Debra Lieberman and Jan Antfolk
It takes, in short,…a mind debauched by learning to carry the process of making the natural seem strange, so far as to ask for the why of any instinctive human act. To the metaphysician alone can such questions occur as: Why do we smile, when pleased, and not scowl? Why are we unable to talk to a crowd as we talk to a single friend? Why does a particular maiden turn our wits so upside-down? The common man can only say, “Of course we smile, of course our heart palpitates at the sight of the crowd, of course we love the maiden, that beautiful soul clad in that perfect form, so palpably and flagrantly made from all eternity to be loved!”
—William James, Principles of Psychology, 1891
A full causal account of a set of phenomena should explain what is absent, and therefore not observed.
—John Tooby, 1989, p. 14
Psychologists interested in sexual attraction and mate choice, have focused, to a large extent, on the features we find attractive in a mate and the circumstances that promote lust, attachment, and deep engagement. But what about the other side of the coin? What traits do we avoid when selecting a mate? Not as much research has been conducted in this area. One reason could be that we suffer from instinct blindness (see James, 1891). So good are the mechanisms that steer us clear of certain individuals sexually that we rarely recognize the absence of close genetic relatives, the very young, and the very old from discussions regarding mate selection. As an illustration of how family members are often overlooked in the mating literature, consider the three factors social psychologists suggest dictate who one will choose as a sexual partner: familiarity, similarity, and proximity (e.g., Berscheid & Walster, 1978). But who best fits this description? Family members! They are familiar—you have known them your entire life. They are similar—you share the same religion, the same culture, and, in the case they are your blood relatives, a strong physical resemblance. Last, they are close by and easily accessible—perhaps even under the same roof and down the hall. Nevertheless, nuclear family members are, typically, the last group of individuals considered as appropriate sexual partners.
Why is this? Intuitively, the answer to this question is that the thought of having sex with close kin is disgusting and repugnant. As James (1891) might say, of course we aren't attracted to close genetic relatives! But why do most people across diverse cultures feel this way rather than perceiving sexual behavior with a close family member as exciting and erotic? One answer is that humans and many other species evolved robust inbreeding avoidance systems that function to identify close genetic relatives and then disqualify them, depending on circumstance, from the list of potential mates. Because relatives typically do not appear on the sexual radar, when researchers think about sexual attraction, kin are commonly not even mentioned.
In this chapter, we address why such powerful inbreeding avoidance mechanisms evolved in humans as well as in other species. We provide a description of what an inbreeding avoidance system might look like in terms of information-processing structure and discuss recent research aimed at uncovering the systems that perform this function. To start, we begin with the critical background condition, sexual reproduction, which created the selection pressures favoring inbreeding avoidance.
There have been a number of profound theoretical questions evolutionary biologists have tackled: How did life evolve? How did eukaryotes evolve from prokaryotes? Why did multicellularity evolve? How can altruism evolve?
One question, which eluded researchers until recently, is why did sexual reproduction evolve? As a system of replication, sex seems strange. Specifically, there are a number of biological costs associated with sexual reproduction that could be avoided if organisms reproduced asexually (Maynard Smith, 1978). First, only half of an organism's genes get passed on with sex. Contrast this with asexual reproduction, in which a genetic clone is produced each generation. Second, for sexual reproduction to take place, specialized internal mechanisms and organs are required, all of which take up energetically costly tissue. Third, there are costs in terms of the time and energy required to search for a potential mate, say nothing of the strategic systems (e.g., psychological, chemical, or structural) required to convince another individual to engage in reproductive activities. As became evident during early forays into a theoretical account for why sex evolved (e.g., see Ridley, 1993), any explanation for the evolution of sexual reproduction required an account for how the benefits of sex outweigh these costs of meiosis, recombination, and mating, respectively.
As Ridley (1993) discusses in The Red Queen, there were many candidate explanations for why sex evolved, including to aid in the evolution of the species (e.g., Crow & Kimura, 1965), to repair or edit the genome (e.g., Bernstein, Byerly, Hopf, & Michod, 1985; Muller, 1964), and to generate variability among offspring to increase fitness upon dispersal to novel environments or when remaining in saturated environments (e.g., Bell, 1982; Williams, 1975). However, these early candidate explanations for why sex evolved failed on theoretical grounds and/or on the ability to explain the ecological distribution of sexual versus asexual species.
In the 1980s a number of researchers provided an answer to the question why sex (Bremermann, 1980; Hamilton, 1980; Tooby, 1982; and, more recently, Morran, Schmidt, Gelarden, Parrish, & Lively, 2011). Sex thwarts the transmission of pathogens from parent to offspring and further interferes with pathogen adaptation, and hence replication (Tooby, 1982). Long-lived multicellular organisms live in a sea of pathogens—on the skin, in the body, in food, and in the air. For an idea of how prevalent potential pathogens are, it has been estimated that there are approximately 106 bacteria in one mL of ocean water (Whitman, Coleman, & Wiebe, 1998).
Pathogens can exert intense selection pressures on hosts with comparatively longer rates of reproduction. The greater the difference in the rate of replication between host and pathogen, the more deleterious the effect of pathogens can be. In long-lived multicellular species, like humans, this difference can be quite large allowing for fast adaptation of the pathogen to the host. “Micro-organisms can go through as many generations in a week as humanity has gone through since the Neolithic revolution” (Tooby, 1989, pp. 108–109).
Not surprisingly, when they have a constant background against which to evolve, pathogens can wreak havoc on their hosts. Consider a human that reproduced asexually: A mass of cells fissions from the body and develops into a clone. Pathogens that got transmitted to the clonal offspring would have another entire generation to become even better adapted to its internal biochemical environment—evolving better strategies for obtaining host resources and evading destruction by the immune system. Each generation, pathogens would become more deleterious to that clonal lineage. If, instead of cloning, one recombined genomes with another organism likely to possess different alleles, this would create a novel internal biochemistry and place pathogens back at square one, having to re-solve the problems of resource acquisition and immune evasion. Pathogens' old keys to locks present in the parent would not work as well (if at all) to unlock the new locks present in the offspring. In general, then, sexual reproduction interferes with the process of pathogen adaptation.
With sexual reproduction came selection pressures regarding the choice of suitable mates. Recombining genomes with an individual who does not possess the same genetic “parts” (e.g., an individual of another species) would render offspring unviable (Tooby & Cosmides, 1990). On the other hand, recombining genomes with someone genetically identical defeats the purpose of sex. On the continuum of genetic similarity among humans, it would have been critical to avoid individuals sharing similar alleles, especially those governing immune defense as they govern the battlefront between pathogens and host. One class of individuals with an increased probability of sharing similar genes is kin—individuals who share genes by virtue of common descent. The closer the genetic relationship to another person is, the greater the probability of sharing similar alleles. Thus, evolution is expected to have led to systems that reduced the probability of selecting a close genetic relative as a sexual partner.
In addition to pathogens, a second selection pressure leading to the evolution of inbreeding avoidance mechanisms was the presence of deleterious recessive mutations. To understand why recessive mutations played an important role in the evolution of inbreeding avoidance mechanisms, it is first necessary to understand how the human genome is organized (see Lewin, 1999, for a review).
Humans are a diploid species, which means that they possess two parallel, homologous sets of chromosomes. One set is inherited from the mother, and the other is inherited from the father. The gene at each location (locus) along a given chromosome can be matched up to a corresponding or homologous gene on the chromosome inherited from the other parent. As a result, each individual possesses two copies of each gene (with the exception of genes located on the sex chromosomes, and extranuclear genes).
Functional genes at a given locus typically provide the sequence information required to build one of the tens of thousands of different proteins necessary for the structure, development, health, and activity of the organism. The two corresponding genes at the same locus can be identical in their DNA sequence, or they can have different forms. These alternative forms of the same gene are called alleles. When the alleles inherited from the maternal and paternal lineage are the same, they are called homozygous, and when dissimilar, they are called heterozygous. When two different alleles are present, it is often the case that the product of one allele masks the phenotypic expression of the other. The allele whose phenotype is expressed is said to be dominant, whereas the allele whose phenotypic expression is masked is considered recessive.
Various biological processes and entropic forces continually interject random mutations into the genome, usually transforming functional alleles into damaged or deleterious alleles. Errors can be made during DNA replication, and background radiation, heat, chemical agents, and other environmental factors can also cause changes. Mutations come in a variety of types (point mutations, frame shifts, deletions, etc., see Lewin, 1999 for a taxonomy of mutations that occur in the human genome). These mutations can disrupt a gene-product's ability to function properly. For example, a DNA replication error may lead to a mutation in an allele coding for an enzyme necessary for the neutralization of commonly encountered dietary toxin. Depending on the exact base changes caused by the mutation, the enzyme may, for example, (a) not be affected at all and, therefore, function properly, (b) have a slight change in the charge or shape of the binding site leading to a reduction in function, or (c) not function at all. If the enzyme is not produced, or no longer functions adequately, this can lead to harmful or even lethal consequences for the bearer. These negative mutations accumulate in the population until the rate at which they enter matches the rate at which they are expressed and selected out. The point at which entry matches exit is called equilibrium. Lethal dominant genes are always expressed, and so they are selected out rapidly after entering the population, staying at very low frequencies at equilibrium. As such, they play no special role in selecting against inbreeding.
In contrast, when a detrimental mutation is recessive, it has a much less harmful effect whenever it is matched with its undamaged dominant counterpart. Such a heterozygous individual expresses a normal phenotype, and her or his fitness is uninfluenced by the presence of the unexpressed injurious mutation. For this reason, deleterious recessives can accumulate until they reach relatively high frequencies in the population. The same negative trait that, if it were dominant, would stabilize at a frequency of roughly 1 in 1 million would approach a frequency of 1 in 1,000 if it were recessive—that is, 1,000 times more frequent. Indeed, it is only when the same recessive damaged allele is supplied from both the mother and the father, creating a homozygous individual, that the damaging trait is expressed, impeding the survival and reproduction of that individual.
Selection only acts against deleterious recessives when they are expressed and, according to Bittles and Neel (1994) “all of us are thought to carry in the heterozygous condition ‘several’ rare recessive genes which, if rendered homozygous, would result in a significant medical handicap, ranging from severe defects of vision and hearing to disorders incompatible with survival beyond childhood” (p. 17). The estimated number of rare lethal genes in a genome is termed lethal equivalents (Cavalli-Sforza & Bodmer, 1971; Crow & Kimura, 1970). Data from a number of studies suggest that each of us possess, on average, somewhere between two (Bittles & Neel, 1994; Carter, 1967; May, 1979) and six (Kumar, Pai, & Swaminathan, 1967) lethal equivalents: alleles that, if homozygous, would cause death before an individual reached reproductive age (Burnham, 1975; Morton, Crow, & Muller, 1956). We are not dead many times over because at the great majority of these loci, we are heterozygous, and the intact gene masks the damaged gene.
What influences the probability that the same deleterious recessive will be supplied from both the mother and the father? If the two parents are unrelated, then these recessives come together by chance. For example, if you have a lethal (or otherwise detrimental) recessive allele at one locus (Aa) and the recessive allele (a) exists in the population at a frequency of 1 in 1000 and you have a child with a random nonrelative, this child has a 1 in 4000 chance of being homozygous (aa) for this particular harmful trait. This is because it is a 1/1000 × 0.25 chance the child has two copies of the recessive allele (it is a 0.75 chance the child inherited at least one dominant allele). In contrast, mating with close kin increases the likelihood that two rare recessive alleles will meet each other at any given loci (Cavalli-Sforza & Bodmer, 1971; Charlesworth & Charlesworth, 1987; Tooby, 1977). Being genetically related means that the two individuals share common ancestors, and so the same deleterious recessives that show up in one relative are also likely to show up in others descended from the same common ancestor. What is a 1 in 4,000 risk in mating with a nonrelative becomes, when mating with a brother or sister, 1 in 8. If you have a hidden recessive allele at one locus, assuming that only one of your parents had this recessive allele (as described earlier, given no prior inbreeding in your family, there is only a 1 in 1,000 chance that both your parents had the allele), the chance is 0.50 that your sibling also has a nonexpressed copy of that allele. The chance for your child to be a homozygote expressing this detrimental recessive allele is then 1 in 8 (0.50 × 0.25). This means, that in this example, the risk of expressing a lethal recessive allele increases no less than 500 times when comparing a sibling union to a union between nonrelatives.
Therefore, if two close genetic relatives mate with one another, there is a greatly increased chance that the resulting offspring will be homozygous for many deleterious recessives, leading to decreased chance of survival and reproduction. The more closely related the parents, the greater the likelihood that the offspring will suffer a decrease in health and viability, and the selection pressures become very intense whenever the two parents are siblings, or parent and child. For this reason, deleterious recessive mutations posed a strong selection pressure against close-kin matings.
In summary, there were at least two recurring selection pressures that would have strongly selected against inbreeding among our hominid ancestors: (1) an increased susceptibility to disease-causing organisms, and (2) the generation of defects through making deleterious recessive genes homozygous. The cost in terms of damage to the offspring resulting from matings between close genetic relatives is called inbreeding depression (Wright, 1921). These two selection pressures would have selected for design features that reliably and cost-effectively caused a reduction in the probability of mating and conceiving with close, fertile relatives. Those individuals who carried such design features would have produced offspring more likely to survive, reproduce, and pass on those design features than individuals who did not.
Evidence from nonhuman species and humans alike illustrates that inbreeding leads to an increased risk of infection and mortality (nonhuman evidence: Acevedo-Whitehouse, Gulland, Greig, & Amos, 2003; Coltman, Pilkington, & Pemberton, 1999; human evidence: Adams & Neel, 1967; Bittles & Neel, 1994; Carter, 1967; Schull & Neel, 1965; Seemanova, 1971). Inbreeding leads to an increased probability of the expression of recessive deleterious genes leading to a greater incidence of major congenital malformations and postnatal mortality (Bittles, Mason, Green, & Rao, 1991). Many studies in humans have focused on offspring of first cousins since this form of marriage is quite common in many cultures around the world (Bittles, 2005). Though the effects of inbreeding depression in offspring of first cousins (r = 0.125) are expected to be much less severe than in offspring of individuals related at an r = 0.5 (parents, offspring, and siblings) there have, nevertheless, been reports of various deformities and deficiencies. Across a variety of populations, compared to unrelated parents, parents that were first cousins produced offspring with twice the population baseline probability of congenital malformation and/or genetic diseases (Norway: Stoltenberg, Magnus, Lie, Daltveit, & Irgens, 1997; Turkey: Demirel, Katlanoglu, Acar, Bodur, & Padak, 1997; Israel: Jaber, Merlob, Bu, Rotter, & Shohat, 1992; Pakistan: Hussain, 1998). In addition to increased probabilities of mortality and congenital malformations and diseases, children of first cousins have been shown to have cognitive impairments (Bashi, 1977, Cohen, Block, Flum, Kadar, & Goldschmist, 1963; Schull & Neel, 1965).
Studies focusing on the effects of inbreeding between siblings have found substantially increased risks when compared with first-cousin matings. Compared to inbreeding depression rates of 2%–6% in offspring of first cousins (compared to population baseline), it has been estimated that sibling matings lead to an inbreeding depression of 45% (Aoki & Feldman 1997; Ralls, Ballou, & Templeton, 1988; Seemanova, 1971). Moreover, since spontaneous abortion—a likely consequence of expressed detrimental alleles or decreased immunological functioning in the embryo/fetus—may go undetected, the effects of consanguineous marriages may be significantly underestimated (Bittles et al., 1991). There have been a handful of studies documenting the fitness consequences of offspring born of two siblings. In all studies, there was an increased risk of mortality, mental deficiencies, congenital malformations, and disease (Adams & Neel, 1967; Carter, 1967; Schull & Neel, 1965; Seemanova, 1971).
Perhaps one of the best studies on the effects of inbreeding depression was by Seemanova (1971) on a Czech population of women who had children fathered by both a close genetic relative (i.e., father or brother) and an unrelated male. The children of nonincestuous matings provided a perfect control group to investigate the deleterious effects of inbreeding between close genetic relatives. Considering only those females who were of normal intelligence (N = 44), 92 offspring were produced with an unrelated father and 50 offspring were produced with the female's father or brother. Of the nonincestuous children, 5.2% died within the first 5 months and 3.4% of the surviving children had impairments or deformities. There were no mental deficiencies found in these children. In comparison, of the incestuous children, 12% died with major deformities, and 45.4% of the surviving offspring were either severely mentally retarded, had major congenital deformities or impairments, or both. So, compared to nonincestuous offspring, the incestuous offspring were roughly 13 times more likely to die or have severe birth defects, an increase similar to that found in a study by Adams & Neel (1967) who looked at the consequences of brother/sister and father/daughter matings. In more recent studies, it has been found that offspring of closely related parents are smaller and weaker (Fareed & Afzal, 2014), suffer reduced fertility (Beer, Quebbeman, Ayers, & Haines, 1981; Schmiady & Neitzel, 2002; Thomas, Harger, Wagener, Rabin, & Gill, 1985), have cognitive impairments (Roberts, 1967; Rudan et al., 2002), and have autosomal recessive diseases, leading to hearing impairments (Zakzouk, 2002).
In summary, studies in humans and comparable nonhuman species have illustrated the deleterious consequences associated with mating with a close genetic relative. These recurring decrements in fitness would have selected for systems that enabled organisms to avoid mating with close genetic relatives. However, the costs of inbreeding are not identical for everyone, nor for the same person across time. Before describing what a system for avoiding inbreeding might look like in terms of information processing, we discuss additional design criteria regarding opportunity costs.
For long-lived species, such as humans, who interact with close genetic relatives throughout periods of sexual maturity, the recurring deleterious consequences of inbreeding as outlined earlier would have led to the evolution of psychological mechanisms for a sexual preference for nonkin. The strength of this preference, however, should depend on the different costs and benefits associated with the choice of one mate over another. To the degree that an individual forgoes an opportunity to mate with a nonrelative by engaging in inbreeding, inbreeding incurs an opportunity cost (e.g., Dawkins, 1983; Haig, 1999). Instead of having a healthy outbred offspring, inbreeding produces an offspring with a greater probability of inbreeding depression, δ. The decreased fitness in inbred offspring can be defined as x (x = 1 – δ). But δ is likely larger than zero but less than one, meaning that some, but not all inbred offspring fail to survive and propagate genetic material to future generations. Thus, depending on circumstance (e.g., pathogen load of the environment; available mating opportunities) inbreeding is a better option than not reproducing at all, but inbreeding is never as good as optimal outbreeding (Antfolk, 2014a).
In addition to inbreeding depression, one has to take into account opportunity costs. Because the number of offspring an individual can successfully produce and raise is limited, producing an inbred offspring and raising this offspring to nutritional independence will affect the possibility of producing and raising outbred offspring. Adding this opportunity cost (c) to the equation, inbreeding becomes costly when the opportunity costs outweigh the relatively decreased reproductive benefit of an inbred child, or (x – c < 0).
Opportunity costs are not equally distributed across groups, situations, and individuals. Opportunity costs are generally higher for females than for males. Because males invest less metabolic energy than females in the production of gametes, a male's reproductive success depends more on his ability to fertilize an egg than on his ability to produce gametes. Conversely, a female's reproductive success depends more on her ability to produce eggs than on her ability to get them fertilized (Bateman, 1948). This notion led Robert Trivers (1972) to establish a theory on how sex differences in the time and energy invested in reproduction leads to different reproductive strategies. Trivers defined parental investment (PI) as any investment an individual directs toward a particular offspring that precludes investment in other offspring. Investment thus includes everything from bestowing sex cells or sacrificing metabolic energy during copulation to risking one's life while guarding offspring. The critical factor is the minimum level of parental investment required to successfully produce an offspring capable of surviving and the minimum for men and women differs drastically. In humans, a male's minimum PI is the time and energy required for copulation, whereas a female's minimum PI is gestation (9–10 months) and would, in ancestral conditions, almost certainly include lactation (2–3 years). During these respective time periods, males and females cannot engage in alternative or additional sexual activities that would enhance their reproductive success. Males can inseminate one female at a time, and females don't ovulate during pregnancy and throughout periods of enduring on-demand breastfeeding. Given the much larger minimum level of investment by females as compared to males, females are expected to be more selective in their choice of a sexual partner and evidence suggests this is indeed the case (Clark & Hatfield, 1989).
Returning to inbreeding, because investing in one child decreases the possibility to invest in another child, and more so for females than males, the opportunity costs of inbreeding are higher for females than males (cF > cM). Although males suffer less direct costs from inbreeding than do females, the cost to a female relative needs to be accounted for when estimating the consequences inbreeding has to a male. This is because an individual's reproductive success is not limited to the number or the biological fitness of his or her own offspring. Rather, reproductive fitness is measured by the total number of allele copies that an individual transmits to subsequent generations, either through direct descendants or indirectly, through offspring of relatives. These relatives naturally include the relative with whom inbreeding takes place. Thus, the fitness consequences of inbreeding to the male can be expressed as (x – cM) + rMF(x – cF), where rMF is the coefficient of relatedness between the male and the female. From the point of view of the female, the fitness consequences can be modeled as (x – cF) + rFM(x – cM). In the case of brother-sister incest, r would be 0.5, meaning that half of the cost to the female is added to the direct cost to the male in the first example. In the second example, half of the cost to the male is added to the direct cost to the female (Antfolk, 2014a; Dawkins, 1983; Haig, 1999). The different levels of costs associated with inbreeding for men and women should be observable in how objectionable inbreeding is thought to be. Indeed, several studies show that human females react more strongly than males to the thought of having sex with close kin (Antfolk, Karlsson, Bäckström, & Santtila, 2012; Antfolk, Lieberman, & Santtila, 2012; Antfolk, Lindqvist, Albrecht, & Santtila, 2014; Lieberman, Tooby, & Cosmides, 2003).
Furthermore, the effect gender has on preferences to tolerate or object to inbreeding should be moderated by the probability of conception. For example, compared to the days in the menstrual cycle when fertility is low, women in the fertile period of the menstrual cycle have more to lose from sex with a relative. Indeed, women who are fertile have a stronger inbreeding aversion (Antfolk, Lieberman, Albrecht, & Santtila, 2014) and are less likely to associate with fathers, a behavior that can decrease the likelihood of inbreeding (Lieberman, Pillsworth, & Haselton, 2011).
Another variable that affects the opportunity costs of inbreeding is access to other mates. The capacity to acquire sexual access to females is unevenly distributed among males, with some males experiencing much greater reproductive success than others. Indeed, Bateman's early studies on sexual selection in fruit flies demonstrated that, whereas most females were successful in producing offspring, only 20% of males sired offspring for the next generation. The greater variation in male reproductive success should have led to adaptations that sensed one's potential success obtaining a high value mate (or a mate at all) and then activated the appropriate mating strategy. For males with high mate value and a pool of potential mates, inbreeding might not be the first strategy employed. But for low status males who detected they were unsuccessful on the mating market, widening the pool of potential mates to include sisters might be a strategy considered. Should a male with no other options engage in inbreeding, he would suffer only the inclusive costs incurred by his female relative, with whom he reproduces, while possibly gaining the benefit, albeit potentially low, of an inbred offspring. Therefore, one might expect males with few mating opportunities and males with no mating partner to be more inclined (or less resistant) to engage in inbreeding than other men. Indeed, a recent study found that individuals who are single (versus married), who have never had sex (versus have had sex), or have a low (versus high) mate value are more inclined to engage in inbreeding (Antfolk, Lieberman, et al., 2014). Moreover, this effect is more pronounced in men compared to women.
Taken together, humans modulate their inclination to engage in inbreeding depending on the perceived opportunity costs associated with having sex with a relative. For this to be possible, the inbreeding avoidance system must take as input variables such as fertility status, mate value, and opportunity costs.
To solve the problem of avoiding close genetic relatives as sexual partners, a system would need first to estimate the probability that another individual is a close genetic relative, and then, second, to inhibit sexual contact with that person as a function of this probability. But how do we detect kin? There are a number of constraints that confine the set of cues selection might have favored to engineer kin detection systems. For instance, barring recent medical technology, we are not able to directly compare genomes to assess kinship. However, other possible kinship cues exist. One possibility includes the use of more evolutionarily novel cultural information such as linguistic kin terms. But these are unlikely to be the primary cues used to detect kin because kin terms can blur genetic boundaries (e.g., aunt in our culture refers both to a parent's sister, a blood relative, and a parent's brother's wife, a nonblood relative). Furthermore, it is unlikely that phylogenetically prior kin detection mechanisms that functioned in the absence of linguistic information were overwritten by more variable and potentially less reliable cultural information.
Rather, it is likely we rely on ecologically valid cues that correlated with genetic relatedness in human ancestral environments. The stable patterns of genetic similarity created by events of sexual reproduction led to “categories” of individuals: mothers, fathers, offspring, siblings, aunts, uncles, nieces, nephews, and so forth. A system that identified recurring attributes unique to each category of “genetically similar other” and reduced the probability of selecting these individuals as sexual partners in a manner that reflected the costs of mating would have conferred a strong selective advantage. Importantly, the cues mediating kin detection might differ depending on the category of kin in question. To the extent that different cues signaled an individual was a specific type of close genetic relative (e.g., mother, father, offspring, or sibling), different detection mechanisms are likely to exist. Additionally, males and females might use distinct cues to identify the same type of kin. For instance, because men can never be fully certain of their relatedness to potential offspring, the cues signaling that an infant is indeed one's own are likely to differ for men and women. That is, ancestrally, a female was always certain (before the miracles of modern medicine) that the child coming out of her was indeed her own. The cue “birth” would have accurately identified offspring. But men don't give birth. Instead, men need to rely on information regarding the probability that the child of a particular female is indeed their own. If a man never had sex with a woman, the probability is zero. If a man did have sex with a woman, then issues such as timing of intercourse relative to birth and fidelity arise. To date, we do not know how men compute paternity and so this is a question ripe for research.
Evolutionary biologists have identified a range of kinship cues by investigating inbreeding avoidance and altruism in nonhuman species (for review, see Hepper 1991). For instance, early association, a spatial cue that identifies likely siblings in species in which offspring require extended maternal care, predicts patterns of social preferences and mate choice in species such as voles, mice, macaques, and chimps. In some species, chemical cues guide kin detection and associated kin-directed behaviors. Studies on house mice, for example, show that mate preferences are guided by assessments of similarity at loci controlling the major histocompatibility complex (MHC). That is, males and females prefer to mate with individuals who are MHC dissimilar from them, a preference thought to protect against the negative effects of pathogens. For MHC disassortative mating to occur, however, individuals require a referent, either themselves or a close relative, to determine what counts as MHC dissimilar. A series of cross-fostering experiments in which individuals were raised by MHC-dissimilar parents showed that individuals preferred to mate with others who were dissimilar from their foster parent's MHC composition. Thus MHC-guided mate preferences appear to use parental phenotypes as referents of one's own genetic composition (e.g., Penn & Potts, 1999). Of course, one should be careful in translating the MHC/HLA observations from nonhuman animal studies to humans. Few studies have used genotyping for humans so it is unclear whether MHC serves as a kinship cue for humans or functions more broadly to promote genetic diversity across kin and nonkin alike.
In the human evolutionary literature, the majority of research has focused on the detection of siblings and the associated development of sexual aversions and sibling-directed altruism. Next, we briefly discuss some recent findings from this literature.
The ancestral social environment of humans was such that a likely reliable cue to siblingship would have been seeing one's own mother caring for (e.g., breastfeeding) a newborn. Indeed, the intense mother-child association that typically occurs surrounding the natal period and continues throughout the first few years of life would have served as a stable anchor point for others to infer relatedness. Thus, if an individual observed an infant receiving care from the individual's own mother (at least the female categorized as one's own mother), then it was highly probable that that infant was the individual's sibling. Further, exposure to this cue would have signaled genetic relatedness regardless of coresidence (or association) duration. That is, regardless of whether one was 5, 10, or 15 years old, maternal-infant directed care would have cued probable genetic relatedness.
However, as potent a cue as mother-infant association might be, it is available only to older siblings already present in the social environment; the arrow of time prevents a younger sibling from having seen his or her older sibling born and cared for as an infant. For younger siblings, then, what cue or cues might evolution have used to identify probable older siblings?
One solution is to track the flow of parental effort. Any child regularly receiving care from one's own mother and father had a higher probability of being kin than children receiving care from other individuals. Moreover, the longer the care, the more likely the individual would have been a sibling. This cue, operationalized as childhood co-residence duration, was first proposed by Edward Westermarck, a Finnish social scientist who noted that children reared in close physical proximity during childhood tend to develop a sexual aversion toward one another later in adulthood (Westermarck, 1889/1891; see also Antfolk, 2014b). This idea, known as the Westermarck hypothesis, has received support from various anthropological and psychological investigations (see review in Lieberman et al., 2003). Perhaps most notable are the cases of the Israeli kibbutzim and Taiwanese minor marriages, two natural experiments inadvertently created by cultural institutions in which unrelated children were reared in close physical proximity throughout childhood. As the Westermarck hypothesis predicts, children reared together throughout childhood rarely marry one another (Israeli kibbutzim: Shepher, 1983), and if forced to marry suffer decreased rates of fertility and increased rates of divorce and extramarital affairs (Taiwanese minor marriages: Wolf, 1995). Together, these studies point to early coresidence as one cue our mind uses to assess relatedness and to dampen sexual desires.
However, they also raise many questions. For example, does coresidence duration predict sexual aversions differently for the younger and older sibling in a sib pair? As suggested earlier, older siblings might rely on a different cue to identify probable younger siblings, one that operates independent of coresidence duration. Also, do the same kinship cues that regulate inbreeding avoidance also regulate kin-directed altruism, the other suite of behaviors relying on assessments of relatedness?
It is not ethical to subject humans to the life-altering experiments used by evolutionary biologists to study kin recognition in nonhuman animals. For this reason, scientists have either had to look for natural experiments, such as those mentioned earlier, or take advantage of the natural variation that exists in families composed of actual genetic relatives. To investigate whether a proposed cue serves as a signal of relatedness, it is possible to match individual variation in exposure to the specific cue (e.g., coresidence duration, maternal-infant association) to behaviors and reactions relating to sexual behaviors with family members. Converging lines of evidence that we use a particular cue to categorize individuals according to genetic relatedness can be found through investigations of altruism. If the same kin detection mechanism serves to regulate both sexual avoidance and altruism, then a cue to kinship should show parallel effects across these two distinct motivational systems.
Using surveys to collect information, Lieberman, Tooby, and Cosmides (2007) found that individuals not exposed to their mother caring for their sibling as a newborn (as it is typically for the younger sibling in a sib pair), their duration of coresidence with an opposite sex sibling predicted aversions to sibling incest, as measured by disgust at imagining sex with one's own sibling and moral sentiments relating to third-party sibling incest. By contrast, for individuals exposed to their mother caring for their sibling as an infant (the older siblings in a sib pair), coresidence duration with an opposite sex sibling did not predict aversion to incest. When measures of altruistic attitudes and behavior were analyzed, the same pattern emerged. That is, coresidence duration with a sibling predicted altruism more strongly for individuals without access to the more potent cue of seeing their mother caring for their sibling as a newborn. These data provide compelling evidence that the mind uses two different cues for identifying older versus younger siblings and for regulating sexual aversions and altruistic motivations.
Upon inspecting the levels of aversions and altruistic inclinations reported, Lieberman et al. (2007) found that older siblings exposed to the cue of seeing their mother care for a younger sibling as a newborn reported intense levels of disgust toward sexual acts with that sibling as well as increased levels of altruism across all durations of coresidence. That is, regardless of whether a subject resided for 15 or only 3 years with their younger sibling, the level of sexual aversion reported in response to sibling incest and altruistic inclinations were close to the maximum. In contrast, for subjects for whom this cue was not available and who relied on coresidence duration as a cue to siblingship (the younger siblings in the dataset), disgust at sexual acts with their older sibling and sibling-directed altruism were low for shorter periods of coresidence and gradually increased with extended periods of coresidence. In fact, data suggest it takes approximately 14–15 years of coresidence for younger siblings to reach the same level of sexual aversions and altruistic effort reported by older siblings who were exposed to the cue of seeing their mother care for their sibling as a newborn.
Taken together, these data provide a first glimpse into the cognitive procedures governing kin detection and kin-directed behavior in humans. The findings indicate that the mind uses at least two cues to detect siblings and mediate inbreeding avoidance and kin-directed altruism: exposure to maternal investments in a newborn (used by older siblings to detect younger siblings) and duration of coresidence throughout periods of shared parental investment (typically used by younger siblings to detect older siblings). Because these same cues were found to regulate aversions and altruism in the same way, it suggests the existence of a single set of kinship-estimating procedures that feed motivational systems guiding mate choice and, separately, altruistic effort. Importantly, additional cues might also play a role in sibling detection. For example, facial resemblance and olfactory cues such as those derived from the major histocompatibility complex (MHC) predict preferences in mate choice (e.g., DeBruine, 2005; Wedekind & Füri, 1997). No matter the cue, if the model of kin detection and kin-directed behaviors outlined earlier is correct, a particular cue to kinship should regulate both sexual aversions and altruism.
Many questions remain unanswered. For example, if coresidence duration mediates sibling detection, is a specific period of coresidence (e.g., ages < 5) required, as some have suggested (Shepher, 1983; Wolf, 1995)? Or does each year of coresidence contribute in equal increments to a computed kinship estimate as research by Lieberman et al. (2007) suggests? Furthermore, what cues might distinguish a full biological sibling from maternal and paternal half siblings? Are the cues used to identify siblings the same as those used to identify parents and offspring? In the coming decades, research in evolutionary psychology can start to answer these and other questions.
Interestingly, humans also display an aversion toward others engaging in incest (Antfolk, Karlsson, et al., 2012; Antfolk, Lieberman, et al., 2012; Antfolk, Lindgvist, et al., 2014; Fessler & Navarrete, 2004; Lieberman et al., 2003; Lieberman et al., 2007; Royzman, Leeman, & Sabini, 2008). This reaction to third-party behavior has been explained as an adaptation, and also, as a by-product. Reactions to third-party inbreeding might merely reflect the activation of one's own inbreeding aversions. That is, when asked to think about an unknown brother-sister pair having sex, the strength of one's opposition could be a function of how opposed one would be to having sex with one's own sibling. There is evidence that this is indeed one source of third-party attitudes as the cues that predict disgust toward engaging in sex with one's own sibling also predict opposition toward third-party sibling inbreeding (Lieberman et al., 2007; Lieberman & Lobel, 2012). Thus, reactions to third party incest might be a by-product of the activation of personal inbreeding avoidance mechanisms.
But reactions to third-party inbreeding could also be an adaptation. The actions and decisions of close genetic relatives can greatly impact one's inclusive fitness. This holds for the decisions close genetic relatives make with respect to their mate choice. Not only are instances of inbreeding potentially costly to the individuals engaging in the act (see earlier discussion), they also impose large fitness costs on their close genetic relatives. Indeed, for each individual within the family, there is an expected inclusive fitness matrix of the costs and benefits associated with the different incestuous unions within the family. Consider an example that illustrates how an instance of inbreeding between a brother and sister not only bears fitness costs to the individuals engaged in the sexual liaison, but also to their sister. Although the sister of the sibling pair does not herself engage in inbreeding, she suffers inclusive-fitness costs by losing outbred nieces/nephews via her brother and via her sister. Extending the mathematical model discussed earlier, we can describe the fitness consequences to a related bystander (a common sister S). From the perspective of S, the fitness costs of inbreeding between a related male and female can be modeled as rSM(x – cM) + rSF(x – cF), where the costs to the sister and the costs to the brother are first weighted by the degree of relatedness of the sister to the male and the degrees of relatedness of the sister to the female respectively, and then summed.
To the extent that close genetic relatives could have pursued one another as a sexual partner in ancestral environments (an almost certain possibility, considering that humans lived in small groups), design features that motivated the interference of sexual unions between one's close relatives in a manner consistent with the costs and benefits associated with each particular incestuous dyad would have become more frequent in the population compared to design features that did not care whether family members mated with one another. That is, in addition to specialized systems for detecting close genetic relatives for the purpose of guiding one's own mating decisions, specialized systems are hypothesized to exist that assess the costs and benefits of other incestuous dyads within the family and motivate the active interference between those dyads imposing elevated costs to one's inclusive fitness (Lieberman, Tooby, & Cosmides, 2001). Intensity of interference should depend on a number of factors including mutation load, pathogen load, and available opportunities to secure unrelated mates (Tooby, 1977).
There is empirical evidence of a third-party aversion to related individuals engaging in incest mirroring the inclusive-fitness costs this behavior potentially has. Antfolk, Lieberman, and Santtila (2012) showed that regardless of a participant being described as involved in incest or not, incest between individuals related to a participant predicted the strength of aversion each incestuous situations elicited in the participant. Lieberman et al. (2001; unpublished data) showed that mate-child sexual behavior was the most objectionable followed, in order, by father-sister sex, son-daughter sex, mother-brother sex, and brother-sister sex. Certainly more work is needed in this area, but thus far, data suggest that humans possess cognitive adaptations for regulating the sexual behavior of family members. A system that used the computed kinship estimates that function to guide one's own sexual motivations to also assess the costs and benefits of particular unions within the family would be on its way to solving this adaptive problem.
There is one question that has been a topic of interest, particularly to sociologists and anthropologists (see Wolf & Durham, 2005): If there is an innate aversion that develops among close genetic relatives, then why do we need rules proscribing incest? There might be many reasons that norms regarding incest exist. Here are two: First, kinship cues are not always present. Siblings might be separated during childhood and, therefore, not observe one another being cared for by the same mother and father. Likewise, men might lack paternity certainty (e.g., because they question the fidelity of their mate or because they rarely witnessed their mate caring for her child) and, therefore, men will be more likely to find a daughter sexually attractive. These events, though likely rare, would have occurred from time to time leading, in some cases, to inbreeding. As discussed earlier, incest within the family carries costs to the inclusive fitness of other relatives. Thus, third parties might be motivated to prohibit certain sexual unions and this prohibition could manifest as cultural norms.
A second reason that norms proscribing incest might exist relates to the evolution of morality and social norms. DeScioli and Kurzban (2009, 2013) proposed a provocative account explaining why humans evolved adaptations to create and spread moral norms. To succinctly paraphrase, one adaptive problem humans faced was choosing sides when a conflict erupted. Costs are minimized to the extent that everyone chooses the same side to support in a conflict. But how does one decide which side to choose? Norms provide one solution: They solve the coordination problem a priori. If it is agreed upon beforehand that stealing is wrong and warrants punishment, then individuals can cost-effectively identify who they support. (Tooby & Cosmides, 2010, offer a slightly different view on the evolution of morality, one that focuses on the adaptive problems that arise once coalitions can form around a common goal, including the exploitation and extermination of less powerful others.)
But which norms to select? Regardless of whether the selection pressure leading to the evolution of social norms was signaling one's alliance in a particular social conflict or group exploitation, there is the problem of content. DeScioli and Kurzban (2009, 2013) suggest that strategically, it makes sense for an individual to support norms that do not impose on personal preferences. That is, to the extent that a moral norm prohibits behavior that one is not inclined to perform, then there is little cost associated with supporting such a norm. Disgust is an emotion that indicates the fitness value of performing particular behaviors relating to pathogen transmission and mate choice and thus might be a privileged emotion when making decisions about candidate moral norms. To the extent that one is disgusted by ingesting certain foods, physically contacting nonhumans or particular groups of humans, or having sex with particular people, these are the behaviors that, if prohibited, would not be felt as curtailing one's preferred course of action. So, one answer to why there exist norms proscribing incest is that inbreeding is a behavior that few people engage in (e.g., because of the fallibility of kinship cues or because of substance abuse, which could impair aversion systems). Given that the majority of people (or people in power capable of enforcing norms) find incest disgusting, proscribing incest is not felt as limiting personal preferences in terms of mate choice and thus would solve the problem of coordination, either for the purpose of choosing sides during a conflict as proposed by DeScioli and Kurzban (2009, 2013) or targeting groups vulnerable to exploitation as suggested by Tooby and Cosmides (2010). Certainly more research will need to be done to test between these alternate functional explanations.
There are many questions still left unanswered. Here we have raised a few of these questions about how humans detect their kin and the cognitive systems that perform the function of inbreeding avoidance. We conclude by emphasizing the need to take seriously the information-processing level of description when describing psychological adaptations. Evolutionary psychology has brought to the study of humans an ultimate level of explanation, that is, why certain attributes exist. In addition, researchers within evolutionary psychology have made much progress in documenting what phenotypes exist (e.g., behavior, preferences, etc.) and connecting them back to ultimate causal processes. In general, however, researchers in the field tend to overlook the intermediate level connecting why and what: how. Developing a cognitive model, no matter how rudimentary, can help clarify the function in question, identify potential moderators, and help direct programs of research. The information-processing model of kin detection discussed herein was a critical tool for thinking about how cues to kinship out in the real world could impact internal sexual preferences. This model also has helped to think about how kinship is integrated with the other attributes guiding mate choice (e.g., health, age, and status) and generated additional hypotheses about how kinship might be traded off against other attributes. Developing a cognitive model starting from inputs to internal computations to behavior can help identify the scope of one's research and help to overcome any instinct blindness that masks the existence and complexity of our evolved psychology.