness as something akin to life-and-death threats to personal well-being provide a clue that our psychology was shaped, ultimately, to ensure the replicative success of our genes. Genetic cuckoldry, reputational damage, and the loss of reproductive value of a child are life-and-death situations for genes. Although the survival of the individual is not sufficient to ensure genetic fitness, it is necessary. This chapter focuses on dangers to this necessary element of genetic fitness: dangers to the survival and physical health of the individual that other humans can pose.
There is no question that humans are dangerous. Among animals, humans are second only to mosquitoes in the number of people they kill each year (GatesNotes, 2014). Of the 5.8 million deaths from injuries that occur globally each year, about 1 in 7 result from homicide or warfare (World Health Organization [WHO], 2008). Among 15- to 29-year-olds, who experience among the highest levels of competition for reproductively relevant resources (Buss, 2003), homicide is the fourth leading cause of death, after traffic injuries, HIV/AIDS, and tuberculosis (WHO, 2008). Warfare is the ninth leading cause of death for this group. Across all ages, almost twice as many men than women die from injuries and violence.
For some ancestral contexts, such as big-game hunting, cooperation would have produced the greatest inclusive fitness benefit for individual hunters, given that their individual contributions to the hunt were proportional to the benefits they received. For other ancestral contexts, however, a behavior that increased the inclusive fitness of one individual would have simultaneously decreased the fitness of another, creating conflict between the parties involved (Buss, 1989). Given adequate fitness benefits to power the engine of selection, any trait can be favored, including those that make humans dangerous to one another.
Many of the human activities that make humans the most dangerous have been proposed to be the result of psychological adaptations. Evidence suggests psychological adaptations have a role in producing spousal violence (Buss & Shackelford, 1997a), aggression (Buss & Shackelford, 1997b; Campbell, 1993; Daly & Wilson, 1988; Wilson, Daly, & Pound, 2002), rape (Thornhill & Palmer, 2000), homicide (Buss, 2005; Duntley & Buss, 2011), and warfare (Tooby & Cosmides, 2010; 1988) (see Pinker, 2011, more broadly, for a comprehensive review of the historical trajectory of these and other dangers). At the core of the selection pressures that shaped these adaptations is conflict and competition for limited resources and social relationships.
Violence can offer a potential solution to a wider variety of adaptive problems than other behaviors can. For example, consider the use of two strategies to obtain resources: violence and clandestine theft. While theft can be effective at getting resources, violence can be used as a strategy to simultaneously aid in theft, demonstrate physical prowess to potential mates, and intimidate rivals.
Although perpetrators of cost-inflicting strategies can gain much through their behavior, victims can incur costs ranging from strategic interference with evolved goals to death. On average, death is the most costly outcome that victims face. It is implausible that selection would not have acted to prevent or stanch the costs of victimization.
Antagonistic coevolutionary arms races are part of the evolutionary history of most species, if not all of them. They can occur between species, as with predators and their prey, or between competitors within species. They can create massive selection pressures, capable of producing rapid evolutionary change (see Phillips, Brown, & Shine, 2004). Any recurrent context of conflict between conspecifics has the potential to be a hotbed for the coevolution of competing strategies to best a competitor and defend against being bested.
The evolution of adaptations to inflict costs creates selection pressures for the coevolution of counteradaptations to defend against them. The amount of selection pressure is a function of the magnitude and frequency of the costs over evolutionary time. The evolution of adaptations to defend against incurring costs creates new selection pressure for refinements of adaptations designed to inflict costs or new adaptations for that purpose. These refined adaptations for cost-infliction, in turn, create new selection pressure for refinements in adaptations to defend against costs. New combinations of shifting adaptations and counteradaptations can lead to antagonistic coevolutionary arms races between adaptations to inflict costs and adaptations to defend against them that go on perpetually, unless the source of the conflict and competition is resolved or otherwise eliminated.
The existence of adaptations that are designed to counter the cost-inflicting strategies of competitors is a source of evidence that the competitor's strategy is the product of adaptations. Counteradaptations to a given competitor's strategy can evolve only when the strategy has been sufficiently recurrent in predictable contexts over evolutionary time. Adaptations are more likely than by-products of adaptations or noise to produce evolutionarily recurrent, contextually predictable behaviors. Moreover, defensive counteradaptations may function by making a competitor's cost-inflicting behavior too costly to perform (e.g., killing a sexual aggressor), which would create selection pressure against the cost-inflicting strategy. A cost-inflicting strategy that continues to persist over evolutionary time despite the costs suggests that it may, on average, be functional in producing a net benefit in a particular context. Evidence of such functionality is necessary evidence of adaptation, but is not sufficient. Additional evidence could be sought in the complexity and specificity of the design features of evolved defenses and their match to design features of the hypothesized adaptation. A complex set of highly functional design features in mechanisms that produce cost-infliction and correspondingly complex and specific defenses against them would strengthen the case that both are adaptations.
There are important differences between the form and function of victim defenses that depend on the timing of their activation. Victims can defend themselves against the costs inflicted by dangerous humans: (1) before the victimization occurs, (2) while the cost-inflicting event is occurring, or (3) after being victimized. The strength of selection pressures operating to shape adaptations to address each temporal context varies as a function of the nature of the costs inflicted. For example, there would be selection pressures on victim adaptations against rape in all three temporal contexts. Women should have adaptations to avoid victimization, to minimize costs during victimization, and to take steps to prevent future victimization in the aftermath of rape. However, there would not be selection pressures on all three of the temporal contexts of adaptations of people who are murdered. The primary victims of homicide are incapable of directly influencing events after their deaths.
The best defense against being victimized is to not become a victim. To the extent that strategies of cost infliction were perpetrated by predictable conspecifics in predictable contexts there would have been selection pressure for the evolution of defensive adaptations to avoid them. Individuals with psychological mechanisms that led them to recognize situations and competitors associated with a higher likelihood of incurring the costs and avoid them would have had a large fitness advantage over those who lacked such mechanisms. Fear while walking through dark alleys at night, of people who seem “shifty,” and stranger anxiety in infants are examples of the hypothesized outcomes of adaptations to prevent falling victim to the cost-inflicting strategies of others.
Selection also shaped adaptations to minimize the costs of victimization while it is occurring. Defensive postures, verbal attempts at manipulation, and seeking or creating opportunities to flee an attacker are defensive strategies hypothesized to have been selected because they decreased the costs of victimization. Curling into a fetal position may help to deflect the blows from an attacker away from a victim's head and internal organs. The use of language to activate sympathy or empathy in an attacker, or to frighten an attacker away, may be effective in decreasing the duration or severity of the cost infliction. Creating or waiting for an event that distracts an attacker, or temporarily incapacitating an attacker, might give victims an opportunity to escape or to hide and seek protection. Selection would have favored any adaptation that decreased the magnitude of costs that victims might otherwise have incurred.
Victim adaptations activated after the occurrence of the cost-inflicting event that function to minimize the impact of the victimization or to prevent future victimization also would have been favored by selection. For example, acting as though the injuries sustained during a fight are not as debilitating as they actually are, or verbal assaults on an attacker that impugn the effectiveness of the person's attack, such as “you punch like a 3-year-old,” may decrease the status loss associated with being beaten in a physical fight.
There are numerous avenues for the prevention of future occurrences of victimi-zation. One is learning cues to danger. By recognizing and subsequently avoiding dangerous contexts and individuals, victims will be less likely to incur costs from them in the future. A person victimized in a certain part of a city, for example, subsequently may be motivated to avoid that part of the city. A victim may avoid future interactions with an attacker. Victims also may be proactive in avoiding conflicts by fortifying defenses against future attacks by conspecifics. For example, carrying a weapon for self-defense may decrease the likelihood of incurring serious costs in future confrontations.
Another avenue for the prevention of future victimization is to retaliate against an attacker. Demonstrating an effective ability to retaliate may decrease the likelihood of future victimization by sending a message to the perpetrator and others that attacks or exploitation will be avenged. Revenge has been suggested to be built into our psychology by natural selection (Buss & Duntley, 2006). fMRI research has demonstrated that pleasure centers of men's brains become activated upon exacting revenge against someone they believe perpetrated a wrong against them (Singer et al., 2006). The research suggests that the motivation for men to seek revenge may have evolutionary underpinnings and supports the contention that maintaining status in social competition was important for the inclusive fitness of ancestral men.
Selection pressures for each temporal category of victim adaptations were unlikely to be equal. Because entirely avoiding being a victim was ancestrally associated with the lowest victimization costs, there probably was more selection pressure for the evolution of previctimization adaptations than for victim adaptations that function during or after victims have incurred costs. As a result, previctimization adaptations are hypothesized to be larger in number than the other temporal categories of victim defense adaptations.
Although it can be useful to conceptualize three distinct temporal categories of victim defenses for the purpose of exploring their design features, it is also possible that some defensive strategies bridge more than one temporal category, such as storing information about victimization in memory and using it to prevent becoming a victim in the future.
In sum, it is useful to consider three temporal categories of victim adaptations: those aimed at avoiding victimization, those that minimize the costs of victimization while it is occurring, and those that function after victimization to minimize its costs and to prevent its recurrence. The nature of the victimization will determine the degree of selection pressure for adaptations in each of these contexts.
A range of strategies may be employed to inflict costs on others, too many to adequately address in this chapter. Behaviors that inflict the greatest costs create the strongest selection pressure for defenses against them. To get the clearest picture of the coevolution of strategies of cost infliction and defenses against them, the next section focuses on three contexts associated with the greatest costs for its victims: Violence, rape, and homicide.
Physically injuring rivals clearly inflicts physiological costs on them. The use of violence can also be an effective competitive tactic. Healthy individuals can compete more effectively than the rivals they injure. Rivals may be more likely to avoid or drop out of competition with individuals who injured them in the past. Individuals who inflict greater injuries than they sustain in a conflict may gain a reputation of being difficult to exploit (Buss, 2011). This reputation may protect individuals against violent confrontations and grant them easier access to resources with less resistance from rivals.
An effective strategy for preventing violence is to avoid the violent confrontation altogether. Because it is easier to attack an individual than a group, human adaptations to form alliances may provide one form of deterrence against violent rivals (Tooby & Cosmides, 2010). Adaptations that lead to the avoidance of contexts likely to make an individual the target of violence may provide another kind of protection against being injured by a conspecific. Humans also may possess adaptations designed to attempt to reason with an attacker, arguing that the costs of the attacker's violent behavior outweigh the benefits, offering some other possible resolution to the conflict, or threatening to use violence to defend themselves. If an attack cannot be avoided, individuals may resort to aggression or even murder to defend against becoming a victim of violence (Daly & Wilson, 1988).
An analysis of the structure of the bones of men's faces suggests that they evolved for fighting, specifically to minimize injury caused by being punched by another man (Carrier & Morgan, 2014). The boney structures argued to provide protection from the blows of a rival provide a good example of physiological adaptations against violence.
Rape is a cost-inflicting strategy with a direct link to reproduction. Rapists may benefit by fathering offspring that they may not have otherwise produced. Rape inflicts not only emotional costs (Block, 1990; Burgess & Holmstrom, 1974; Jerin & Moriarty, 2010) and physical costs (Geist, 1988) on women, but also fitness costs by bypassing female choice in mates and the timing of reproduction (Buss, 2011; see also Perilloux, Duntley, & Buss, 2012 for a discussion of the range of costs). Although scholars have concluded that there is not enough evidence to determine whether men have adaptations to rape (Buss, 2003, 2011, 2014; Symons, 1979), ethnographies and historical records suggest that rape occurs cross-culturally and was recurrent over human evolutionary history (Buss, 2003, 2011; McKibbin & Shackelford, 2011).
Numerous researchers have proposed the existence of anti-rape adaptations. The formation of alliances with men and other women for protection has been argued to represent evolved counterstrategies to rapists' tactics (Smuts, 1992). The “bodyguard hypothesis” proposes that women's preference for mates who are physically formidable and high in social dominance is, in part, an adaptation to prevent rape (Wilson & Mesnick, 1997). As noted, Carrier & Morgan (2014) argue that the bony structures in men's faces are particularly well adapted for fistfights with other men. These bony structures, which include a larger and broader jawbone, thicker cheekbones, and more pronounced bones around the nose and eyes, are present only in men. Women lack any facial bones that approach the thickness and strength of those possessed by men. Research has demonstrated, however, that women exhibit a mate preference for men with more masculine faces, particularly when they are ovulating (Gildersleeve, Haselton, & Fales, 2014). Women's preference for masculinized male faces, particularly at the time that women are most fertile, could function to help women choose to associate with men best able to protect them from rapists. Women's evolved mate preference for masculinized facial features can be thought of as creating, through sexual selection, adaptations in male physiology that enable the men with whom they prefer to associate to better defend their female partners from attacks by other men.
Specialized fears that motivate women to avoid situations ancestrally predictive of an increased likelihood of being raped have been proposed to help preemptively defend against rape. To prevent conception resulting from rape, women may avoid risky activities during ovulation (Bröder & Hohmann, 2003; Chavanne & Gallup, 1998). The psychological pain of rape motivates women to be more vigilant in the future (Thornhill & Palmer, 2000). Women blame themselves for being victimized more than others blame them (Perilloux, Duntley, & Buss, 2014), perhaps creating a sense of personal control to alter behavioral patterns to better avoid future victimization. In addition, women may possess adaptations to minimize the costs of rape after it has occurred. To avoid the reputational damage that can be associated with rape or to avoid losing their romantic partner, women may feel motivated to keep their ordeal a secret. Women may feel a strong urge to bathe themselves after being sexually coerced, washing physical evidence of the forced encounter away so it cannot be detected, especially by their romantic partner. Women may seek revenge against their attacker by marshalling male relatives and allies to attack him, especially if the rapist represents a persistent threat to the women or their female relatives. Spontaneous abortion, premature delivery, and infanticide may also represent female defenses to avoid investing in a rapist's child (for a detailed review, see McKibbin & Shackelford, 2011).
Homicide is a strategy capable of solving or contributing to the solution of conflict with other individuals (Buss & Duntley, 1998, 1999, 2003, 2004, in progress; Duntley & Buss, 1998, 1999, 2000, 2001, 2002, 2011). According to homicide adaptation theory, homicide is unique from nonlethal solutions to conflict because it represents an absolute end to the competition between individuals. Living and dying are drastically different outcomes of behavior, large enough to have created selection pressure for the evolution of cognitive algorithms capable of guiding behavior toward either nonlethal or lethal outcomes. Addressing conflict with competitors with strategies that leave them alive allows them to create the same problems in the future that they did in the past. Once dead, a person can no longer damage reputations, steal resources, prevent others from attracting romantic partners, or poach others' mates. It would be astonishing if selection did not operate differently on cognitive algorithms that produce lethal and nonlethal outcomes.
The fundamental and profound difference in the outcomes of nonlethal and lethal behaviors leads to the hypothesis that homicide is the designed output of evolved psychological mechanisms. Inflicting a lethal injury on a rival is the evolved function of homicide adaptations. Killing conspecifics could have helped to solve a variety of ancestral problems (Duntley & Buss, 2008, 2011), including: (a) preventing the exploitation, injury, rape, or killing of self, kin, mates, and coalitional allies by conspecifics in the present and future; (b) reputation management against being perceived as easily exploited, injured, raped, or killed by conspecifics; (c) protecting resources, territory, shelter, and food from competitors; (d) eliminating resource-absorbing or costly individuals who are not genetically related (e.g., stepchildren); and (e) eliminating genetic relatives who interfere with investment in other vehicles better able to translate resource investment into genetic fitness (e.g., deformed infants, the chronically ill or infirm).
The infliction of an unrecoverable injury that slowly kills a victim through infection or other gradual decline in health can be just as effective as causing the victim's instant death, but is more subtle and may motivate less vengeance in the victim's kin and social allies. With the help of time, age, starvation, pathogens, parasites, and poor wound healing, killers could achieve the evolved goal of eliminating a rival while maintaining some plausible deniability about their intentions to kill.
Conspecific killing was a recurrent feature of human evolutionary history (Chagnon, 1988; Keeley, 1996; Trinkaus & Shipman, 1993). A victim's death has a much larger impact on his or her inclusive fitness than just the loss of the genes housed in the person's body. The inclusive fitness costs of dying at the hands of another human can cascade to the victim's children, spouse, and kin. The specific costs include:
A homicide victim's fitness losses can become rivals' fitness gains. Killers can benefit from the residual reproductive value and parenting value of the surviving mate of their victim, sometimes at the expense of the victim's children with that mate. A killer can ascend into the vacancy in a status hierarchy left by his victim. The children of killers would thrive relative to the children of homicide victims, who would be deprived of the investment, protection, and influence of a genetic parent. Many family members who would have survived if the person was not killed will die before they can reproduce. Many children who would have been born in the family will never be born.
The magnitude of rivals' fitness gains will be heavily dependent on group size and the presence of formidable rivals. In smaller groups and with fewer rivals present, a slight local increase in resources or mates, following a murder, can bring a substantial benefit to the murderer. In larger groups and when more rivals are present, however, the fitness benefits could be diluted because the newly available resources could be harder to control.
One hypothesized design feature of homicide avoidance mechanisms is sensitivity to high-risk contexts. Cues to the presence of such contexts include:
Perhaps the most effective defense against being killed is to completely avoid situations associated with an increased risk of being a victim of homicide. The experience of fear may be one adaptive mechanism that helps us to avoid such situations. In his book The Gift of Fear (1997), Gavin De Becker argues that fear can function as a signal that exists to aid in our survival, protecting us from violent situations. It is adaptive to experience fear, he argues, when the fear is enabling—allowing an individual to effectively address the danger he or she faces. Real fear, according to De Becker, “occurs in the presence of danger and will always easily link to pain or death” (p. 285).
Marks (1987) argued that fear and anxiety can be protective in four primary ways. First, it can lead a person to become immobile, which could conceal an individual from a predator or hostile conspecific, allow for assessment of the situation, and perhaps avoid being attacked. This is a valuable strategy when there is uncertainty about whether one has been spotted or cannot determine the exact location of the threat. Second, fear can motivate an individual to escape or to avoid danger in the environment. Third, a person may adopt a strategy of aggression in self-defense. Finally, an individual can adopt of strategy of submission to appease the source of the hostility, a common tactic among social mammals, including humans (Buss, 2014).
Because homicide has unique fitness consequences, the fear of being killed may be a distinct emotional state accompanied by specific decision rules that function to help individuals defend themselves. Rather than consisting of a single, consistent emotional experience, fear of being killed is proposed to be expressed in a range of discreet states. As a victim defense, a variety of fears may be experienced, which include, for example: mild anxiety about groups of unknown strangers in the distance; terror that motivates curling into a fetal position if an attacker has knocked one down and is kicking one in the head; battle-numbness that allows one to ignore moderate injuries if there is still imminent danger from an attacking horde; and a specific aversion to sharp incoming projectile weapons that likely would cause hemorrhaging or infections.
It is interesting that people in modern environments so willingly expose themselves to experiences that they evolved to fear. More than half of the programs at the top of Nielsen Ratings in a typical week (when the NFL playoffs are not occurring) are homicide dramas or documentaries. Murder mystery novels, monster movies, TV crime series, haunted houses, and Halloween masks all activate victim psychology. Why people, especially teenagers and young adults, voluntarily subject themselves to seemingly aversive stimuli may involve the calibration and practice of victim defenses.
Another protection against homicide is defending against the attacks of others. These strategies can take three primary forms:
Some of these strategies may be implemented up to the moment that a homicidal behavior is enacted upon a victim. The implementation of these defensive strategies may not always be enough to derail a homicidal strategy in favor of a nonlethal alternative. If not, the person targeted by a killer would have no recourse but to violently defend against attempts at lethal aggression.
Physically incapacitating a killer is another strategy a victim can use in self-defense. Victims can fight back themselves or enlist canine allies. Some research suggests that one of the functions of our ancestors' domestication of dogs was to act as watchdogs and bodyguards against hostile conspecifics (Clutton-Brock, 1999; Shipman, 2010). Invariably, an incapacitation strategy involves physically attacking the would-be killer. At a minimum, the intended victim must incapacitate the attacker enough to flee or buy enough time for help to arrive. Sometimes, the most practical strategy may be to kill the killer in self-defense. Killing in self-defense is likely to be influenced by contextual features such as: a lack of kin or allies in close proximity to help; the failure of nonlethal strategies to incapacitate the attacker or otherwise derail the progression of his lethal behavior; and a lack of other options.
One of the key differences between a would-be killer and victim in hostile confrontations is that the killer is more often prepared to carry out a homicidal strategy than the victim is to defend against being killed. The killer can select the time and place when it is best to kill. Selection would have favored adaptive design that led killers to catch victims alone and by surprise, reducing the possible costs of killing (e.g., being injured or killed by a victim or the victim's kin). Because the genetic relatives of a homicide victim suffer fitness costs, adaptations to defend against being killed should be also found in victims' kin.
At least two forces may have selected for adaptations in kin that function to stanch the negative consequences of a family member being killed. First, damage to a homicide victim's family reputation may be repaired by inflicting reciprocal costs on the killer. A family that is capable of striking back against the killer may be able to demonstrate that it is no longer exploitable. Second, the killer may be a persistent threat if he were to continue to live. Avenging the death of a family member by killing the killer may eliminate a source of recurrent fitness costs.
Homicide defense adaptations are costly for killers. The evolution of adaptations to defend against being killed would have created selection pressures for the evolution of refined adaptations for homicide that were capable of circumventing the evolved defenses. The presence of refined homicide adaptations, in turn, would have selected for further refinements to homicide defenses, and so on, setting up an antagonistic coevolutionary arms race between adaptations to kill and adaptations to defend against being killed.
Several sources of evidence suggest that mechanisms dedicated to conspecific killing could have evolved. The first source of evidence is comparative. In some insect and arachnid species, where mate-killing and cannibalism is known to increase the number or viability of offspring (including mantids, black widow spiders, and scorpions), males cautiously approach females to mate and then retreat quickly. During copulation, males of sexually cannibalistic species use diverse strategies to decrease their chances of being cannibalized (Elgar & Crespi, 1992): Male scorpions sometimes sting the female after deposition of the spermatophore (Polis & Farley, 1979); male black widows (Gould, 1984) and crab spiders (Bristowe, 1958) often restrain females in silk prior to copulation. Conspecific killing, as well as mechanisms to prevent getting killed, appear to be common among insects and arachnids.
Among the roughly 5,400 species of mammals, many also have well-documented patterns of conspecific killing. Male tigers, lions, wolves, hyenas, cougars, and cheetahs have been observed to kill the infants of rival males (Ghiglieri, 1999), hastening the estrus of the mothers, which often mate with the killers. Among primate species, conspecific killings have been well documented among langur monkeys (Hrdy, 1977), chacma baboons (Busse & Hamilton, 1981), red howler monkeys (Crockett & Sekulic, 1984), savanna baboons (Collins, Busse, & Goodall, 1984), mountain gorillas (Fossey, 1984), chimpanzees (Bygott, 1972; Suzuki, 1971), blue monkeys (Butynski, 1982), and others (Hausfater & Hrdy, 1984). The killing of conspecific rival males has also been well-documented among chimpanzees (Wilson et al., 2014) and mountain gorillas (Fossey, 1984). If conspecific killing was favored by selection in other animals, it could have been favored in humans as well.
Homicide has the potential to occur wherever there are humans interacting with other humans. This is as true of mother and child as it is of enemy nations. It is even true of the relationship between a pregnant mother and her developing fetus. For a woman, the fetus she carries may not represent her last opportunity to reproduce. Women were selected to invest more in those offspring who will yield the greater reproductive benefit, even in utero. If a fetus is not viable, it would make more sense, in terms of fitness, for a pregnant woman to forgo her investment in its development in favor of investing in a subsequent pregnancy. Most fertilized eggs do not result in a full-term pregnancy. Up to 78% fail to implant or are spontaneously aborted (Nesse & Williams, 1994). Most often, these outcomes occur because the mother detects chromosomal or other developmental abnormalities in the fetus. The mother's ability to detect such abnormalities is the result of adaptations that function to prevent the mother from investing in offspring that will likely die young. Most miscarriages occur within the first 12 weeks after conception (Haig, 1993), at a point when the mother has not yet invested heavily in a costly pregnancy and a spontaneously aborted fetus is less likely to lead to infection (Saraiya et al., 1999). The fetus, however, is not passive in its mother's evolved reproductive strategy. The fetus has only one chance to live. The production and release of human chorionic gonadotropin (hCG) by the fetus into the mother's bloodstream, which is normally an honest signal of fetal viability, may be a fetal adaptation against being spontaneously aborted. This hormone prevents the mother from menstruating, allowing the fetus to remain implanted. Maternal physiology reacts to the production of hCG as a sign that the developing fetus is viable (Haig, 1993). After a child is born, other humans do not cease to be dangerous. Ample evidence can be found by examining child-killing by parents and parent-substitutes.
A newborn infant has few options for defending itself from homicidal attacks perpetrated by adults. To defend against maternal infanticide, a newborn's best strategy may be to display cues that it is a vehicle worthy of investment. Immediately after birth, an infant should display cues to its health and vigor, cues capable of satisfying maternal adaptations that evolved to judge the probability of fitness payoffs for investing in the infant (Soltis, 2004). Newborns who nurse in the first hour after birth stimulate a surge in maternal oxytocin levels, strengthening the bond between mother and newborn. Nursing mothers' priorities become shifted. They become less motivated to self-groom for the purposes of attracting a mate and more motivated to groom their infants (Insel, 1992). By contrast, new mothers who do not nurse are more likely to suffer from postpartum depression (Papinczak & Turner, 2000; Taveras et al., 2003), a condition associated with higher rates of maternal infanticide (Hagen, 1999; Knopps, 1993; Spinelli, 2004) and maternal thoughts of harming their babies (Jennings, Ross, Popper, & Elmore, 1999; Kendall-Tackett, 1994). More active newborns are less likely to die (Chong & Karlberg, 2004; Morales & Vazquez, 1994), and would be a wiser object of maternal investment than newborns that are less active. Selection may have favored early nursing, the production of loud cries, and robust movements in newborns as defenses against maternal infanticide.
As they develop, infants are increasingly able to move about on their own. As a result, they are increasingly likely to encounter dangers while outside the range of their caregivers' protection. Infants who possess some ability to recognize potential dangers in the environment would have a significant advantage over infants with no such ability. Selection would have favored fears of specific dangers, to steer infants away from threats to their survival. The developmental timing of the emergence of fears provides evidence that selection played a part in shaping them. For example, the fear of heights emerges when children begin to crawl, which corresponds with infants' greater risk of falling. Fear of strangers emerges at about the same time (Scarr & Salapatek, 1970), corresponding with a greater risk of encountering hostile conspecifics. Stranger anxiety prevents children from approaching those they do not know well and motivates them to seek parental protection. It has been documented in countries and cultures from Guatemala and Zambia, to the !Kung and the Hopi Indians (Smith, 1979). Infant deaths at the hands of unrelated conspecifics have been documented among nonhuman primates (Ghiglieri, 1999, Hrdy, 1977; Wrangham & Peterson, 1996) and in humans (Daly & Wilson, 1988; Hrdy, 1999). Human children are more fearful of men than of women strangers, which corresponds to the greater threat posed by unrelated males over evolutionary history (Heerwagen & Orians, 2002). If a fear of strangers prevented even a tiny fraction of children from being killed in the evolutionary past, stranger anxiety would have been favored by selection.
Strangers are not the only threat to the lives of children. With a stepparent in the home, children are between 40 and 100 times more likely to be killed than children raised by two genetic parents (Daly & Wilson, 1988). Stepfamilies were likely a recurrent feature of ancestral environments. Without modern medical treatments, disease killed many adults. Fathers sometimes died in battles or on hunts. Mothers sometimes died during childbirth. After their partner's death, it probably was common for a surviving parent to find a new mate. New long-term relationships bring benefits to single parents, but also carry the potential for great costs to their children. The increased risk of their existing children being killed may affect single parents' mate preferences or the decision of whether to seek a new mate at all. Single parents' preferences for new partners could reflect, in part, evolved defenses against the homicide of their existing children (Buss, 2005).
Stepchildren also may possess adaptations to help defend against potentially homicidal stepparents, including the ability to predict a stepparent's likelihood of being homicidal and inflicting other costs. Children's evolved intuitions about potential stepparents may lead them to influence their custodial parent's mate choice, decreasing the children's risk of being killed. Genetic parents may have done well to pay attention to their children's preferences: Bringing a preexisting child into a new long-term relationship is a predictor of intimate partner homicide as well (Campbell, Glass, Sharps, Laughon, & Bloom, 2007).
Selection also may have favored adaptations that lead stepchildren to minimize their costliness to their stepparent by keeping a low profile and demanding few resources. Stepchildren also should recognize opportunities to make themselves valuable to their stepparent, such as contributing to the care of half siblings that result from the relationship between their genetic parent and stepparent. A possible strategy for stepchildren who feel their life is in danger may be to sabotage their genetic parent's long-term relationship by inflicting costs on the stepparent or inflicting costs on themselves, which could drive the stepparent away or redirect their genetic parent's investment away from a new mateship to ensure the offspring's survival. Engaging in delinquent behaviors, self-mutilation, disordered eating, drug use, and suicide attempts may be strategies children use to redirect their genetic parent's investment. Living in a stepfamily compared to living with two genetic parents more than doubles a child's risk of engaging in juvenile delinquent behavior (Coughlin & Vuchinich, 1996; Zill, 1994).
The presence of a stepparent is a good example of a recurrent context of increased risk of homicide that may have selected for antihomicide defenses in stepchildren and their kin. These adaptations become activated in stepchildren, but remain dormant in children who reside with genetic parents. Specialized adaptations to defend against homicide are proposed to exist for all contextual domains where there was a recurrent risk of being killed. Many situations, however, do not provide complete information about the probability that a person may fall victim to homicide. Because being killed is so costly, selection may have fashioned adaptively patterned biases that lead people to systematically overestimate the likelihood that they will be killed in conditions of uncertainty.
Because many inferences about whether one will be targeted by a killer are clouded by uncertainty, contexts of homicide can be considered compatible with the logic of error management theory (Haselton, 2003; Haselton & Buss, 2000). In situations involving uncertainty, making an erroneous inference about the intentions of others can carry high fitness costs. In contexts ancestrally predictive of homicide, it would be better, on average, to infer that rivals might want to kill you when they really do not, than to infer that rivals do not want to kill you when they actually do. In this way, people would avoid making the more costly of the two errors. A hypothesized design feature of the psychology of homicide avoidance is a cognitive bias that leads people to overestimate homicidal intent in the presence of cues to adaptive problems historically solvable by homicide.
The amount of uncertainty surrounding a potentially high-cost situation is also likely to have an effect. In conditions of uncertainty about the identity of another person, in unclear social situations, and in the absence of information to the contrary, the safer error would be to overestimate a conspecific's hostile intentions. In fact, the safest error may be to assume that the other person intended to kill you. Selection would have favored decision rules that are quickly sensitive to potentially costly meetings with conspecifics. When facing uncertainty from environmental cues, selection should mold psychological design to assume that the worst possible fitness event is going to occur, so its heavy costs can be more effectively avoided. The strategies people employ to defend against homicide (e.g., avoiding the context, fleeing, or killing one's attacker) would simultaneously defend against a number of nonlethal, cost-inflicting strategies. As a result, strategies capable of defending against homicide also can help to protect an individual from a range of other dangerous situations.
In summary, uncertainty about the nature of situations, including uncertainty about the identity or history of an individual, provided selection pressures that influenced the design of human error management psychology. Adaptations to minimize costly errors evolved in the form of cognitive biases that overestimate the likelihood that another individual intends to inflict costs proportional to the uncertainty surrounding the individual and the context. The bias toward inferring that another individual intends to inflict costs should increase as uncertainty about the individual and the context increases. This is not to say that such an error management bias will be applied equally to different individuals. The bias should be proportional to the ancestral threat that different individuals posed. It should be especially strong for those who posed the greatest threat, such as young adult males, and less strong or absent for others (e.g., infants, young children, the elderly).
There is evidence that people's perceptions are biased in the direction predicted by error management theory (Haselton & Buss, 2001). Experiments using schematic facial stimuli demonstrate that different facial expressions are not processed the same way (Öhman, Lundqvist, & Esteves, 2001). Participants identified threatening faces more quickly than happy faces from among distracters. Additionally, faces with V-shaped eyebrows of a schematic angry facial display were identified more quickly and accurately than were faces with inverted V-shaped eyebrows (friendly faces), among distracters. These results are consistent with a perceptual bias predicted by error management theory that leads individuals to be especially sensitive to the presence of potentially hostile conspecifics. Natural selection would have favored a greater sensitivity to angry faces than to friendly faces, as those with hostile intentions would have posed an adaptive problem often requiring immediate action to avoid incurring potentially heavy costs, particularly from out-group members (Ackermann et al., 2006).
Despite sensitivity to dangerous humans, many people still enter into situations that could get them killed. People have extramarital affairs, derogate competitors, and poach the material resources and mates of others. What makes them think that they can get away with their lives?
The answer may lie in the use of secrecy as a defense against being killed. People only become homicidal if they are aware that they are being wronged. Their ignorance can provide those who sneak behind their backs some measure of protection from being killed. A sexual relationship behind the back of one's partner, for example, could benefit men in the form of additional offspring and benefit women in the form of access to superior or different genes, and to additional resources from an affair partner (Greiling & Buss, 2000). Selection should have favored the use of secrecy to defend against the costs of an infidelity being discovered, which includes being killed by a jealous mate or rival. In the case of sexual infidelity, there is a clear pattern in the risks of being killed. Men are more likely than women to kill their partner for a sexual infidelity (Serran & Firestone, 2004; Wilson & Daly, 1992). As a result, selection pressures may have been stronger on women than on men to adopt clandestine tactics to conduct their affairs. This may help to explain why men indicate a greater amount of uncertainty about whether their romantic partner is having an affair than women do (Buss, 2000): Men encounter fewer cues to their partner's infidelity. Clandestine strategies, however, are not always successful. Sometimes men discover their partner's infidelity. As homicide statistics demonstrate (Buss, 2005; Daly & Wilson, 1988; Ghiglieri, 1999), perhaps the most dangerous human a woman will encounter in her lifetime is her romantic partner.
In a review of 223 appellate opinions of the cases of battered women who killed their male partners in Pennsylvania, 75% of the homicides occurred while the woman was being assaulted by her romantic partner (Maguigan, 1991). In a study of mate homicides in North Carolina between 1991 and 1993, violence perpetrated by men preceded 75% of cases in which women killed their romantic partners. In contrast, there is no evidence that violence perpetrated by women preceded any of the homicides committed by men (Smith, Moracco, & Butts, 1998). It can be argued that the majority of women who kill their romantic partners do so in self-defense or to protect their children or other kin (Serran & Firestone, 2004). Female-perpetrated mate homicide may be an example of the ultimate anti-homicide defense: killing an attacker before the attacker kills you.
The ancestral costs of being murdered were substantial enough to select for adaptations designed to eliminate the threat of homicidal conspecifics by killing them. Killing someone to prevent them from killing you would have had distinct evolutionary advantages over strategies of nonlethal violence. By killing a homicidal conspecific, you eliminate any future threat the person may pose. Whereas an injured rival can recuperate and attempt to kill you again, a dead rival cannot. By killing the person who would kill you, one also demonstrates a willingness and ability to end a life, sending a powerful signal to others that attempts on your life will be met with the ultimate cost.
Most legal systems do not treat homicides committed in self-defense the same as other homicides. The law considers killing in self-defense to be a form of justifiable homicide if the killer “reasonably believes that killing is a necessary response to a physical attack that is likely to cause serious injury or death” (Costanzo, 2004, p. 83). In the evolutionary history of adaptations to produce preemptive homicides, however, the management of errors in conditions of uncertainty would have played a pivotal role in determining what a person reasonably believes. Individuals in the past who erred on the side of preemptively killing those perceived to be a credible threat to their lives or the lives of their kin would have had an advantage over others who erred in the opposite direction. The consequence of this overestimation is the preemptive killing of some individuals who would not have become killers. In the calculus of selection, however, it is better to be safe and alive than dead.
Adaptations for homicide need not be involved in the production of all homicidal behavior. When not agnostic about whether some adaptations function to produce homicide, Daly and Wilson (1988, 1990; see also Daly, Chapter 26, this Handbook, Volume 2) have argued that homicides may be the by-products of psychological mechanisms favored by selection for their nonlethal outcomes. For example, adaptations that produce nonlethal spousal violence to prevent the defection of a mate could overreact and mistakenly generate levels of violent behavior that result in a spousal homicide. Following this logic, the death of a child from neglect could be the accidental by-product of a failure in the activation or engagement of adaptations responsible for the production of parental solicitude.
Homicide adaptation theory does not preclude the possibility that some homicides are by-products of the activation of other mechanisms or simply mistakes (Duntley & Buss, 2011). However, there is disagreement about whether evidence supports the theory that the majority of homicides are the designed output of psychological adaptations (Buss, 2005; Duntley & Buss, 2008). One source of evidence is the conscious thoughts that people report having about killing others (e.g., Kenrick & Sheets, 1993). Daly (Chapter 26, this Handbook, Volume 2) dismisses evidence from studies of homicidal fantasies, homicidal intent, and willingness to kill because people experience fantasies of video game playing more often than fantasies of killing and formulate plans to do many things that were not targets of selection, such as watching TV.
First, homicide adaptation theory does not purport to be a general explanation for all conscious thoughts or intentions; it is not intended to provide an explanation for the frequency of fantasies about videogame playing or about people's intentions to do things that were not targets of selection. (Although, it is interesting that first person shooter video games were the top seller in 7 of the 10 past years (TheCHIVE, 2014), and that 6 of the 10 most popular TV shows involve murder mysteries (TV Guide, 2014).) Instead, Duntley and Buss (2011) propose that some design features of the psychology that produces lethal violence can be explored through examining people's thoughts of killing, a strategy no different than using sexual fantasies to better understand psychological adaptations that influence sexual behavior (Ellis & Symons, 1990). Second, the relative frequency or duration of thoughts about any topic would not seem to provide conclusive evidence about whether those thoughts are the functional outputs of adaptations. People think about sex less frequently than food, sleep, personal hygiene, social contact, time off, coffee, watching TV, checking email, and using other social media (Hofmann, Vohs, & Baumeister, 2012). But it would be difficult to argue that thoughts of sex are not products of adaptations because they occur less frequently than thoughts about other topics, some of which do not have obvious adaptive significance.
There is also disagreement about what source of evidence of the fitness impact of being a killer should be used to evaluate homicide adaptation theory. Daly (Chapter 26, this Handbook, Volume 2) argues that homicide does not promote individual fitness in the human groups that he has observed. However, it is difficult to draw conclusions about the fitness outcomes of homicide based on observations of a limited number of individuals for a few generations. Selection operates on genetic variability in populations over thousands and thousands of generations. Over deep time, a trait providing as little as a 1% average fitness advantage can be favored by selection (Nilsson & Pelger, 1994). Rather than relying on a single source of evidence to evaluate the fitness outcomes of killing conspecifics, additional sources should be considered (Duntley & Buss, 2008, 2011).
For example, evidence that tracks the historical transmission of Y-chromosome variants suggests that homicide was fitness promoting. One study found that as many as 0.5% of the world's total population could be descendants of Genghis Khan. Roughly 16 million men living in the former Mongol empire are argued to carry Khan's Y-chromosome (Zerjal et al., 2003). Khan's reproductive dominance was the result of his crushing military might, which resulted in the killings of thousands of his same-sex rivals, putting the reproductive value of the women of those he vanquished under his control.
Although there likely is no single source of evidence that clearly favors Homicide Adaptation Theory over the by-product hypothesis of homicide, the total weight of growing evidence supports the view that the function of some of our psychological adaptations is to produce behavior that is lethal to conspecifics (Duntley & Buss, 2008, 2011). Whether homicides are the functional output of adaptations or not, lethal aggression was a powerful selective force over human evolutionary history.
The evolution of adaptations to inflict costs created selection pressures for the coevolution of adaptations in victims to help them avoid or prevent incurring the costs. These coevolved victim adaptations, in turn, created selection pressures for the evolution of refined and new adaptations for cost-infliction, setting up antagonistic, coevolutionary arms races between strategies to inflict costs and strategies to defend against them. Coevolutionary arms races can be extremely powerful. They can exert selection pressures on numerous physiological and psychological systems simultaneously, leading to rapid evolutionary change and great complexity of adaptive design. Adaptations for homicide and adaptations to defend against homicide are argued to be results of just such an antagonistic coevolutionary arms race. The costs of being killed are among the greatest an individual can incur at the hands of a conspecific. These tremendous costs created unique and powerful selection pressures for the evolution of adaptations to defend against being killed.