Lawrence S. Sugiyama
If all humans of the same developmental stage and sex share the same attractiveness-assessment psychology, why don't they all find the same individuals attractive? How and why do our minds work to produce these effects? This chapter frames these questions in terms of a set of basic principles for understanding physical attractiveness: the Evolutionary Rules of Attraction. Research using evolutionary theory to understand human attractiveness is vast and growing, so not all can be covered herein. This chapter, therefore, (a) reviews alternative evolutionary explanations for an attraction; (b) highlights the general components of attraction systems; (c) identifies causes of variation in attractiveness assessment; (d) identifies domains of social value for which attractiveness assessment is relevant; (e) reviews evidence for some attractiveness-assessment adaptations in those domains; and (f) highlights research avenues calling for increased attention. In so doing, it updates my earlier argument (Sugiyama, 2005) that human physical attractiveness assessment is generated by adaptations functioning to evaluate evolutionarily relevant cues to human social value across multiple domains of interaction. It also extends my argument that evolutionary human life history theory and data from small-scale foraging societies are instrumental in generating predictions about these domains of social value, the cues or signals associated with them, adaptations selected to regulate attraction to them, and trade-offs predicted among them.
Objects are not intrinsically disgusting or attractive: These feelings are generated by cognitive adaptations. Humans are strongly attracted to some stimuli and repulsed by others, while remaining emotionally neutral to most. Attraction and disgust were relevant to many adaptive problems faced by our ancestors, such as what to eat, where to camp, whom to ally with, whom to mate with, and which juveniles to provision. Because the cues associated with, and behavior resulting in the fitness-promoting aspects of ancestral environments varied from task to task, different preference suites evolved for different tasks. No general attractiveness adaptation applies to all stimuli. A simple thought experiment illustrates why: If we assessed potential mates using the same attraction mechanisms with which we assessed food, we would find meat and fruits as sexually arousing as healthy, sexually mature members of our own species.
Attractions do not operate like an on-off switch: They are regulated in response to local evolutionarily relevant parameters. The physiological bases of an attraction include a network of complex biochemical and cellular information-processing systems designed to receive sensory information that identifies cues to the relevant attractiveness contexts. These systems activate the assessment processes, and direct attention to environmental cues to take as input. For each cue occurring above threshold, input must be encoded and its value computed. That information must be sent to mechanisms that integrate it with values of other cues relevant to the assessment, produce an emotional response scaled to the value of that cue integration, and then motivate specific behavior of a given intensity. These systems encompass receptors, neural organization and firing, neurotransmitters, and hormonal pathways and processing. Receptors, signal transmission, assessment processing, and emotional outputs are all components of attraction systems, and the same signal must produce different levels of attraction in different circumstances to function adaptively.
Although progress is being made (e.g., Roney, Simmons, & Lukaszewski, 2010), the genetic and neurophysiological bases of human interpersonal attraction adaptations remain largely to be explored, but taste mechanisms illustrate that attraction is based on coordinated gene action on cascades of biochemical and cellular processes. Humans are attracted to taste cues of “umami” or savory taste. Umami attraction begins with two closely related proteins, TAS1R1 (taste receptor type 1+1) and TAS1R3 (taste receptor type 1+3) that when chemically bound function to detect L-amino acids, particularly L-glutamate, an amino acid in meat and other foods such as mushrooms. TAS1R1 is encoded by the TAS1R1 gene, and TAS1R3 by the TAS1R3 gene. TAS1R1+3 receptors are located on the tip of complex biological structures—taste buds (which require other genes for other aspects of their production and function). When L-amino acid concentrations in the mouth reach a threshold, about 0.0007 M, signaling molecules trigger calcium release, activating melastatin 5 (TrpM5), which leads to membrane depolarization and release of the neurotransmitter ATP. Umami receptors don't have synapses. The ATP activates serotonin release from neighboring taste receptors that do, and these signals are transmitted via gustatory chorda tympani (and facial, glossopharyngeal, and vagus nerves) to multiple processing centers of the brain via additional structurally and functionally specific and physiologically complex biochemical and neurobiological processes. Resultant analyses of umami are then integrated with results of other taste system analysis (e.g., sweet, bitter) to generate different aspects of taste, including quality, intensity, pleasantness (or unpleasantness), location, and persistence. These, in turn, are integrated with visual, olfactory, and hunger regulatory mechanisms. The final output regulates attraction and motivates behavior.
A functional analysis provides an ultimate explanation for these adaptations. Amino acids are critical to many life functions (e.g., building muscle tissue, antibodies, enzymes; transportation of molecules such as hemoglobin). Umami taste signals the presence of amino acid L-glutamate, and attraction to such foods is rewarded by pleasurable taste, motivating their pursuit and consumption. This attraction evolved because it increased the probability of our ancestors consuming L-glutamate rich foods, increasing frequency of alleles generating the structures producing the preference (e.g., TASIR1+3 receptor arising with jawed vertebrates), and it was retained, built upon and fine-tuned by selection over time (Roper, 2007; Zhang et al., 2003). Conversely, revulsions discourage us from interacting with agents and substances in ways that were detrimental to fitness. For example, ingestion of moldy fruits can lead to intestinal distress from mycotoxins, and many plants have evolved biochemical defenses to keep predators from eating them (Jensen et al., 2013). Ingestion of these toxins can have high fitness costs, including death. Bitterness receptors respond to some of these chemotoxins and, as predicted for this threat-detection system, have a much lower activation threshold than sweet or umami receptors. The ultimate causal explanation expressed in cognitive terms (savory, taste, attraction, revulsion) is necessary for understanding these mechanisms, provides a useful explanation in its own right, and is consonant with the proximate genetic, biochemical, physiological level, and the functional explanation makes sense of the biochemical processes.
One possible explanation for an attraction then, is that it is the function of an adaptation. The most compelling case for adaptation is when one can show: (a) the species' ancestors recurrently faced a particular adaptive problem; (b) the structure has a complex functional design so improbably well-suited to solving the adaptive problem that pure chance must be rejected as an alternative hypothesis; and (c) the organism shares that design (or a facultative developmental program that builds that design) with all normally developing conspecifics.
However, we may also be attracted to objects exhibiting cues that were associated with a fitness-enhancing entity in ancestral environments but lack the fitness-enhancing properties themselves (Symons, 1987; Tooby & Cosmides, 1990, 1992). Consider soft drinks. Sweetness receptors initially evolved because carbohydrates in solution and sugar content provided were a statistically reliable cue of nutritious, energy-packed foods. The receptors were retained in humans because they generated attraction to foods such as fruit and honey. In modern environments, soft drinks exhibit similar cues, but do not provide the same fitness benefits (Eaton, Shostak, & Konner, 1988; Nesse & Williams, 1994). Sweet receptors and attraction respond (in some individuals) to aspartame and saccharine, even though attraction to these substances was not under selection. Similarly, pornography exhibits cues of willing, available, and fertile mates, but time spent viewing one's computer screen lacks the fitness-enhancing consequences for which attraction to these cues evolved.
Another possibility is that an evolved preference may bias responses to similar cues in other domains. Burley and colleagues accidentally discovered that finches have species-typical mate preferences with respect to the bands put on their legs by researchers. Female zebra finches prefer males with red rather than blue leg bands, whereas double-bar finches prefer light blue rather than red bands (Burley, 1986; Burley, Krantzberg, & Radman, 1982). Each species prefers colors similar to its own species' plumage, suggesting that these preferences are a by-product of species-recognition mechanisms. Similarly, some guppy species exhibit preferences for orange-colored foods, which may explain over 90% of the variation in female preferences for orange spots on males (Rodd, Hughes, Grether, & Baril, 2002). Humans may exhibit similar nonfunctional preferences, and complete understanding of human attractiveness will need to distinguish these.
Once a sensory bias evolves in a given domain, it may be subject to further selection in others. When subjected to selection for a different function, the result is an adaptation. In the natural world of mating, when sensory drive is present, it is likely to exert selection pressure on the target sex, so that preference and target trait become linked, and selection favors both trait and preference (e.g., Kokko, Brooks, Jennions, & Morley, 2003; Payne & Pagel, 2001). Distinguishing between an attraction due to sensory bias and sensory bias that has been under further selection requires fine-grained analysis. For instance, instead of a mate preference for facial symmetry being an adapation for attraction to high genetic or phenotypic quality, an alternate hypothesis is that symmetry preference is the result of sensory bias (Enquist & Johnstone, 1997). However, more symmetric faces are more attractive than less symmetric faces when right side up but not upside down, (Little & Jones, 2003, 2006). Greater symmetry preference is also associated with greater preference for sexually dimorphic facial features in the opposite sex, so sensory bias alone cannot account for facial symmetry preference (Little et al. 2008). Because there is no principled way to predict nonfunctional preferences, it is more useful to focus on adaptive problems for which attractions are expected to have evolved, and test by-product, sensory bias, pathology, or evolutionary mismatches as alternative explanations.
Finally, locally variable contingencies between cues we have evolved to assess and other associated cues may explain attractions that otherwise seem variable or idiosyncratic. This is because these associated cues may enhance local signal quality, or become emotionally attached to it. For instance, Damasio (1994) shows how an emotional response may link an adaptation's functional operation with stimuli recurrently associated with the stimuli for which an attraction evolved. If so, locally variable attractions may be the joint effect of attraction adaptations, association adaptations, and recurrent linkage between them. This is probably the basis for how many particular local cultural products become associated with stimulation of attraction responses, when such attractions per se could not be the direct products of selection. Of course, humans also employ artifacts to enhance those cues for which we have assessment adaptations (e.g., high heels, makeup, symmetrical face paint).
Naive cultural determinism does not offer an alternative scientific means of explaining attraction. The belief that “learning” or our “capacity for culture” accounts for attractions overlooks or grossly simplifies the psychological architecture requisite to generating preferences and other cultural phenomena (Tooby & Cosmides, 1992). On the cultural determinist view, who and what are found attractive varies arbitrarily across cultures: Individuals assess the physical attractiveness of both sexes based on local cultural dictates, and tend to prefer the sex that society tells them to. If this view were correct, standards of attractiveness would vary randomly across the cultural and geographic landscapes of human experience, but they do not (e.g., Sugiyama, 2005). Moreover, we would not expect to see behavior and attractions that are nonnormative within a culture, except perhaps as pathology. In 1950s American society, for example, we should find no straight parents with gay offspring, because the heteronormative determinants of straight culture should produce only straight individuals: This was not the case.
In sum, we have five basic options when formulating hypotheses regarding an attraction. It is either: (1) an adaptation; (2) a necessary by-product of an adaptation; (3) a nonfunctional effect of an adaptation that was not under selection to produce that effect per se; (4) the result of chance, pathology, or an idiosyncratic stochastic event; or (5) the product of an adaptation and statistical cue association, where the recurrent association of a local stimuli with an evolved preference cue binds the attraction to the local stimuli.
Adaptations for attraction and disgust are expected to be facultative within reaction norms. We expect them to be sensitive to the cost–benefit structure of their bearer's current circumstances and phenotypic condition, and also to the local conditions under which they develop. For social relationships, we expect these mechanisms to be sensitive to the kind of relationship (e.g., mating, trading, alloparental) for which an individual is being assessed. Evolutionary life history theory (LHT) points to age- and context-related trade-offs that were likely at play in the evolution of attractiveness-assessment mechanisms. Thus, with insights derived from the study of small-scale societies, LHT provides a framework for generating predictions about why and how the operation of attractiveness-assessment mechanisms varies under different circumstances.
Because success in the human ecological niche depends on multiple social relationships, the first prediction suggested by LHT is that attractiveness assessment will vary depending on the social value being assessed. A second source of variation in attractiveness assessments is the phenotypic state of the assessor, including the individual's sex, developmental stage, and reproductive state. For example, female assessments of male sexual attractiveness vary across the ovulatory cycle, and assessments of others' sexual attractiveness are affected by the assessor's relative sexual attractiveness as a short- or long-term mate. Individuals also vary in their socio-sexual orientation, or the degree to which they desire, approve of, and engage in committed versus uncommitted sexual relations, and this too can affect the relative attractiveness of different traits (Penke & Asendorpf, 2008; Simpson & Gangestad, 1992). Disentangling the effects of genetic variation on phenotypic state is critical here, as some kinds of variation are more relevant than others to explaining why different assessors find different individuals attractive. For example, the major histocompatibility complex (MHC), a gene complex involved in immune function may affect sexual “chemistry” between individuals.
Variation in reproductive and sexual strategies provides a third basis for variation in assessments of sexual attractiveness, and this same principle should apply to other strategies and social values. Consider cooperation: When there are physical cues associated with differences in cooperative strategies, individuals exhibiting different cues should be more or less attractive depending on the cooperative strategy of the assessor. For instance, an individual deploying a hawk strategy should find individuals exhibiting cues associated with a dove strategy more attractive as potential prey than individuals exhibiting cues associated with a hawk strategy.
Another source of variation in attractiveness assessments is local differences in the range of available variation in the cues being assessed (Sugiyama, 2005). This principle should hold regardless of the social value(s) being assessed, because attractiveness for any social value is relative to the available alternatives within a particular context. Research suggests that some of the available range of variation is based not only on local variation in the cues being assessed but on the phenotypic condition of the assessor.
In mating, another source of assessment variation is the range of competitors and the relative competitiveness of the assessor. The costs and benefits of different strategies are contingent on the assessor's phenotypic condition relative to competitors and the assessor's access to resources relative to competitors. Consequently, the mate value of a given individual will vary, in part, depending on the mate value of the individual making the assessment.
Human survival and reproduction are dependent on solving adaptive problems associated with multiple, partially overlapping spheres of social interaction, such as mates, offspring, kin, and allies. Individuals who were attracted to conspecifics exhibiting cues of high social value would have been more successful than those who were less discriminating. Our evolved attractiveness-assessment psychology is, therefore, expected to index social value across these domains. Human evolutionary life history provides the key to understanding these domains of human social value and the physical cues correlated with them.
Life history theory examines how and why natural selection produces age-related allocation of resources to different life functions both across and within species. Because resources are finite, there are trade-offs in the allocation of energy to different life functions (e.g., Charnov, 1993; Del Giudice, Gangestad, & Kaplan, Chapter 2, this volume; Stearns, 1992; Williams, 1966). Within a species' typical life history pattern, then, selection is predicted to produce adaptations that generated adaptively “strategic” trade-offs in energy allocation in response to evolutionarily relevant variables (e.g., Hill & Hurtado, 1996; Stearns, 1992; Trivers, 1972). These variables include extrinsic and intrinsic factors. Extrinsic factors include mortality and morbidity risk, the relative value of available resources, their spatial distribution, and costs of acquisition. Intrinsic factors include ego's sex, developmental and life-history stage, phenotypic condition, fertility, health, constraints, and options. Determining how individuals use local environmental cues to adjust their allocation of life resources is a main goal of understanding variation within a species' general life-history parameters (e.g., Hill & Hurtado, 1996).
Human life history includes a unique constellation of traits including: altricial birth; a long period of postweaning dependence; delayed reproduction; short interbirth intervals resulting in multiple dependent offspring; menopause; long postreproductive lifespan; allo-maternal investment; intra- and intergenerational resource transfers; facultative care of the sick and injured; and high levels of skill and knowledge acquisition, and social knowledge transmission (e.g., Flinn, Geary, & Ward, 2005; Hill, Barton, & Hurtado, 2009; Kaplan, Hill, Lancaster, & Hurtado, 2000; Leigh, 2001). Humans devote more of their energy to brain function, and have higher diet quality to support it (based on data from extant foragers), than expected of a primate our size. Conversely, we expend relatively less energy on muscle tissue, and more on fat reserves during infancy and early development, and to support fetal cognitive growth (Lassek & Gaulin, 2008; Leonard, Robertson, Snodgrass & Kuzawa, 2003). Food transfers, health care, and allo-maternal care help offset these costs.
At a general level, age-related trade-offs are expected between somatic (growth and maintenance) and reproductive (mating and parental) investment. However, adaptations regulating these investments are expected to be fine grained. For example, trade-offs documented among Shuar forager-horticulturalists include: quantity and quality of offspring (Blackwell, 2009); growth, body fat, and immune function (e.g., Blackwell, Pryor, Pozo, Tiwia, & Sugiyama, 2009; Blackwell, Snodgrass, Madimenos, & Sugiyama, 2010; Urlacher et al., 2014); branches of immune function (Blackwell et al., 2010), pregnancy, lactation, bone maintenance, and reformation (Madimenos, Snodgrass, Liebert, Cepon, & Sugiyama, 2012); pregnancy and lactation and mate's activity levels (Madimenos, Snodgrass, Blackwell, Liebert, & Sugiyama, 2011); helminth (intestinal worm) infection and disgust sensitivity, mediated by local costs of avoidance of fecal contamination (Cepon-Robins et al., 2013; Cepon-Robins et al., 2014); growth and childhood stress (Liebert et al., 2014). Trade-offs between Shuar childrens' growth and immune function are apparent during the week of illness and for a month thereafter, and are mediated by the child's body fat reserves (Urlacher et al., 2014). Tradeoffs between quantity and quality of offspring differ by sex and region, apparently due to differences in growth costs and age-related value of juvenile male and female labor across Shuar territory (Blackwell, 2009; Blackwell, Tiwia, et al., 2010). Further, different dimensions of growth are prioritized over others: Shuar have high prevalence of stunting (short height for age) but almost no wasting (low weight for height). These and studies in a variety of other populations point to complex temporal and socioecological regulation of life history trade-offs, and long-term dynamic and cumulative effects on phenotypic quality, reproduction, and health-related outcomes associated with fitness (e.g., Bribiescas, Ellison, & Gray, 2012; Ellison, 2003; Hill & Hurtado, 1996; Jasienska, 2009; Kramer, Greaves, & Ellison, 2009).
In short, adaptations regulate a vast number of trade-offs that ultimately affect how we look, smell, and sound, and these trade-offs vary based on a plethora of local and individual conditions. Phenotypic and behavioral results of these trade-offs are complex and varied, both across and within socioecological contexts. Documentation of trade-offs requires large sample sizes, repeat measures, and data on multiple variables, and presents a host of other methodological and theoretical challenges. Phenotypic correlation—where interindividual variation in physical condition or access to resources leads to positive or no correlation, instead of negative correlation, among traits that trade off (Stearns 1989)—is also an issue in interpreting study results. The fact that many aspects of attractiveness covary within individuals is at least in part a function of phenotypic correlation.
The past decade has seen a small but welcome increase in cross-cultural studies of attractiveness among nonwestern, nonindustrialized, nonstudent populations. However, cross-cultural testing has required a rethinking of hypotheses and methods and tempering of conclusions in light of variation in the socioecological context in which these adaptations play out (e.g., Schmitt, 2014; Scott, Clark, Boothroyd, & Penton-Voak, 2012; Scott et al., 2014). A life history approach predicts that, depending on the functional design of adaptations for attraction, local conditions will often produce variation in attraction to different levels of a trait and in the prioritization of traits used to make assessments. This is because different trade-offs are expected within and across cultures and mating strategies deployed. Thus, we expect variation in behavioral outputs (e.g., assessments of relative attractiveness) as functional products of evolved attractiveness-assessment mechanisms.
Conspecifics are of value to us in a range of social contexts that affect fitness. Individuals differ in their ability and willingness to provide benefits to us—and also in their ability to inflict costs on us—in each of these domains. For each domain of social value, then, some individuals are more valuable than others to the assessor. Accordingly, we would expect selection to have produced mechanisms that assess conspecifics in terms of the degree and kinds of social value they hold for us (e.g., Sugiyama, 2005).
For example, women face the problem of assessing a potential long-term partner's ability and willingness to invest in themselves and their offspring, which they must then weigh against the candidate's ability and willingness to invest in other women. Males face a similar assessment for long-term mating. Lukaszewski and Roney (2010), therefore, predicted that mate preferences for personality traits would depend on the target of the partner's behavior. Subjects rated kindness and trustworthiness toward the subject or his/her family much higher than the same behaviors directed toward others. Conversely, subjects preferred partners who were dominant to members of the partner's sex over partners who directed their dominance toward the subject.
Methodologically, studies of attractiveness must be precise in stipulating the social value being assessed and trade-offs between different aspects of social value. Some cues of social value will be recurrent across domains, whereas others will be domain specific. For example, across all domains of social value, cues of health are predicted to influence attractiveness because health is a valuable asset for all positive interaction partners (Sugiyama, 2005). Another factor to consider is the temporal scope and informational integration of the assessment process: The perception of a person's relative attractiveness in a given domain may change with long-term observation or behavioral interaction. For example, an otherwise physically beautiful person who too highly overvalues him/herself relative to the assessor's evaluation will be found less attractive as a cooperative partner (Sell, Tooby, & Cosmides, 2009; Tooby, Cosmides, Sell, Lieberman, & Sznycer, 2008). A methodological concern is the terminology used to elicit preferences and assessments from subjects: attractive, cute, sexy, and handsome do not mean the same thing, and each term appears to reflect a different constellation of social value traits (Sugiyama, 2005). The question of how different attractiveness adaptations relate to each other and to different aspects of social value requires much additional work, but headway is being made.
Finding, attracting, and maintaining a relationship with another individual long enough to reproduce is complicated, fraught with potential missteps, and conflicts of interest. Raising offspring entails additional problems. Only a very limited subset of attractions and behaviors lead to successful reproduction under a given set of circumstances. Mate selection and mating strategies have therefore been under intense selection.
Human mating is flexible, exhibiting both long- and short-term mating strategies, serial monogamy, some degree of polygyny and lesser degrees of polyandry (e.g., Buss & Schmitt 1993; Daly & Wilson, 1983). Both sexes may engage in extra-pair copulations (e.g., Greiling & Buss, 2000; Thornhill & Gangestad, 2008). Mating effort includes the identification and assessment of potential mates, and the allocation of time and energy to courtship. People differ in mate value, defined as the degree to which an individual would promote the reproductive success of another individual by mating with him/her. Mate value includes residual reproductive value—the probable number of future offspring a person of a certain age and sex will produce (Symons, 1979). Human reproductive value is usually discussed in species-typical perspective, but because reproductive value may vary across local and individual conditions, we must frame our predictions accordingly. Over time, selection favors alleles that organize developmental properties that identify, assess, and integrate cues of high mate value, and motivate individuals to be attracted to conspecifics exhibiting these cues, because these preferences likely led to more successful reproduction than alternative designs. The sum of these assessments contributes to our perception of a potential mate's “physical attractiveness.”
Components of human mate value include species, sex, age, degree of relatedness, health, status, kindness, intelligence, and willingness and ability to mate with ego and invest in ego's offspring. Our mate-selection psychology must assess a potential mate for cues associated with each of these components, weigh their relative importance under current and probable future conditions, integrate these inputs to arrive at a comprehensive estimation of mate value, and regulate a graded emotional and behavioral response. Some features associated with high male-mate value differ from those associated with high female-mate value; criteria of male and female attractiveness are expected to differ when this is the case (e.g., Buss, 1989; Symons, 1979).
Some individuals have higher mate value than others. The result is competition for access to mates, especially high-quality ones. Darwin referred to the selective force created by this competition as sexual selection. Intrasexual selection refers to the selection of traits (e.g., tusks, body size, and musculature) that enhance their bearer's chances of gaining sexual access to the opposite sex relative to same-sex competitors. Intersexual selection is the process whereby individuals with a given trait are preferred by the opposite sex as mating partners, with the result that said trait is spread, elaborated, or maintained in the population even if it has no survival value (Darwin, 1871). If members of the choosing sex are sexually attracted to a feature of the chosen sex (e.g., a longer-than-average tail), and if offspring inherit these traits and preferences, then the preferred trait can become highly exaggerated (e.g., the peacock's tail), a phenomenon referred to as runaway sexual selection. Alternatively, good-genes sexual selection hypothesizes that attractive individuals who have higher mating success may also have other high-fitness attributes associated with heritable genetic variation, such as lower mortality or greater parasite resistance (Hamilton & Zuk, 1982). Mate choice for these attributes (or their correlates) would thereby increase the genetic quality of the offspring. Besides good genes selection, a number of species appear to exhibit mate choice for material provisioning and protection of mate and offspring.
Biologists differentiate cues from signals: Signals are traits selected for because they carried specific meanings that changed the behavior of recipients in ways that benefited the receiver, whereas cues were not modified by selection to carry meaning per se (e.g., Bradbury & Vehrencamp, 1998; Smith & David, 2003). In practice, it is sometimes difficult to distinguish signals from cues, because what are initially cues can be shaped by sexual selection to carry specific meanings that change the behavior of perceivers. Costly signaling theory posits that traits associated with good genes or provisioning of material benefits can evolve into elaborate displays, which function as “honest” signals about underlying phenotypic and genotypic qualities of their bearers (Grafen, 1990; Zahavi & Zahavi, 1997). When a trait signals information about its bearer that is useful for the bearer to transmit and for the recipient to receive, then false signals might also be selected for, undermining the signal value of the trait for both sender and receiver. However, if the cost of producing the signal is such that only some individuals can afford to fully develop it, and that cost is linked to the underlying phenotypic or genotypic quality being signaled, then recipients can be assured of the signal's “honesty.” Elaborate anatomical features, such as the peacock's tail, can evolve this way: Only high-quality males can produce the finest displays, so peahens can reliably use male display in their mate choices, and the fitness costs of the display to the peacock are offset by his increased mating opportunities. Costly signals are not restricted to mating: They can evolve whenever the conditions outlined previously are met (Grafen, 1990; Zahavi & Zahavi, 1997; c.f. Donaldson-Matasci, Bergstrom, & Lachmann, 2013). Without clear evidence that a trait used in attractiveness assessment has evolved as a signal, we should first consider hypotheses that it is used as a cue.
Because mating competition is costly, selection produces adaptations that assess one's mate value relative to potential rivals. This saves time, energy and physical costs by averting competition with rivals the individual is unlikely to outcompete. Conversely, our ancestors could also increase their mating access by driving off, dominating, outshining, or undermining (e.g., poaching mates from) rivals against whom they had a reasonable chance of success. This evaluation entails same-sex attractiveness assessment—not for the purpose of mating, but for determining one's relative mate value and intrasexual competitiveness (e.g., Pawlowski & Dunbar, 1999; Puts, 2010).
For example, men have traits indicating strong selection for intrasexual competition. They can accurately assess the relative strength of other males using body, vocal, or facial features, and these correspond to assessments of fighting ability and intrasexual competitiveness (Puts, 2010; Sell et al., 2009). These physical traits develop under regulation from an increase in androgens, particularly from puberty (e.g., Bribiescas, 2006). Females, in turn, exhibit mate choice for these features of size and musculature associated with strength (Sugiyama, 2005). Dijkstra and Buunk (2001) show that males experience more jealousy in response to a potential rival with higher shoulder-to-hip ratio (i.e., a V-shaped torso). Taller and more dominant men are less sensitive to these cues than shorter and less dominant men, and are also less jealous, indicating regulation of these perceptions in relation to self-other relative competitiveness (e.g., Buunk, Park, Zurriaga, Klavina, & Massar, 2008; Watkins, Fraccaro, et al., 2010; Watkins, Jones, & DeBruine, 2010). Men find these traits unappealing in mating competitors but may find them attractive in allies or cooperative partners, whereas women prefer cues of strength and intrasexual dominance in mates. We may thus expect some concurrence in male and female assessments of an individual's sexual attractiveness to members of the opposite sex. But we should also expect systematic variation in these assessments, partly affected by context (including relative intrasexual rivalry between assessor and assessee) and the social value of these cues in other domains.
For women, male-mate value includes traits associated with genetic quality, health, and physical formidability, as well as traits associated with ability and willingness to invest in a woman and her offspring (Symons, 1979). Assessments based on these two criteria may diverge. In long-term relationships, women often have to trade off physical attractiveness for willingness to invest, because they often have to trade off genetic and phenotypic quality for investment (Gangestad, Thornhill, & Garver-Apgar, Chapter 14, this volume). However, these trade-offs are context dependent: Women place more importance on physical characteristics in short-term and extra-pair sex partners, and during the fertile phase of their ovulatory cycles. A woman's own attractiveness, mating status, and preference for short-term mateships affect the degree of this ovulatory cycle effect, because they affect the relative trade-off she faces between a mate's genetic or phenotypic quality. For instance, in western and some nonwestern contexts, women show greater preference for “masculine” faces, voices, and bodies in short-term mates than in long-term mates. Recent research also shows cross-cultural variation in women's preference for degree of facial masculinity (Scott et al., 2014). In assessing rivals, women focus on physical cues and behaviors related to female mate value (e.g., nubility, youth, fertility, fecundity, health). Women's intrasexual competition enhances these for self and downplays them in rivals, mediated by their own physical attractiveness. Conversely, women downplay their own promiscuity and denigrate it in others, especially in long-term mating competition (e.g., Bleske & Shackelford, 2001; Buss & Dedden, 1990).
It follows that cue/signal detection, assessment, integration, and motivational adaptations are integrated-but-separable components of attractiveness psychology, just as receptors, cue-assessment integration, and pleasure responses are separate components of umami attraction. In assessments of male sexual attractiveness, for example, women might experience feelings of desire (if the male were judged attractive), repugnance (if the male were judged unattractive or identified as close kin), or indifference. In contrast, men might experience feelings such as submissiveness (if the male were judged attractive or dominant) or self-confidence and dominance (if the male were judged unattractive or less dominant). Men and women have different adaptive objectives when evaluating the sexual attractiveness of a given male or female. Men must decide whether they should provoke, avoid a confrontation with, or cooperate with another male, and have, therefore, been under selection to evaluate the prowess of other males vis-à-vis their own. Women must decide whether they should copulate with, ally with, or avoid a given male, and have, therefore, been under selection to evaluate males in terms of the fitness costs and benefits they present as mates and fathers. In evaluations of female attractiveness, men must decide whether they should copulate with, cooperate with, or avoid a given female, and have, therefore, been under selection to evaluate females in terms of their fertility and sexual accessibility. Women must decide whether they should provoke, avoid a confrontation with, or befriend another female, and have, therefore, been under selection to evaluate the attractiveness and dominance of other females vis-à-vis their own.
Note that costs and benefits associated with these outcomes may vary by sex, individual, and circumstance. This gives rise to interindividual variation in attentiveness to different cues/signals, cue/signal assessment and integration, perceptions of attractiveness, and behavior. In long-term contexts, some factors are important for both sexes (e.g., kindness, social status, physical attractiveness) but differ in relative importance to men and women. For example, Shackelford, Schmitt, and Buss (2005) used factor analysis to examine trade-offs among four dimensions of mate preference: dependable/stable versus good looks/health; love versus status/resources; education/intelligence versus desire for home/children; and sociability versus similar religion. Physical cues are hypothesized to be involved in assessment of all these dimensions of mate value, except preference for similar religion. Further, well-documented sex differences indicate that overall, men place more value on good looks in a long-term mate than do women, whereas women place more value on mate's status and resources (e.g., Buss, 1989).
Determination of mate value entails a number of adaptive problems, the solution to each of which will affect perceptions of attractiveness on the part of the assessor. The first step in this process is identification of viable mates. Obviously, copulation with inanimate objects, other species, or sexually immature humans is ineffectual for reproduction. Copulation with members of the same sex is similarly ineffectual for reproduction per se, although it may be of indirect benefit as a means of recruiting allomothers. Copulation with carriers of contagious disease entails fitness costs, and copulation with individuals bearing genetic anomalies can result in pregnancies that produce non- or less-viable offspring, as can copulation between close genetic relatives. Forced copulation with fertile human members of the opposite sex entails opportunity, reputational, and possibly retributive costs. In assessing rivals, men may, therefore, focus more on physical cues associated with potential rivals' formidability and dominance in assessing themselves vis-à-vis competitors because these attributes could spell death or loss of a mate at the hands of a rival or rival coalition. Data from psychological studies, homicide patterns, and intratribal conflict support the view that various aspects of mating competition are often causes of violence, and size and strength are assets in this competition (e.g., Buss, 2006; Chagnon, 1988; Daly & Wilson, 1988; Macfarlan, Walker, Flinn, & Chagnon, 2014; Puts, 2010; Scalise Sugiyama, 2014).
Women's attractiveness-assessment psychology is predicted to include mechanisms for evaluating cues associated with male genotypic quality. One cue to genotypic quality is phenotypic condition, part of which is heritable. Male mate value also includes material provisioning of mates, offspring, and other adults: Among foragers, men provide about 85% of the protein and 65% of the calories to the diet (Cordain et al., 2000; Kaplan et al., 2000; Marlowe, 2001), with positive effects on female fecundity and offspring immune function, health, and survival (Gurven & Hill, 2009). Across societies, women appear to assess prospective long-term mates using cues of willingness and ability to invest in a mate and her offspring, such as kindness, intelligence, industriousness, and ability to acquire resources (e.g., Buss, 1989). Male ability and willingness to invest is important for females because of the high costs of pregnancy, lactation, the long period of juvenile dependence, and short interbirth intervals resulting in multiple dependent offspring (e.g., Kaplan et al., 2000). Ache juveniles with father living suffer a third lower mortality than those whose father has died (Hill & Hurtado, 1996), and hunter-gatherer males contribute significantly to subistence (e.g., Gurven & Hill, 2009), although effects of fathers on correlates of offspring fitness vary across social and ecological contexts (e.g., Bribiescas et al., 2012; Hewlett & Macfarlan, 2010; Marlowe, 1999a, 1999b, 2001, 2005; Sear & Mace, 2008).
Human males grow for a longer period, mature more slowly, and reproduce later than females (e.g., Bogin, 1999). They also exhibit higher interindividual variance in reproductive success than females (e.g., Betzig, 2012). Because paternity is less certain than maternity, men's age at first reproduction is harder to track directly, but males in foraging societies appear to begin reproducing in their early 20s—several years later than females. Age-related changes in male fertility among the Ache, !Kung, and Yanomamö indicate a rise in fertility beginning in the late teens and peaking in the mid-30s to early 40s. Mean age at last birth for 23 Ache men who lived to at least age 60 was 48 years: Although half of the men ceased reproducing as early as women did, the other half reproduced for longer periods, including six who continued past their mid-50s. Further, male foraging success peaks relatively late in life, ranging from the 30s to almost age 50 (Kaplan et al. 2000; Walker, Hill, Kaplan, & McMillan, 2002). Apicella (2014) found that strength predicts reproductive success and reputation for hunting ability among Hadza foragers of Tanzania; however, strength peaks much earlier than hunting return rates, indicating the important role of knowledge in hunting (e.g., Gurven, Kaplan, & Gutierrez, 2006). Because male mate value is not so closely linked to youth, female preference mechanisms are expected to target cues of genotypic and phenotypic quality and productive ability rather than youth per se (Buss, 1989). Selection is expected to have favored female assessment for phenotypic cues of male fertility. However, since one fertile male can potentially inseminate multiple females, preference for cues to fertility per se is less intense in women than in men (Symons, 1979).
Women can benefit from pursuing a mix of long- and short-term mating strategies, to reduce trade-offs inherent in each (e.g., Buss & Schmitt, 1993; Gangestad & Simpson, 2000). It is now clear that studies should incorporate methods that disentangle these considerations. From a female perspective, poor health and genetic quality are liabilities in any prospective mating partner. However, women are expected to find physical traits linked to underlying genetic qualities relatively more important in short-term than in long-term mates. Long-term mateships entail childrearing; thus, prospective long-term partners must be evaluated for their parenting abilities and good-partner qualities as well as their physical attributes. Thus, size, strength, pugnacity, and physical dominance may be traded for ability and willingness to invest in the woman and her offspring, although attractive women don't face these trade-offs so they desire high levels of both (Buss & Shackelford, 2008). For women, parenting skills are less important in a short-term mate, for obvious reasons. Because men relax their standards for short-term mates, short-term mateships can offer some women access to higher genetic quality sires for their offspring than they could acquire in a long-term partner, and many of the traits associated with aggressive formidability—for example, size, strength, and facial masculinity—are proximate cues of genetic quality (e.g., Buss & Schmitt, 1993; Gangestad, Merriman, & Thompson, 2010; Thornhill & Gangestad, 2003).
Males face investment trade-offs between mate quantity and mate quality. Local paternal effects on offspring fitness affect the costs and benefits associated with each and the relative costs and opportunities of obtaining multiple mates. The latter will be affected by a given male's mate value, local degree of effective polygyny or operational sex ratio, and relative values of long-term and short-term mating for women. Some physical traits are associated with differences in male-mate value that may influence the male's propensity to pursue short- or long-term mating strategies. Females may use these traits as cues to probable male-mating behavior. Women may be expected to use these same criteria in their assessments of the relative social value of their fathers, brothers, and other male kin to others, but to weight the criteria differently.
Access to women's reproductive capacity constitutes a primary constraint on men's relative reproductive success. Human female mate value is, therefore, closely linked to age-related reproductive life stage, health, fertility, and parity (see Sugiyama, 2005). Women have delayed maturity compared to that expected for a primate of our size, and cease reproduction some 20 years prior to death, resulting in a compression of the reproductive life span. Age at first birth for female chimps is around 12 years, whereas for female foragers in natural fertility conditions it is about 17 years (Hill & Hurtado, 1996; Kaplan et al., 2000; Thompson et al., 2007). A woman's reproductive value is highest just before she begins fertile ovulatory cycles, because the number of reproductive years ahead of her is highest and the probability that she will die prior to reproduction is lowest. Fertility varies across the reproductive lifespan. Peak age-specific female fertility in industrialized nations is around 22 years and may show significant declines by 27. Data from foraging populations indicate peak age-specific fertility varying from about 22 to 25 years among the !Kung of Botswana and the Yanomamö of Venezuela to about 28 to 35 years among the Ache of Paraguay. Diet, work effort, pathogen and social stress, and other social variables affect hormonal indices of female fertility and fecundity, suggesting that female reproduction varies with socioecological variables affecting energetic availability (e.g., Ellison, 2001, 2003; Jasienska, 2009; Valeggia & Ellison, 2009).
Women's minimum necessary maternal investment is high. It includes accumulation of bodily reserves and maintenance of a positive energy balance, placentation, gestation, and the mortality risk associated with bearing a large-headed offspring through a relatively narrow pelvic opening (e.g., Ellison 2001, 2008; Rosenberg & Trevathan, 2002). Fecundity depends on hormonally regulated ovarian function, which tracks energetic availability and demands. For women of normal BMI, pregnancy increases energetic requirements by roughly 90, 300, and 466 calories a day during the first, second, and third trimesters, respectively. Breastfeeding increases energy requirements by about 450 to 500 calories per day in healthy western women who are not particularly active (Butte, Wong, Treuth, Ellis & Smith, 2004). Lactation suppresses reproductive function in relation to a woman's energy budget (e.g., Ellison, 2003). Humans have shorter interbirth intervals than expected given these costs. The mean weaning ages for 30 hunter-gatherer groups reported in R. L. Kelly (1995) averages 30.9 months (Sugiyama, 2005). The interbirth interval for women in a group of 11 foraging societies is 3.47 years, and the average total fertility rate is between five and six children. The costs of pregnancy, lactation, short interbirth intervals, and multiple dependent children appear to be offset by slow child growth and allomaternal work effort and food provisioning (e.g., Gurven & Walker, 2006; Hrdy, 1999).
Women experience reproductive decline earlier than senescence of other bodily functions (e.g., Hill & Hurtado, 1996; Thompson et al., 2007). Female foragers may thus live well past their reproductive years, although maternal and grand-maternal investment of resources and social support in offspring may continue into offsprings' adulthood. Among Ache women, the average age of last birth is 42. By age 46 the yearly probability of birth is 0 (Hill & Hurtado, 1996). R. L. Kelly (1995) lists data on mean age at last birth for women in 10 foraging societies; the average mean is 34.9 years (Sugiyama, 2005).
The human female reproductive environment of evolutionary adaptiveness (EEA) was such that for most of the time between menarche and menopause a woman was not fecund. Symons (1995) calculated that a Yanomamö woman can possibly conceive on just 78 of 8,030 days during her average reproductive lifespan. My calculations (Sugiyama, 2005) based on R. L. Kelly's (1995) data on foragers yielded a broadly similar conclusion. With an average age at first birth of 17 and average age at last birth of 42 (for Ache), an average female forager's potential fertile lifespan is about 25 years, during which she is likely to have five children. On average, she would have been pregnant or lactating for 5,985 days—almost two-thirds of her reproductive lifetime. With 3 fertile days per month she might be fecund on only 314 days in her 9,125-day fertile lifetime, assuming she suffered no ill health, food deficiencies or other stressors that limited fecundity.
Since female reproductive value declines with age after menarche, cues associated with advancing age are expected to be negatively correlated with female sexual attractiveness (Symons, 1979). With each birth, the average forager woman loses another sixth of her reproductive value. Thus, cues associated with parity are expected to be negatively correlated with female sexual attractiveness. Because some cues to fecundity are observable, we may also expect adaptations that use statistically reliable cues to fecundity-related hormonal status in assessments of female mate attractiveness. Symons (1979, 1995) predicted that this would result in males being attracted to cues of nubility, or highest reproductive value (i.e., female has begun ovulatory cycling but not yet given birth). Since women do not advertise estrus (or do not do so widely), attraction to cues of nubility would dramatically increase a male's chances of reproducing. A man who maintained exclusive mating access to a woman over her reproductive lifetime could, on average, sire five or six children with her. Preference for cues of peak fertility would increase probability of conception, particularly for short-term mates (e.g., Symons, 1979).
Women with positive energy balance and good health are likely to be more fertile than those with negative energy balance and poor health. Thus, men are expected to have evolved preference mechanisms that find cues of good health and nutrition attractive, and women are expected to use the same cues in assessments of their reproductive rivals. Even though selection may have produced attraction to cues of nubility, attraction to these cues alone might compromise long-term mateships, and would have the effect of concentrating male reproductive effort on fathering only the first of a woman's average six offspring. Other cues that a woman is resuming ovulatory cycling postpartum, such as lightening of the skin (Symons, 1995) or having a child approaching weaning age, should predict some of the variance in real-world female sexual attractiveness. Although self-report measures and other studies have failed to find male preference for peak residual reproductive value, experimental tests including the relevant range of stimuli do (Blackwell & Sugiyama, 2008). Across cultures, physical attractiveness ranks high among the criteria that men look for in mates (e.g., Buss, 1989). Even where self-report findings suggest that attractiveness is not so highly valued, more detailed methods reveal strong preference for physical attractiveness (Pillsworth, 2008). Women appear to be sensitive to this preference, as indicated by the highly lucrative cosmetics and associated beauty industries.
Anthropologists have long recognized that, in prestate societies, social relationships are organized by kinship and kinship-like institutions. All known human cultures include three basic kinds of social relationships based on relatedness: marriage, descent, and kinship classification systems. Classificatory kinship systems are based on three conceptual primitives: sex, descent, and generation. These systems fall into several basic types, which vary in terms of how kinship is parsed along these basic dimensions. Who is and is not an appropriate marriage partner is often based on both classificatory kinship and descent (e.g., Chagnon, 1997). These common features of social organization reflect the value placed on kinship cross-culturally (e.g., Brown, 1991; D. Jones, 2003).
Kin selection and parental investment theories help explain these values, even though classificatory kinship and biological kinship do not completely overlap. Individuals can increase the alleles they bear not only via their own reproduction, but also via aid to those with whom they share those alleles by virtue of recent common descent. Hamilton (1964) thus showed how one evolutionary pathway by means of which altruism can arise: when the cost to the altruist is less than the benefit to the recipient devalued by the probable degree to which they are related. From ego's perspective, then, others vary in kin value both as potential investors and as foci of investment.
Trivers (1972) predicted the adaptive problems parental investment mechanisms need to solve, and the logic of his argument can be extended to kinship more generally and to other domains of human social value. People vary in (a) their probable degree of relatedness to ego, (b) their ability to translate investment by ego into fitness or inclusive fitness, (c) the opportunity costs to ego imposed by that investment, and (d) their willingness and ability to invest resources in ego, ego's offspring, and ego's other kin. The probability that kin can translate investment into successful reproduction is affected by their phenotypic and genotypic quality, including the related variables of health, age, fertility, fecundity, and sex, all of which are associated with physically observable cues. To the degree that potential kin exhibit reliable cues to these values, they are expected to be more attractive than others as sources or targets of investment. Conversely, because inbreeding with close genetic relatives increases the probability that offspring will be homozygous for deleterious alleles, close kinship should negatively affect attractiveness as a sex partner (or sexual value; Tooby et al., 2008).
Data from nonwestern, natural fertility populations show that kin effects on fitness-related traits are significant and vary locally by context. For instance, Hagen, Barrett, and Price (2006) found older brothers had positive effects, and older sisters negative effects, on younger sibling growth among Shuar (see also Hagen & Barrett, 2009). In a larger regional sample of Shuar villages, Blackwell (2009) found quantity/quality trade-offs between number of household siblings and growth. However, they also found a U-shaped relationship between distance to road and boys' effects on siblings' growth, and an inverse U-shaped relationship between distance to road and girls' effects on younger siblings' growth. These results are hypothesized to reflect local sexual division of labor, resulting in sexual variation in productive abilities depending on where villages are located.
The regulation of behavior based on kinship requires cues statistically associated with relative degree of relatedness and adaptations that assess these cues to estimate kinship. Estimated kinship, in turn, regulates emotional outputs and discriminative behavior in relation to the social value at issue (e.g., Lieberman, Tooby, & Cosmides, 2007). Tooby et al. (2008) refer to this kinship regulatory variable as the kinship index. The kinship index can be based on contextual cues, prior association during development, or phenotype matching (Axelrod, Hammond, & Grafen, 2004; Mateo, 2015). In humans, we see evidence of all three, and their effects on attractiveness.
Mother-offspring mutual recognition occurs quickly, based on olfactory, visual, tactile, auditory, and behavioral interactions. One cue of sibship is one's mother caring for an infant (maternal perinatal association, or MPA). For older siblings, one's mother nursing an infant is a statistically reliable cue of sibship or half-sibship. For younger siblings, these cues are not available, and duration of childhood coresidence is used instead (e.g., Lieberman et al., 2007). Kinship index influences at least two different aspects of attractiveness, but in opposite ways. For those indexed as siblings, it increases attractiveness as a target of altruism, but decreases attractiveness as a sexual partner (Lieberman et al., 2007). The same logic holds for parent-offspring altruism and sexual aversion. Kinship index should also regulate assessments of altruistic, reciprocal, and coalitionary value. Other evolutionarily relevant cues of an individual's relatedness to ego are: observing a close female relative give birth to or nurse said individual; observing kin exhibiting altruistic behavior or sexual avoidance toward said individual; and perhaps the use of kinship labels by kin in reference to said individual (e.g., Lieberman et al., 2007).
Evidence suggests that phenotype matching based on visual and olfactory cues is also used in kin estimation and attraction. Facial self-similarity increases trust and cooperation, but decreases sexual attraction (e.g., DeBruine et al., 2011, Lieberman et al., 2007). Phenotype matching based on olfactory cues is also apparent, based in part on underlying genetic influences (e.g., Porter, Balogh, Cernoch, & Franchi, 1986; Roberts et al., 2005). Moreover, father-daughter and brother-sister pairs show odor-based mutual sexual aversion (Weisfeld, Czilli, Phillips, Gall, & Lichtman, 2003). The interest of kin in the mating behavior of others is mediated by degree of perceived physical similarity (Faulkner & Schaller, 2007). Tellingly, independent raters can discriminate relatedness of others based on photographs (e.g., Alvergne, Faurie, & Raymond, 2010; DeBruine et al., 2009; Kaminski, Dridi, Graff, & Gentz, 2009).
People differ in their probable value to kin of ascending generations in terms of their social value as reproductively successful descendants and as contributors to other kin. Parental investment (PI) theory focuses on resource allocation trade-offs among existing offspring, current and future offspring, and the quantity and quality of offspring (e.g., Trivers, 1972). Trivers predicted that parental investment should be allocated in response to three assessments: the probability that the juvenile is (1) one's own progeny, (2) is able to translate investment into future reproductive success, and (3) is a better investment of those resources than alternate potential uses. Some cues to (1) and (2) are observable, and selection has produced adaptations that use these cues to assess a juvenile's attractiveness as a target of investment.
In both women and men, offspring recognition is effected in part via adaptations that enable rapid learning of olfactory, visual, auditory, and tactile cues associated with the infant (e.g., Porter, 1991). These cues regulate attention and attraction by means of hormonal and neurological mechanisms that motivate bonding and caretaking behavior (e.g., Swain et al., 2014; Winberg, 2005). For example, the odor of newborns stimulates greater activation of the dopaminergic system in new mothers than in nulliparous women (Lundström, Boyle, Zatorre, & Jones-Gotman, 2009; Lundstrom et al., 2013). Oxytocin levels are affected by pregnancy, birth, and lactation, and are positively associated with a mother's attraction to her infant, including infant-directed gaze, monitoring of infant, and use of “motherese,” affect, and touching. Women also exhibit attentional biases that appear to help regulate childcare. Mothers pay more attention to infant faces than do nulliparous women. This enhanced attention is not a general response to all faces. Mothers pay more attention to infant than child, adolescent, or adult faces, and more attention to infant and child faces when they exhibit distress. Fathers' oxytocin levels parallel those of mothers, but generate slightly different behavior, including more stimulatory contact with the infant, exploratory encouragement, and infant-directed attention to objects (Feldman, Weller, Zagoory-Sharon, & Levine, 2007; Gordon, Zagoory-Sharon, Leckman, & Feldman, 2010). Individual differences in mothers' oxytocin levels are regulated by epigenetic processes linked to early childhood environment, and to mechanisms that take in more immediate contextual variables to compute available resources and probable and actual allomaternal aid (Hrdy, 2009; Swain et al., 2014).
Conversely, research indicates that stepparents invest less in their stepchildren than their biological offspring, and that stepchildren are less attractive or even aversive as investment targets. A number of studies show that stepfathers invest less than biological fathers (e.g., Anderson, Kaplan, Lam, & Lancaster, 1999; Anderson, Kaplan, & Lancaster, 1999; Flinn, 1988). For example, Hadza men living with stepchildren bring in less food than those living with biological children only (Marlowe, 1999a). Stepparents also invest less time and energy in supervision, making stepchildren more vulnerable to fatal accidents (Tooley, Karakis, Stokes, & Ozanne-Smith, 2006). Children living with stepparents are also at elevated risk of abuse and homicide (Daly & Wilson, 1985, 1988). Stepparental investment by males does occur but, tellingly, research suggests that it is used as a means of forming a mateship with the mother (Anderson, 2000). Stepfathers are reported by women as perpetrators of sexual abuse much more often than are biological fathers (Russell, 1984), as expected if sexual attraction is down-regulated by kinship index.
Depending on how and why data were gathered, an estimated 1.7%–30% of children are estimated to be sired by men who are not the putative father (Anderson, 2006). Ability to recognize their own versus others offspring via phenotype matching would provide beneficial input into the kinship index, regulating paternal investment in putative offspring.
Resemblance to self affects attractiveness for investment in hypothetical and real life contexts. When presented with different facial morphs, created using each subjects' image and those of children, males were more likely to choose their own child/face morphs over others as recipients of aid in hypothetical investment scenarios (Platek, Burch, Panyavin, Wasserman, & Gallup, 2002; Platek et al., 2003; Platek et al., 2004). Volk and Quinsey (2007) found facial resemblance more important to men than women in hypothetical adoption scenarios, but other studies found that both men and women used facial self-similarity in investment decisions (Bressan & Zucchi, 2009; DeBruine, 2004). Functional magnetic resonance imaging also showed men's and women's neural activation patterns differed when viewing self but not nonself-morphs, suggesting sex differences in neural processing of facial self-resemblance cues, and Alvergne, Perreau, Mazur, Mueller, and Raymond (2014) found specific facial features used as paternity cues, and that these features are those that change less with development. Apicella and Marlowe (2004) found that men reported greater investment in their children when they thought their children had greater psychological and physical resemblance to themselves. Burch and Gallup (2000) found that among men in a domestic-abuse facility, resemblance was positively associated with men's self-reports of relationship quality with their children and negatively with severity of spousal abuse. Alvergne et al. (2010) found that although mothers' assessments of facial resemblance to children corresponded with fathers' perceptions, fathers' but not mothers' self-reported emotional closeness to children was predicted by actual facial self-resemblance. Mothers also exhibit interest in the resemblance between offspring and putative fathers, as it provides a cue to probable paternal behavior (Daly & Wilson, 1982; Regalski & Gaulin, 1993).
Porter, Cernoch, and Balogh (1985) found that third parties could correctly match mother-offspring odors but not husband-wife pairs, suggesting an underlying genetic mediation of olfactory kin recognition. Among Senegalese coastal populations, Alvergne, Faurie, and Raymond (2009) presented subjects with a child's face or odor, and asked them to pick the true father from amongst either three (facial) or two (odor) choices. Subjects correctly identified the father more often than expected by chance using both visual and olfactory cues. Facial and olfactory resemblance was also positively associated with paternal investment as rated independently by the child's mother, and paternal investment was positively related to the child's BMI and upper arm circumference. Olfactory phenotype matching can be based at least in part on a resemblance to self, because our ancestors could obviously experience the chemosensory cues to their own as well as others' odors. This is particularly useful in generating estimations of close relatedness. Kinship estimate can also be based on chemosensory exposure to individuals exhibiting other cues of close kin, useful in determining “family” resemblance for slightly more distantly related kin.
Understanding the socioecological context in which phenotype-matching systems evolved is critical to hypothesis formulation and testing. For example, facial morph studies primarily use self-based resemblance, but it is unlikely that human environments of evolutionary adaptedness (EEA) provided sufficient opportunities for self-observation to support the evolution of solely self-referential facial phenotype matching. It seems more likely that visual phenotype matching is generated using a kin template constructed by observing close kin (e.g., DeBruine, Jones, Little, & Perrett, 2008). To test this, Bressan and Zucchi (2009) took facial photos of 17 monozygotic and 18 dizygotic Italian twin pairs and morphed each of these with a model's face to produce images composed of 65% of the model's face and 35% of the subject's face. Two months later, subjects were presented with the morphs of their own and their twin's faces and asked to choose (a) which they would help in case of danger and (b) which they would encourage an opposite-sex sibling to marry. Subjects could not recognize the faces as self and twin morphs. For both questions, though, the self-morph was chosen significantly more often than the twin morph, with no difference by twin type or sex.
Bressan and Zucchi conclude that this shows self- rather than twin-referential phenotype matching, arguing that because subjects saw their own twins more often than their own faces, a kin-based phenotype matching template system would generate a sibling rather than self-referential bias. However, all subjects in the study had access to mirrors, so their own facial features may have been input disproportionately into the kin phenotype template (e.g., DeBruine et al., 2008). A functional facial phenotype-matching template generator could not be indifferent to the kinship index of individuals used in generating the template, for the simple reason that our ancestors lived in a densely kin-populated environment. Clearly, more research is needed to illuminate these processes. One obvious test would be to determine whether people with little access to mirrored surfaces also use self-referential similarity over kin-referential templates. Or, less optimally, where there is variation in mirror access, one could determine whether greater access is associated with greater use of self-referenced cues. The number of populations in which such a study is possible is rapidly approaching zero, so it will need to be done soon.
The probability that a juvenile will translate a given amount of investment into successful reproduction is related to that individual's sex, age, genotypic and phenotypic condition, and to socioecological context (e.g., Trivers, 1972; Trivers & Willard, 1973). Physical cues that were evolutionarily correlated with good health and high genetic quality provide observable correlates of a juvenile's probable ability to translate investment into reproduction, and are expected to be found attractive in offspring. Physical cues of low genotypic or phenotypic quality are associated with reduction in parental care, suggesting that these traits are unattractive to parents. For instance, physical deformity is a recurrent proximate cause for infanticide cross-culturally (Daly & Wilson, 1988), and vocal qualities associated with premature birth are aversive to adults (Furlow, Armijo-Pruett, Gangestad, & Thornhill, 1997; Mann, 1992). Poor physical tone, lethargy, or lack of pedomorphic characteristics in infants increase risk of abuse and maternal neglect when resources are scarce (e.g., Daly & Wilson, 1981; Hrdy, 1999; McCabe, 1984). Conversely, physical cues associated with infancy such as large eyes, small noses, and a rounded head are attractive to parents and others (Alley, 1983; Zebrowitz, 1997). Parents of attractive infants are more attentive and affectionate toward them (Langlois, Ritter, Casey, & Sawin, 1995), as are those in allomaternal roles (e.g., teachers) and nonrelated others (Glocker et al., 2009). Meta-analysis shows that less attractive children receive less caregiving (Langlois et al., 2000), and parents rate less attractive infants as older and developmentally more mature (Ritter, Casey, & Langlois, 1991), even though this is objectively not the case.
Resources are finite, and parents must decide whether, and how much, to invest in existing offspring, future offspring, own somatic resources, and mating effort. Trade-offs between investment in quantity and quality of offspring are documented in some cases (e.g., Blackwell, 2009; Gillespie, Russell, & Lummaa, 2008; Hagen, Hames, Craig, Lauer, & Price, 2001; Hagen et al., 2006; Sellen, 1999; Strassman & Gillespie, 2002). The outputs of mechanisms that make these assessments, in conjunction with outputs of the kin index and phenotypic quality evaluators, are thus expected to up- or down-regulate the attractiveness of offspring. Of course offspring are not passive recipients of whatever care others might deign provide. They are predicted to have adaptations that evaluate their own condition, that of their potential caregivers, and other options for acquiring resources, which in turn generate behavior to enhance their own ability to survive and reproduce (e.g., Hewlett & Lamb, 2005; Konnor, 2010; Sugiyama & Chacon, 2005). Possible responses include attempting to increase investment in themselves by others, reducing risks of losing investment, or acquiring more resources on their own. Crying is used by infants to gain attention and investment, and foragers children can and do contribute to their diet by acquiring resources on their own, although this varies across ecological conditions (e.g., Bliege Bird & Bird, 2002; Blurton Jones, Hawkes, & Draper, 1994; Sugiyama & Chacon, 2005). Parents and alloparents, in turn, are expected to be sensitive to juveniles' ability to contribute to their own welfare, and to adjust their reproductive and investment strategies accordingly (e.g., Blurton Jones, Hawkes, & O'Connell, 1997; Daly & Wilson, 1988; Trivers, 1972).
The human ecological niche is characterized by a high degree of cooperation. Studies of modern and prehistoric foraging societies indicate that ancestral cooperative activities included mate acquisition (Apostolou, 2007), child rearing (Hill & Hurtado, 2009; Hrdy, 2007), foraging (e.g., Alvard, 2003, 2005; Hill, 2002), information transmission (e.g., Scalise Sugiyama, 2011), warfare (e.g., Chagnon, 1997; Ember & Ember, 1997; Keeley, 1996), and aid during health crises (Sugiyama, 2004a). Although people probably lived among relatively more kin than we do in the United States, not all those allies would have been close kin (Apicella, Marlowe, Fowler, & Christakis, 2012; Chagnon, 1979; Hill et al., 2011). Even individuals with whom ego does not directly cooperate can have social value when they yield positive externalities such as increasing ego's food supply, attracting potential mates to ego's proximity, deterring attacks, providing information, or helping ego's allies (e.g., Tooby & Cosmides, 1996). Conversely, individuals may have unintended negative effects upon us. For example, unhealthy individuals may increase disease exposure, and impulsively aggressive individuals may incite conflict. Health, physical abilities, generosity, cooperativeness, and intelligence provide at least some cues to a person's value with regard to these recurrent problems of human life.
Although he overstated the case, Levi-Strauss saw marriage in what he called “primitive” society as an exchange between men (i.e., the consanguinial male relatives of the bride and groom). Certainly, who mates with whom is of interest not only to the principals. With its concomitant social, economic, and reproductive rights and obligations, the universal institution of marriage reflects the fundamental interests of individuals in the mateships of their offspring, siblings, and other close relatives. Mateships build alliances and serve as vehicles for a descent groups' reproductive future, and sons- and daughters-in-law play integral social and economic roles. Accordingly, family members regularly assess potential daughters- and sons-in-law regarding their coalitional, productive, and reproductive assets, and the ethnographic literature reveals that many marriages are arranged (Apostolou, 2007). Parents' and offspring's assessments may overlap with regard to long-term mate value (except for assessments contingent on individual's phenotypic condition), but in other respects they may differ. For example, parents might place more importance than offspring on a prospective mate's cooperative qualities and coalitional ties.
Another critical sphere of cooperation is child rearing. Human life history is characterized by a high degree of investment in juveniles provided by individuals other than the biological mother, including biological and social fathers, aunts, uncles, and grandparents. Evidence suggests humans are cooperative breeders, with multiple females and males cooperating in the raising of offspring (e.g, Hill & Hurtado, 2009; Hrdy, 2007; Kramer, 2010; Mace & Sear, 2005). People may thus cultivate relationships with others based on their suitability as alloparents. Relevant cues in making this choice may overlap with cues of long-term mate value, but will diverge in some areas. Obviously, sex of alloparent is less important than sex of mate. Fertility and fecundity might oppositely affect mate and alloparent value: A postmenopausal woman has low reproductive value, but could provide valuable benefits as an alloparent, and would not face a trade-off between investment in allochildren and her own current reproduction. Similarly, prereproductive age females often provide alloparental support for younger siblings. But the opportunity costs of doing so increase as they have children of their own, thus decreasing their alloparental value to their parents.
Cooperation is also critical in the context of coalitional violence, where estimated male mortality from violence in tribal societies ranges from 10% to 30% or more (e.g., Beckerman et al., 2009; Chagnon, 1997; Patton, 2000; Pinker, 2011; Walker & Bailey, 2013), and the evolution of neuroendocrine regulation of coalitionary behavior (Flinn, Ponzi, & Muehlenbein, 2012). In a world without police, standing armies, or hereditary leadership, a reputation for being willing and able to strategically use violence is a deterrent to attack, and a necessary component of becoming a “headman” (e.g., Chagnon, 1997; Patton, 2000). And in a world of close-range, nonmechanized weaponry, individual strength, size, speed, and agility are highly advantageous. Headman often translates as “big” or “big man,” and tribal leaders are often bigger than average (Brown, 1991). Leadership, organizational abilities, and strategic acumen are also valued in coalitional politics, and the value of a coalitional partner is also based in part on his/her reliability, loyalty, intelligence, and willingness and ability to back up coalitional interests with force (e.g., Chagnon, 1997).
Some of these abilities may be assessed through physical and behavioral cues. For example, reliability and ability to defend coalitional interests will be affected by health: Individuals in frail health will be less reliable and less able defenders, and immune-compromised individuals may increase disease transmission among coalition members. Because they further success in foraging, fighting, and deterrence of violence, physical prowess and aggressive formidability are linked to male survival, social status and, consequently, their social value to other males. Thus, cues of physical prowess and aggressive formidability are likely to be important in assessments of male attractiveness by males. Men are expected to display these qualities to other males, and be adept at predicting the outcomes of physical conflicts based on assessment of traits correlated with these qualities (e.g., dominance, tenacity, pugnacity, pain tolerance, agility, strength, endurance). All else equal, men should find males who exhibit these cues attractive as coalition partners. Because successful coalition building, maintenance, and deployment also require certain social and intellectual skills, traits associated with these qualities should also be found attractive in potential coalition partners. Male coalitional assessment psychology must, therefore, be able to weigh the degree to which a given male possesses these abilities, and their relative importance to the coalition in question. A coalition of brawny, athletic warriors lacking planning ability could benefit from adding to its ranks a man who is physically deficient but strategically brilliant.
Phenotypic condition refers to an individual's ability to efficiently acquire resources and convert them into fitness. It can include components such as metabolic efficiency, robustness, foraging efficiency, and toxin clearance. One aspect of phenotypic condition is health, defined as relative presence or absence of injury and/or infectious, chronic, or genetic disease (e.g., Tybur & Gangestad, 2011). Direct benefits of good health for social value include lower infectious- disease-transmission risk, and greater ability to provide the fitness benefits of the value at issue (e.g., Sugiyama, 2005). Healthy associates also reduce costs associated with health-care provisioning and loss of productive contributions (e.g., Sugiyama, 2004a; Sugiyama & Chacon, 2000). Indirect benefits of healthy associates include reduced replacement and buffering costs (e.g., search, pursuit, and opportunity costs of replacing a mate, offspring, or alliance partner; Sugiyama & Chacon, 2000). To the extent that phenotypic condition and health are heritable, mating with healthy individuals also confers those benefits to offspring (Tybur & Gangsestad, 2011). Factors affecting health are complex. Current conditions (e.g., energy stores, diet, exposure to pathogens) affect health, but environmental variables during fetal and childhood development also affect adult health, via their effects on life-history trade-offs. For example maternal nutrition and endocrine status can have epigenetic effects on glucocorticoid receptors affecting stress sensitivity or resilience, offspring metabolism, fat deposition, and muscle development (e.g., Gluckman, Hanson, & Mitchell, 2010; Kuzawa, 2012; Nepomnaschy & Flinn, 2009), each of these are hypothesized to be used in assessments of attraction. However, health consequences of these early life factors are often not apparent until adulthood, and their impact may not be directly evident via just one cue. For example, the concept of allostatic load—the long-term effects of stress on degrading biological function—is now used to generate health measures of the effects of stress, because single measures do not adequately capture or measure these effects (e.g., McEwen, Nasveld, Palmer, & Anderson, 2012). Similarly, if adaptations exist to assess health, single phenotypic health cues may show very little relationship with health outcomes, whereas multiple cue integration may show larger effects.
When testing links between health and attractiveness in postepidemiological transition societies, we must bear in mind that the causes of mortality and morbidity are quite different for most of us than for our foraging ancestors (Harper & Armelagos, 2010; Nesse & Williams, 1994). Foragers and forager-horticulturalists exhibit different mortality profiles than modern industrial populations. Illness was the leading cause of death in 12 out of 13 forager and forager-horticulturalist groups with available data, causing 71% of all deaths overall (n > 3,000; Gurven & Kaplan, 2007). Some deaths may have been due to introduced diseases, but many were not: respiratory illness accounted for 23.7%, gastrointestinal illness for 13.8%, fever for 7.3%, and other diseases for 16.6%. Accidents accounted for about 8% of deaths overall. Mortality rates among these groups were higher across the lifespan than in the United States—30, 100, and 10 times higher for infants, children, and adolescents, respectively. Although male and female mortality rates differed somewhat, in general, age-specific mortality was high during infancy, dropped steeply until the mid-teens, and then remained fairly level until it hit an adult modal around age 72. Conversely, deaths due to chronic disease were few, although they are often harder to identify.
Illness and injury entail fitness costs besides death. Either can significantly reduce productivity, thereby jeopardizing ability to provision self, offspring, and allies. Among Shiwiar forager-horticulturalists, injuries causing disability over a month in duration are common across the lifespan (Sugiyama, 2004a). Among Yora forager-horticulturalists, topical bacterial infection accounted for the majority of days on which individuals were disabled and could not forage or garden, with significant effects on production (Sugiyama & Chacon, 2000). Infected wounds sometimes lead to loss of limbs (Chagnon, 1997; Sugiyama, 1996), with obvious negative effects on productivity.
The immune system is energetically costly to develop, maintain, and deploy. For example, mounting a fever increases total adult resting energy use by 7%–13% per degree Celsius rise in temperature (Hotamisligil & Erbay, 2008). Even in the absence of fever, immune response to mild respiratory infection increased resting metabolic rate of otherwise healthy men 8%–14%, and was associated with a decrease in serum testosterone of 10%–30% (Muehlenbein, Hirschtick, Bonner, & Swartz, 2010). These costs reduce fecundity, and are complicated by pregnancy and lactation, when energetic needs are high (e.g., Ellison, 2003; McDade et al., 2012). For comparison, humans allocate about 20%–25% of our resting metabolic rate to brain function (Leonard & Robertson, 1994).
C-reactive protein (CRP) is a critical component of innate immunity linked with acute inflammatory response. In a 4-week repeated-measures study, 34% of Shuar adults (n = 54) had elevated CRP, indicating new infection at one of the four weekly measures (McDade et al., 2012). In a separate sample of over 300 Shuar, over 50% of participants showed presence of at least one type of helminth, and many had co-infections, with infection prevalence and intensity varying across Shuar territory (Cepon-Robins et al., 2014). Bites or infestation from ectoparasitic insects (e.g., mosquitoes, ticks, chiggers) are common. Besides small but recurrent blood loss and immune activation, many of these result in a secondary infection from being scratched and infected, particularly among children (Sugiyama, 2004a; Chagnon, 1997). Ectoparasitic insects are major disease vectors causing high morbidity and mortality, and people vary in susceptibility to bites and infection (e.g., D. W. Kelly, 2001; Lindsay, Adiamah, Miller, Pleass, & Armstrong, 1993). In some areas of the world, selection pressure from Malaria falciperum is so intense it maintains sickle cell trait, even though, in the homozygous condition, sickle cell anemia is fatal (e.g., Nesse & Williams, 1994). Parasite resistance is a critical feature in the evolution of mate choice, and sexual reproduction itself may have evolved in an arms race against rapidly coevolving pathogens (e.g., Hamilton & Zuk, 1982; Tooby, 1982).
Illness and injury also negatively impact growth. Among Shuar children, elevated CRP is associated with lower growth rates across 1- to 3-week periods. These trade-offs with growth are mediated by body fat reserves mobilized for energy during illness (Urlacher et al., 2014). Immunoglobulin E (IgE) provides a biomarker of past intensity of helminth infection and current infection, and children with higher IgE are shorter. Catchup growth, particularly at the growth spurt, doesn't appear to fully offset these diversions of energy from growth to immune function among the Shuar, but may to some degree in other groups, depending upon age at peak infection (Blackwell et al., 2011).
Individuals vary in susceptibility to illness and accidents due to differences in: (a) developmental and current energetic availability; (b) immune development and function, (c) chemical and behavioral factors affecting exposure to insects that are disease vectors, (d) individual factors associated with disgust sensitivity, risk taking, and coordination (e.g., Cepon-Robins et al., 2013; Mukabana, Takken, Coe, & Knols, 2002). At least some of this variance is heritable. Cues associated with health, phenotypic, and genotypic quality are therefore expected to be attractive across all social value domains. However, optimal level of health and phenotypic quality for any domain will be different. Accordingly, relative preferences for these cues are expected to vary systematically across domains, and also by age, sex, and individual and socioecological conditions that structure the costs and benefits of health for social value. For instance, in large cross-cultural samples, relative strength of mate preferences for good health, physical attractiveness, and youth (amongst females) increased with evolutionarily relevant health risk/pathogen prevalence (e.g., Gangestad & Buss, 1993; Gangestad, Haselton, & Buss, 2006).
This idea has been extended to particular traits hypothesized to be related to health. For example, trade-offs exist between investment in immunity and testosterone, and testosterone appears to have immunosuppressant effects. The immunocompetence handicap hypothesis proposes that traits developing under the influence of testosterone, such as facial masculinity or musculature, are costly signals of underlying genetic quality and immunocompetence. Because the relative value of immunocompetence is greater in high pathogen environments, female preference for facial masculinity was predicted to be positively correlated with pathogen prevalence. Various studies find positive association between degree of female preferences for masculinity in mens faces, voices, and bodies and pathogen prevalence, and between masculinity preference and degree of pathogen but not other kinds of disgust (e.g., DeBruine, Jones, Crawford, Welling, & Little, 2010; DeBruine, Jones, Tybur, Lieberman, & Griskevicius, 2010; B. C. Jones et al., 2013). However, recent data and analysis suggest the testosterone-immunocompetence hypothesis may require reevaluation (e.g., Boothroyd, Scott, Gray, Coombes, & Pound, 2013; Puts, 2010; Scott et al., 2012; Scott et al., 2014) or more nuanced assessment of relative trade-offs (e.g., Schmitt 2014). There may be more direct intrasexual and coalitional competitive benefits to men with masculine features—e.g., musculature, dominance, authority—as well as relative costs and benefits to women of mating with them under various circumstances (e.g., relative cost-benefit structure of men's contributions to subsistence, risks of violence by the male and by other males, desertion, women's reproductive status, and so forth). Other cues may provide more direct indicators of health status, such as skin, hair, oral, movement patterns, or olfactory qualities (e.g., Grammer, Keki, Striebel, Atzmüller, & Fink, 2003; Sugiyama, 2005).
Skin functions in protection, regulation, and sensation. Smooth, unblemished skin indicates less exposure to or damage by parasites and/or disease. Skin condition also provides a window on strength of immune function, indicated by ability to heal without infection (Singh & Bronstad, 1997; Sugiyama, 2004a). Skin damage accumulates with age, such that smooth skin and even skin tone are associated with youth (e.g., Jablonski, 2013). Skin quality also reflects current and chronic nutritional state (e.g., Piccardi & Manissier, 2009). Dandruff can indicate vitamin insufficiency or scalp microbiome imbalance, and psoriasis is a T-cell mediated inflammatory disease linked to immune dysregulation, oxidative stress, and genes of the MHC (e.g., Feng et al., 2009; Nestle, De Meglio, Qin, & Nickoloff, 2009; Tsoi et al., 2012).
Intrapopulation variance in skin color is associated with nutrition, disease, and fertility. For example, hepatitis and anemia can produce a pallid skin cast. In contrast, betacartonoids increase yellowish hue, while cardiovascular efficiency, sexual excitement, anger, and other emotional states are related to greater red coloration. Women tend to have lighter skin than men, probably to increase vitamin D absorption for calcium needs during pregnancy and lactation (Jablonski & Chaplin, 2000). Skin darkens with age, such that comparatively light skin is predicted to be attractive in females, as a cue of youth (Symons, 1979). Women's skin has many estrogen and progesterone receptors, which change skin characteristics over the ovulatory cycle, such as lightening of the skin during the fertile phase, and with age. In addition, lipid secretion, skin thickness, fat deposition, elasticity, hydration, and skin microbiota change over the cycle (Farage, Miller, Berardesca, & Maibach, 2009). Fine lines and wrinkles increase with age, as does unevenness in skin tone.
A growing body of research is testing the longstanding prediction that smoother and relatively lighter skin are related to female sexual attractiveness (e.g., Darwin, 1871; Symons, 1979; van den Berghe & Frost, 1986). Fink, Grammer, and Thornhill (2001) presented subjects with standardized face shapes varying in texture, and found that skin texture significantly influenced attractiveness ratings. Attractiveness ratings of small skin patches are positively correlated with those of the whole face, as well as perceived health and ratings of male facial attractiveness (B. C. Jones, Little, Burt, & Perrett, 2004). Effects of skin color homogeneity and texture independently affect attractiveness. Fink, Grammer, and Matts (2006) had male subjects rate three-dimensional computer models of female faces standardized for shape and skin texture, varying only in skin color homogeneity. Male subjects rated more homogeneous skin color more attractive, healthy, and younger looking.
Multiple variables affect skin color distribution, including melanin, carotenoids, hemoglobin concentration and oxygenated blood, and tanning (Coetzee et al., 2012; Stephen, Coetzee, Smith, & Perrett, 2009; Whitehead, Coetzee, Ozakinci, & Perrett, 2012), and these affect both male and female attractiveness (e.g., Fink, Bunse, Matts, & D'Emiliano, 2012). Matts, Fink, Grammer, and Burquest (2007) used images of skin from cheeks from photos of 170 girls and women from 11 to 76 years of age, and had 353 subjects rate them for attractiveness, healthiness, and youthfulness. Skin homogeneity was positively correlated with perceived attractiveness and youthfulness, and negatively associated with perceived age. Image maps of both hemoglobin and melanin distribution each showed the same results, indicating multiple components of skin homogeneity contribute to this effect. Fink and Matts (2008) used a subsample of images of the women over 40 to examine the relative effects of skin texture and color on attractiveness. From these, they created four sets of images: original, skin texture removed, skin color smoothed (homogenized), and skin texture removed and color smoothed, which subjects rated for age and health. Results indicated that color-smoothed images were rated most healthy, indicating that evenness of skin color is a cue to health. Significant age differences were found among all image sets, with the largest differences found between the original set and the set with texture removed and color smoothed. The latter set was rated youngest, with greatest effect due to texture cues.
Carotenoids, important for protection of tissues and DNA from oxidative stress damage, also play roles in immune activity, increasing cell surface expression of MHC molecules. Because carotenoids accumulate in the skin, yellowness of skin may provide a cue to health and, for this reason, be found attractive. Similarly, redness of the skin is associated with vascularization and oxygenation of blood, and women's estrogen levels. In some nonhuman primates, it is associated with reproductive status, diet, lack of parasites, and immunity, and is also used in sexual signaling. Stephen et al. (2009) suggest that this may explain why redness in human skin is associated with health and found attractive. Using the CIELab color system, they allowed male and female subjects to digitally shift facial image skin color along one or two of three dimensions (light-dark, red-green, and yellow-blue) associated with lightness and the effects of melanin and carotenoids, respectively. As predicted, subjects increased both yellow and red hues of faces to optimize perception of health. Coloration was not increased to extremes, suggesting the existence of optimal levels of these color components. Subjects also enhanced sexual dimorphism in skin color, by lightening women's faces more than men's and increasing redness and yellowness in men's faces more than women's. Similar results hold for South African and Asian subjects (Whitehead, Coetzee, et al., 2012), suggesting that these components of skin color are attractive across populations. Even small changes in frequency of fruit and vegetable intake had measurable effects on carotenoid-related skin color; these changes are detectable and sufficient to affect perceived health and attractiveness (Whitehead, Re, Xiao, Ozakinci, & Perrett, 2012). Studies also suggest that coloration related to carotenoids is relatively more important than melanin in perception of health and attraction (e.g., Stephen, Coetzee, & Perrett, 2011).
Differences in the cue value of these skin features are likely stronger under the dietary, pathogen/parasite, and fertility conditions characteristic of the human EEA. Ideally, future research should be aimed at systematically documenting the relative strength of their cue values to health, age, sex, and reproductive status across conditions more representative of ancestral environments. Examining the relationship between skin tone and biomarkers of health and immune function under these conditions would be particularly instructive: Proposed attraction to healthy skin tone may increase under conditions of dietary and immunological challenge. However, doing so will be challenging, due to lack of uniform lighting at most relevant field sites and the costs of measuring hypothesized correlates such as pathogen prevalence and control variables.
Hair grows at the rate of about one half inch per month, until it falls out upon reaching 2 to 3 feet in length. Energy constraints or illness affect growth rate or hair loss. Hair also gets thinner and dryer with age, as cebum (scalp oil) production is reduced. Graying hair is also associated with reduction in melanocytes and aging-related oxidative stress, and is closely tied to underlying genetics (Commo, Gaillard, & Bernard, 2004). As a result, greying occurs at different times in different populations (Trüeb, 2009). Given the visible effects of aging on hair condition, we would expect male choice for reproductive value and/or peak fertility to be negatively related to gray hair in women. Conversely, female choice for status and productive ability in males, and the later age at which these occur, may be related to female preference for some level of gray hair in males, although association with actual age will vary across populations.
Hair also reflects nutritional and health status. Starvation, nutritional deficiencies, and stress increase hair loss, damage, and fragility, and malnourishment causes observable changes in hair color (Rushton, 2002). Dry, brittle, and/or dull hair is associated with lower dietary keratin, fatty acids, protein, vitamins A, B, folic acid, and minerals such as iron, zinc, calcium, magnesium, and copper (Haneke & Baran, 2011). Iron deficiency, for example, is related to reduction in hair growth and hair loss a few months later (Karadag, Ertugrul, Tutal, & Akin, 2011). Iron is critical to cellular function, and iron withholding may be used as a part of the body's innate immune system. Anemia affects about 2 billion people worldwide, and is associated with fatigue, headaches, low blood pressure, shortness of breath, reduced infection resistance and, over the long term, poorer growth, development, cognitive function and reproductive outcomes (WHO/UNICEF/UNU, 1998). Anemia is of particular concern for women and children in developing countries because pregnancy and rapid growth increase iron requirements. Micronutrient deficiencies are also related to disruption of enzyme function and metabolism regulation (Park, Choi, & Nam, 2009). Zinc deficiencies, for example, are associated with impairments of neurological function (e.g., in autism, depression, and other psychiatric conditions) and impaired immune response and healing (Grønli, Kvamme, Friborg, & Wynn, 2013; Prasad, 2013; Priya & Geetha, 2011; Stechmiller, 2010). Zinc in hair is also negatively correlated with age and with below-normal testosterone levels. Vitamin D deficiency is associated with rickets, and suboptimal levels with a variety of health problems, mood disturbances and poor hair health (e.g., Amor, Rashid, & Mirmirani, 2010; Holick, 2007).
Hair, therefore, provides an observable continuous record of an individual's health, stress and nutrition over a 2- to-3-year period. It also reflects heritable genotypic quality (Etcoff, 1999) and age. Shiny, strong hair provides a cue to recent good health, developmental condition, phenotypic, and genotypic quality. The longer the hair, the longer the record of health. Tellingly, long hair is often preferred across cultures, and long, lustrous hair is often associated with beauty (Etcoff, 1999). Grammar, Fink, Thornhill, Juette, and Runzal (2002) found that hair length was significantly correlated with female attractiveness. In a sample of over 200 women aged 13–73, Hinsz, Matz, and Patience (2001) found that younger, higher-reproductive-value women tended to have longer hair than older women, as predicted if higher-reproductive-value women use their hair as an advertisement of that fact. Interestingly, hair grows fastest among women around the ages of peak fertility (Etcoff, 1999), with the result that evidence of environmental damage has less time to accumulate before new hair grows in, and evidence of health or dietary problems would reflect a shorter period of time.
Body hair and beards are male secondary sex characteristics associated with age and androgen profiles, but have a heritable component and vary across populations (e.g. Dixson & Vasey, 2012; Puts, 2010). Body hair has been hypothesized to be costly to produce and a signal of ability to afford testosterone, as it may entail trade-offs with immune function. Research to date shows women's sexual preference for body hair varies across populations. Dixson, Halliwell, East, Wignarajah, and Anderson (2003) found that male body line drawings with trunk hair were rated as older and more attractive than those without by both British and Sri Lankan women. Women from Cameroon rating shaded photos of male figures found those with body hair sexually attractive, but less so than British women, while women in the United States, China, and New Zealand found male figures without trunk hair most attractive (Dixson, Dixon, Bishop & Parish, 2010; Dixson, Dixson, Morgan, & Anderson, 2007). In Finland, postmenopausal but not premenopausal women found male images with chest hair attractive (Rantala, Pölkki, & Rantala, 2010). Although cross-cultural data is especially welcome and needed, these studies again point to an intrinsic problem in understanding cross-cultural variation: They do not take into account assessment of local trade-offs and sources of interpopulation variance in relative value of qualities associated with higher androgen profiles. Local information such as the relative value to women of male investment, cost of mate desertion or extra-pair copulation, costs and benefits of mate's use of aggression; pathogen prevalence; local range of observable distribution of torso hair; rater's menstrual cycle phase, number of offspring, and sociosexual orientation are needed to make sense of this variation.
Male facial hair also grows in association with androgen surges beginning at male puberty, and continues to cover more area and become thicker with age into adulthood. Beard and eyebrow growth make male lower faces and brows look larger, emphasizing testosterone-related traits (Guthrie, 1970), and may thus signal reproductive maturity, dominance, status or aggressive formidability associated with testosterone levels. Long beards (and long head hair) also provide handles with which to control and harm one's opponent in fights, leading Zahavi and Zahavi (1997) to hypothesize that they provide costly signals of intrasexual competitive ability. A number of studies make it clear that bearded faces are perceived as older, more dominant, and/or more aggressive (e.g., Dixson & Brooks, 2013; Dixson & Vasey, 2012; Neave & Shields, 2008), supporting the claim that facial hair is the product of intrasexual selection. Studies on effect of beards on women's assessments of attractiveness are inconsistent, perhaps due in part to variation in the range of stimuli presented. These inconsistencies may also reflect trade-offs between mating for good genes and ability/willingness to invest, which in turn vary depending on whether the woman is pursuing a short- or long-term mating strategy. Neave and Shields (2008) therefore made three sets of five computer-generated young adult male faces (age 18–25), ranging from clean-shaven to full beard for each. Results from 60 female undergraduate raters showed the predicted positive linear relationships between amount of facial hair and masculinity, dominance, aggression, social maturity, and age. Attractiveness ratings showed a curvilinear relationship with amount of facial hair, with light stubble rated most attractive and full beard rated least attractive. In long-term contexts, light stubble was also the most favored but clean-shaven faces were less favored. For the age range of participants tested, the ability to display signs of dominance in the future (light stubble) trumped actual possession of the signal, probably due to subjects' relatively young age. Obvious next steps are to replicate the study with a wider rater age range.
Developmental instability (DI) results from an individual's inability to buffer development against environmental stressors, and is thought to be negatively related to fitness. Many features of animals' bodies are designed to be bilaterally symmetrical, and although developmental disturbances are expected to affect development on both sides of the body equally, mutational load or homozygosity may increase small random variations from symmetry during development, known as fluctuating asymmetry or FA (e.g., Watson & Thornhill, 1994). FA is therefore hypothesized to be a measure of DI and a proxy for fitness (e.g., Dongen, 2006; Polak, 2003). If so, it provides a potentially assessable cue of phenotypic and genotypic quality (via ability to buffer developmental stress and/or less exposure to developmental disturbances), and an explanation for why symmetrical individuals are found more attractive than less symmetrical individuals. Epigenetic effects of early life stress on hypothalamic–pituitary–adrenal axis (HPAA) regulation may also be related to individual differences in FA, although results are complex and more research is needed (e.g., Flinn, Duncan, et al., 2012). Because maintaining symmetrical development in the face of developmental disturbances is costly, FA has also been hypothesized to be an “honest” (costly) signal of genotypic and phenotypic quality related to a number of fitness-related outcomes. For example, because testosterone exhibits trade-offs with immune function, male display of physical features associated with higher testosterone was thought to amplify signal quality of FA (e.g., Thornhill & Gangestad, 1993), and a similar case has been made for estrogens (Little et al., 2008). Even if low FA provides a measure of an individual's ability to withstand developmental disturbances, low FA preference need not be based on costly signaling per se. FA is a partially heritable trait (Dongen, 2000), since homozygosity and mutational load increase susceptibility to environmental developmental stressors, and since condition-dependent traits will necessarily include some heritable variation (Gangestad & Thornhill, 1999; Rowe & Houle, 1996).
Nonhuman animal research suggests that FA is negatively correlated with a wide range of fitness-related measures of growth, survival, fecundity, intra-sexual competitiveness, and mating success, but results are mixed (Dongen, 2006). In human research, studies report FA associated with health measures, fetal outcomes, psychological outcomes, hormones or morphological correlates of sex hormones (masculine/feminine features), number of sex partners, and attractiveness, but results are also mixed depending on methods, study population, and the hypothesized correlates of fitness measured (Møller, 2006; Thornhill & Gangestad, 2006).
For instance, Thornhill and Gangestad (2006) found facial and body FA positively associated with 3-year retrospective self-reported respiratory infections but not intestinal sickness in a sample of 406 men and women (see also Shackelford & Larsen 1997, 1999). In a sample of over 900 26-year-old men and women, Milne et al. (2003) found FA significantly related to the likelihood of women, but not men, self-reporting two or more medical conditions. Waynforth (1995) found FA associated with higher morbidity among Mayan men in Belize, and Gangestad, Merriman, and Thompson (2010) found oxidative stress positively associated with FA. However, Rhodes et al. (2001) found that facial symmetry did not predict illness in medical records of 294 U.S. 17-year-olds born between 1920 and 1929. In a large study of 4,732 English children, Pound et al. (2014) found no association with FA and longitudinal health measures, but found a very small negative relationship between FA and IQ (see also Banks, Batchelor, & McDaniel, 2010). Other studies also found positive association between FA and psychiatric dissorders (e.g., Arboleda-Florez, Ramcharan, Hreczko, & Fick, 1998, Markow & Wandler, 1986; Mellor, 1992).
FA and cues of phenotypic quality are also reportedly linked with sexually dimorphic traits under developmental regulation by androgens and estrogens. For example, more symmetrical men are reported to have greater musculature (Gangestad & Thornill, 1997), body size (Manning, 1995), grip strength (Fink, Weege, Manning, & Trivers, 2014), testosterone-related facial cues of dominance and reproductive health (Gangestad & Thornhill, 2003), and lower resting metabolic rate (Manning, Koukourakis, & Brodie, 1997) than less symmetrical males. Men with low FA report earlier age of first intercourse, higher numbers of sex partners, higher number of extra-pair copulation partners, and shorter time elapsed until sex with a new partner (Gangestad & Thornhill, 1997; Thornhill & Gangestad, 1994). More symmetrical men also are reported to have more sperm per ejaculate, as well as higher motility (Manning, Scutt, & Lewis-Jones, 1998; Soler et al., 2003). Among Mayan men, FA was associated with lower fecundity, and marginally associated with higher age at first reproduction and fewer lifetime sex partners (Waynforth 1995). Gangestad, Thornhill, and Garver-Apgar (2010) found oxidative stress positively associated with FA and negatively associated with attractiveness and ratings of health and masculinity. Development of these traits is predicted to be condition-dependent: Higher-quality males may be best able to develop and maintain large size, musculature, and high testosterone, and to buffer oxidative stress (Gangestad et al., 2010; Gangestad & Thornhill, 1997, 2003). These masculine traits evolved for intrasexual competition, and may also be the target of female mate choice. A number of studies show that women find masculine traits particularly desirable in short-term mates and extra-pair sex partners (Schmitt, Chapter 11, this volume), and are more attracted to and more likely to have sex with men exhibiting these “masculine” traits during the fertile phase of their ovulatory cycle.
This pattern appears to be the product of facultative mating strategies conditional on a male's relative attractiveness and hormone-mediated sociosexual strategies that covary with FA (Gangestad & Simpson, 2000). This may exacerbate the trade-off women face between good genes and investment. Evidence suggests that female attraction to low-FA males increases with a woman's current fecundity and in short-term (or extra-pair) mating contexts, as does preference for male physical attractiveness and its correlates generally (e.g., Gangestad, Thornhill, & Garver-Apgar, Chapter 14, this volume). Degree of male symmetry predicts a significant amount of their partners' copulatory orgasms (Thornhill, Gangestad, & Comer, 1995), which may bias paternity toward symmetrical males, and women experience more frequent orgasm with extra-pair mates (Thornhill & Gangestad, 2003). Women showed greater preference for symmetrical male faces as a function of probability of conception based on phase of ovulatory cycle (Little, Jones, Burt, & Perrett, 2007; cf. Peters, Simmons, & Rhodes, 2009). Similarly, when presented with T-shirts worn by different men, non-hormonally contracepting women preferred the body scent of more symmetrical men, but only during the fertile times in their cycle; hormonally contracepting women showed no shift (Gangestad & Thornhill, 1998; Rikowski & Grammer, 1999). Score on the sociosexual orientation inventory (SOI) is also associated with female preferences. When 99 women chose between pairs of original and symmetrically manipulated versions of 10 male and 10 female faces, women with higher SOI showed greater preference for more symmetrical faces (Quist et al., 2012).
Studies also report negative correlation between FA and female health and fitness-related variables. Manning (1995) shows an association between body weight and FA in women. Milne et al. (2003) found that female FA was associated with body mass index (BMI) and overall reported number of medical conditions, but not with blood pressure, cholesterol, or cardiorespiratory fitness. This may be due to relatively low levels of environmental stressors in Westernized societies, leading to more homogeneity in FA. For example, among Hadza foragers, FA is higher than in U.S. college students, suggesting that the Hadza experience more developmental stress (Gray & Marlowe, 2002), and more strongly prefer symmetry than a UK sample, and when pregnant or nursing (Little, Apicella, & Marlowe, 2007). More symmetrical women were also found to have earlier age at first birth and more offspring (Manning, Scutt, Whitehouse, & Leinster, 1997; Møller, Soler, & Thornhill, 1995). This may be because attractive women have greater mating opportunities and thus marry earlier, have higher-socioeconomic status mates, and more lifetime offspring. Jasienska, Lipson, Ellison, Thune, and Ziomkiewicz (2006) found more symmetrical Polish women had significantly higher mid-menstrual cycle estradiol levels, indicating significantly greater conception probability, than less symmetrical women, even when controlling for height and BMI. Because estradiol levels in reproductive-age women are related to their size at birth, and FA is hypothesized to be linked with developmental stress, Jasienska et al. (2006) suggest this may form a link between FA and later estradiol levels. This link may provide a direct benefit of preferring symmetrical women as mates. Further, they point out that estrogen levels are linked with many aspects of female health, and that estrogen in premenopausal women actually has an immunostimulant effect, providing potential links between lower FA and female health.
FA is negatively correlated with facial attractiveness ratings of both males and females (e.g., Rhodes, 2006; Rhodes, Louw, & Evangelista, 2009). Most studies on natural variation in facial symmetry show a positive relationship between symmetry and attractiveness (e.g., Sugiyama, 2005). Rhodes et al. (2007) found that most of this association was due to perceived health. They had subjects rate Western and Japanese faces, and found symmetry associated with perceived health, and most effects of symmetry eliminated when perceived health was statistically controlled. B. C. Jones et al. (2001) also found that the relationship between attractiveness and facial symmetry is mediated by the association of symmetry and apparent health, although the direct effect of facial symmetry on attractiveness was small (see also, e.g., Fink, Neave, Manning, & Grammer, 2006). These effects remain even in genetically identical twins (Mealey, Bridgestock, & Townsend, 1999).
If low FA is associated with the ability to withstand developmental disturbance, such that symmetry is correlated with other cues of phenotypic condition, then low FA individuals may be found attractive because of those other cues, in addition to symmetry per se. If so, the link between symmetry and attractiveness would not be direct. Scheib, Gangestad, and Thornhill (1999) found that, when presented with male half-faces (split along the vertical midline), women's attractiveness ratings of half-face images were associated with symmetry of the full face, just as strongly as the women's ratings of the full faces. More symmetrical men had longer lower jaws and more prominent cheekbones, features that appear to reflect developmental influence of testosterone. B. C. Jones et al. (2001) also found that the relationship between attractiveness and facial symmetry is mediated by the association of symmetry and apparent health, while the direct effect of facial symmetry on attractiveness was small.
Body symmetry is also associated with facial symmetry and ratings of attractiveness, health, and fitness, supporting the idea that FA is related to underlying features of phenotypic condition. Thornhill and Gangestad (1994) measured seven nonfacial body traits of 122 undergraduates and found a positive correlation between age at first copulation and degree of asymmetry. They also found negative correlation between FA and self-reported number of lifetime sex partners, even when age, height, ethnicity, marital status, physical attractiveness, and physical anomalies were controlled. FA was important in evaluations of both male and female attractiveness. Gangestad and a naïve research associate measured FA of men from a small village on Dominica using nine different body traits. Both male and female college students rated facial photographs of the more symmetrical men more attractive (Thornhill & Gangestad, 2003). Hume and Montgomerie (2001) studied the relationship between facial attractiveness ratings, FA (based on 22 traits), BMI, health, and age, among male and female subjects, whose attractiveness was then rated by a large number of other men and women. For both males and females, there was a negative association between attractiveness and FA. BMI and past health problems were the best predictors of female attractiveness; for males, it was the socioeconomic status of the environment in which they were raised. However, Hönekopp, Bartholomé, and Jansen (2004) found women's facial attractiveness associated with physical fitness, but that symmetry did not mediate this association, and Tovée, Tasker, and Benson (2000) found no association between FA and bodily attractiveness. Brown et al. (2008) used 3-D body scans of 40 male and 37 females to create 360-degree videos of the body shapes, and had 87 subjects rate them for physical attractiveness. Among male bodies, FA was negatively related to height, shoulder breadth, and torso volume, and positively related to WHR and torso-to-leg length. Among female bodies, FA was positively associated with height and torso volume, but negatively related to WHR and leg length. Principle components analysis (PCA) revealed a single component associated with masculine body shape accounting for 60% of variance in attractiveness, with shoulder breath, torso volume, WHR, and height variables loading positively on this component, and breast size and longer slender legs loading negatively. Lower body FA was less than upper body FA, which the authors suggest may be in part because symmetry is important for locomotor efficiency.
In a recent meta-analysis, Dongen and Gangestad (2011) found 94 studies testing 293 hypothesized relationships between FA and health and disease, reproductive and fetal outcomes, psychological problems, sexually dimorphic hormones or their correlates, and attractiveness. After accounting for publication bias and sample size, they estimate a small but statistically robust association between FA and outcome. Further, they show variation in strength of associations across outcome variables. Considering attractiveness and mate choice, facial but not body FA predicted facial attractiveness, probably because the effect of facial symmetry on attractiveness is direct, whether or not it is the result of DI per se (e.g., Dongen, Cornille, & Lens, 2009; Haufe, 2008). Effects of FA on scent, body, movement, and vocal attractiveness were also apparent, as were associations with health. Associations between FA and correlates of reproductive success (particularly males' number of sexual partners) were also clear, as were effects on maternal risk factors and fetal anomalies. Effect sizes for health-related outcomes were smaller but apparent. Importantly, variation in results across studies could be due to low sample sizes: Obtaining 80% power to detect the effect if present requires a sample size of 350 subjects, well over that included in most studies. A separate meta-analysis found no relationship between FA and morphological correlates of investment in sex hormones related traits (facial masculinity/femininity, digit ratios). However, confidence intervals were wide, so no firm conclusion could be reached either way. Importantly, the largest effect sizes were for those studies that measured hormone profiles directly (Dongen, 2012).
These studies highlight a number of issues relevant to many studies of attractiveness. For instance, many studies use college student subjects and/or western populations, and associations between FA and health outcomes might be greater in evolutionarily relevant populations under higher disease, pathogen, and dietary stress. Most health measures are via self-report, and are limited in scope. Several studies do not measure hormonal profiles, stress, or immune activation directly or longitudinally, when effects of DI on FA are expected to be most salient. These methodological considerations matter because FA is a weak measure of individual differences in developmental instability for multiple reasons (Dongen & Gangestad, 2011). Individuals do not experience the same levels of developmental stress, nor do they experience them at the same times during development (e.g., Blackwell et al., 2011; Urlacher et al., 2014). Traits may vary in how well they are buffered from developmental stability: Those under recent directional selection or sexual selection may be less buffered than functionally critical traits, such as pelvic or leg length symmetry (Clarke, 2003). Associations between FA and fitness-related traits may be apparent only under relatively high stress where lower-quality individuals are significantly challenged, such as we would expect under evolutionarily relevant conditions. However, resource allocation to health trades off against reproduction, and the optimal allocation will be dependent on individual condition. To the degree that, at some point, investment in sexually dimorphic traits associated with reproduction has relatively greater effects on fitness than additional investments in health, individuals in better condition may benefit more from greater investment in reproductive effort than those in poorer condition. This may result in positive correlation between sexually dimorphic traits and fitness, but little to no relationship between sexually dimorphic traits and general health. Under conditions of intense sexual selection, high quality individuals may benefit from investing so much in reproduction that health or survival suffers (Getty, 2002; Kokko et al., 2003; Puts, 2010). Further, different forms of stress may affect DI in unpredictable ways and not in a linear or additive fashion.
Evidence suggests that human female reproductive state, across both the reproductive lifespan and the ovulatory cycle, affects trade-offs in female mate choice which may obscure links between FA and mating preferences if not well controlled for (e.g., Gangestad, Thornhill, & Garver-Apgar, 2005; Sugiyama, 2005). Research on female mate preferences must distinguish between short- and long-term female mating context, subjects' SOI, preferences during fertile and nonfertile phases of the ovulatory cycle, reproductive lifestage, and subjects' own attractiveness. Because hormonal regulation of ovarian function is highly context-sensitive, studies in human biology ideally seek to measure hormonal variation across the cycle to determine individual baseline and variation; future research will need to take this into account. Finally, greater variance in male reproductive success suggests that optimal trade-offs between investment in health and intrasexually selected traits may be lower for men, while trade-offs between health and intersexually selected traits may be higher for women. Finally, similar to the case of the testosterone as costly signaling hypothesis, we must reconsider the claim that FA is primarily a costly signal, rather than a cue to phenotypic and underlying genotypic quality the correlates of which have been under intersexual selection.
Pathogen prevalence and intensity of infection affect allocations to different branches of immunity during development, with consequences in adulthood that vary depending on the timing of health insults (e.g., Blackwell et al., 2011). The number of pathogen antigenic molecules is vast, and changes via pathogen coevolution to immune defenses. The immune system thus includes mechanisms functioning to increase the number of antigens that can be recognized. The major histocompatibility complex (MHC; also called human lymphocyte antigen [HLA] in humans) generates cell surface molecules that bind to specific foreign proteins and present them to Killer T-cells that attack the pathogen directly, or Helper T-cells that signal other systems to coordinate attack. The MHC shows evidence of intense selection for diversity in binding surfaces. It is highly polygenic, highly polymorphic, and codominant. Finally, there is somatic mutation in the genes, which generates additional antigen receptors (Janeway, Travers, Walport, & Schlomchik, 2004).
Selection has produced mechanisms to increase MHC diversity, but a given individual has only a small subset of these alleles. If cues to MHC diversity are assessable, these could be used in mate choice, as well as kin and offspring recognition. Hypothesized mechanisms primarily involve olfactory detection of biochemical cues to MHC, either directly or via its effects on variation in the microbiome of different individuals. Potential benefits of MHC-biased mate choice include inbreeding avoidance, increased offspring heterozygosity, recruitment of rare alleles to counter coevolving pathogens, and increased variability between offspring under conditions of uncertain and changing pathogen pressure (e.g., Brown, 1997; Havlíček & Roberts, 2009; Oliver, Telfer, & Piertney, 2009; Penn & Potts, 1999; Tybur & Gangestad, 2011). However, local selection may favor MHC alleles that bind to antigens of locally prevalent pathogens; hence, there may be selection favoring particular MHC alleles within local populations (Neff & Pitcher, 2005). This strong localized selection may promote mate attraction for MHC similarity, at least at relevant loci. For example, Coetze et al. (2007) found no relationship between HLA heterozygosity and self-reported illness in women, but women with more common alleles reported fewer illnesses and better health than did women with rare alleles. MHC dissimilarity between mother and fetus may also cause problems in pregnancy. Conversely, Lie, Simmons, and Rhodes (2009) found a small but significant relationship between both MHC and non-MHC allelic diversity and self-reported health over a 4-month period. However, some alleles are under stronger selection for diversity than others (Huchard, Baniel, Schliehe-Diecks, & Kappeler, 2013).
In nonhuman organisms, results are complicated, sometimes indicating mate choice for MHC similarity, dissimilarity, or a balance between good genes and heterozygosity (e.g., Bernatchez & Landry, 2003; Piertney & Oliver, 2006). Research indicates that MHC-based choice in humans is similarly complex, although it generally shows a role of MHC in attraction (Havlíček & Roberts, 2009). Sorting out the variation will require good cross-cultural data from evolutionarily valid populations. Wedekind, Seebeck, Bettens, and Paepke (1995) had male subjects wear T-shirts for two nights, and had female subjects rate the shirts for odor intensity, sexiness, and pleasantness. Orally contracepting women preferred MHC-similar men, while noncontracepting women preferred the odor of MHC-dissimilar men. The smell of MHC-dissimilar men also reminded women of their current or previous mates. A follow-up study showed male odor preference for MHC dissimilarity regardless of the sex of the T-shirt wearer (Wedekind & Füri, 1997). Using a similar design, Thornhill et al. (2003) found male preference for both MHC dissimilarity and common alleles. Women's facial attractiveness was positively associated with their scent attractiveness. Men also preferred the scent of women during the fertile phase of the menstrual cycle, based on other scent cues. During the anovulatory phase of their menstrual cycle women preferred the odor of MHC heterozygous men, but during the ovulatory phase they exhibited no MHC preference. There was no relationship between facial attractiveness and MHC heterozygosity. Roberts, Gosling, Carter, and Petrie (2008) also found no odor-based MHC preference during the fertile phase of women's ovulatory cycles. However, single women preferred the odor of MHC-similar men, while women in long-term mateships preferred MHC-dissimilar men. In three studies, women using oral contraceptives preferred the scent of individuals with MHC-similar genotypes (Roberts et al., 2008; Wedekind et al., 1995; Wedekind & Füri, 1997). Using EEG, Pause et al. (2006) found more rapid and intense preattentional brain activity in response to odors of MHC-similar than to MHC-dissimilar individuals. Subjects also rated MHC-similar individuals less attractive as potential partners, suggesting MHC-based incest avoidance. Brain activity patterns for male (frontal lobe) and female (parietal lobe) raters differed when exposed to MHC-similar people of the same sex, suggesting sex differences in MHC odor processing. Jacob, McClintock, Zelano, and Ober (2002) had 49 nulliparous single women choose which of six T-shirts they would “prefer to smell all the time.” Women preferred the smell of males with greater MHC similarity. Women could discriminate odor based on differences in one MHC allele, but only for paternally inherited alleles, leading the authors to suggest that paternal MHC-related odors may be used in offspring-father recognition. This study is often taken as evidence of mate preference for MHC similarity, but subjects were not asked to choose based on mate preference. Ferstl, Eggert, Westphal, Zavazava, and Müller-Ruchholtz (1992) also found evidence for MHC similarity bias among friends, perhaps the result of kin recognition biases expressed in evolutionarily novel circumstances.
A number of studies have examined MHC-biased facial attraction. Roberts, Little, et al. (2005) showed that women preferred faces of MHC-heterozygous men, perhaps assessable via skin quality. Counter to the MHC dissimilarity bias hypothesis, women rated MHC-similar male faces more attractive than MHC-dissimilar faces. Among the Tswana, Coetzee et al. (2007) found no relationship between male ratings of female facial attractiveness or health, and women's heterozygosity or allele frequency (although women with more common alleles reported better health). Coetzee et al. (2012) point out that Tswana practice consanguineal first cousin marriage, which may remove deleterious alleles and decrease heterozygote advantage.
Lie, Simmons, and Rhodes (2010a) had male and female subjects rate opposite-sex facial attractiveness of 80 females and 79 males, respectively. Men rated faces of MHC-dissimilar women more attractive for both short- and long-term mates. In contrast, female raters showed no effect of MHC or non-MHC dissimilarity on ratings of male facial attractiveness. Men found non-MHC genetic diversity attractive in female faces, while women found MHC genetic diversity attractive in male faces for both long- and short-term mating. This provides some evidence for both the genetic dissimilar mating and good genes heterozygosity hypotheses.
Studies of mated couples show mixed results. Among 411 Hutterite couples, Ober et al. (1997) found fewer MHC haplotype matches between spouses than expected based on population genotype frequencies. Ober, Hyslop, Elias, Weitkamp, and Hauck (1998) showed significantly greater fetal loss for MHC-similar couples on individual MHC loci, and even greater loss when couples matched on all 16 loci studied (Ober, 1999). Markow et al. (1993) also show evidence of balancing selection in MHC-biased mating among the Havasupai. However, Hedrick and Black (1997) examined 10 MHC alleles in 194 couples from 11 Indigenous South American groups and found no bias toward MHC-dissimilar mating. Ihara, Aoki, Tokunaga, Takahashi, and Juji (2000) similarly found no evidence of MHC-dissimilar mate preference in Japanese married couples. Chaix, Cao, and Donnelly (2008) found more mate dissimilarity in MHC than in the rest of the genome in 30 Euro-American couples from Utah, but no evidence for MHC-dissimilar mate choice among 30 Yoruba couples from Nigeria. In a study of 145 Australian university students, Lie, Rhodes, and Simmons (2010) found women but not men with higher genetic diversity, both generally and at MHC loci, had greater number of sexual partners.
This pattern of mixed results points to a number of problems with study design. Large samples like the Hutterite are needed to find small but significant effects. However, Markow et al. (1993) did show evidence of MHC-biased mating among the Havasupai, and point out that populations like this are more likely to show evidence of MHC mating bias than large, nonevolutionarily relevant ones. In most studies, only a few alleles were tested, with no a priori predictions about local selection for diversity or “good genes” (e.g., Huchard & Pechouskova, 2014). Only a very large effect size could have been detected in the Indigenous South American sample because only a few alleles were studied across several populations, each with a small sample size, without controlling for degree of inbreeding (Beauchamp & Yamazaki, 1997; Penn & Potts, 1999). The Yoruba have a long precolonial history of monarchy and gerontocracy, and the sample size for the study was only 30. Finally, most studies used marriage as a proxy for sexual preference rather than actual mating behavior.
Garver-Apgar, Gangestad, Thornhill, Miller, and Olp (2006) looked for more subtle evidence of MHC-biased mating preference among 48 romantic couples. They found a negative association between number of MHC alleles a couple shared and women's sexual responsiveness and satisfaction with how much their partner aroused them. Further, MHC similarity was positively correlated with women's number of extrapair partners while with her current partner, but not previous ones. Conversely, MHC similarity was not associated with nonsexual aspects of relationship satisfaction. Men showed no relationship between MHC similarity to partner and sexual responsivity or arousal to her, or to extra-pair partners.
Evidence suggests that MHC-related odors and faces affect preferences, but methodologies are diverse and wording of preference questions is often equivocal (Havlíček & Roberts, 2009). Consequently, results are mixed. Genererally, MHC similarity appears to down regulate sexual attraction but may upregulate nonsexual attraction, while dissimilarity is associated with sexual attraction. Future studies will require larger sample sizes and a greater number of targets. Questions regarding sexual attraction and/or behavior must be more explicit, and must determine whether MHC similarity is related to kin selection. Both MHC and non-MHC alleles should be tested, and analyzed to determine whether they are under selection for diversity or commonality. Tests should examine choice for dissimilarity and diversity. Sociosexual orientation and, for female subjects, phase of menstrual cycle and oral contraceptive use should be included as variables. Systematic inclusion of these variables will go a long way toward determining how MHC affects mate choice, kin recognition, and kin selection.
Sexually dimorphic traits provide cues to the relative social value of both men and women, although the cues associated with each sex are expected to differ in certain predictable ways. Different morphological traits may be associated with higher or lower social value in a given domain, and be more or less important depending on local context. Our attractiveness-assessment psychology is thus expected to generate different assessments of these traits based on local environmental features. For both men and women, the trade-off between additional growth and reproduction can have significant effects on lifetime fitness. The optimal trade-off point can be formally modeled and tested, and is expected to be affected by variables such as diet, workload, stress, birth spacing, and age-specific extrinsic mortality risk (Hill & Hurtado, 1996). Because individuals and their condition vary, some may invest less, more, or the optimal amount in growth prior to reproduction. This opens the door for the evolution of attractiveness-assessment mechanisms that use environmental and social cues to arrive at local preferences for various sexually dimorphic traits, for example height, body size, and muscle development and fat distribution.
Because mammalian males have higher variance in reproductive success than females, intrasexual competition is typically higher among males, and males correspondingly larger or possessing of armaments. If females preferentially mate with more formidable males, selection for these traits is enhanced. Human sexual dimorphism in muscle and fat composition is much greater than overall dimorphism in body size suggests. Men have 75% more arm muscle mass, 50% more leg muscle mass, and 61% more total muscle mass than women, as well as 90% greater upper body strength and 60% greater lower body strength (Lassek & Gaulin, 2009), indicating strong intrasexual selection for musculature (Puts, 2010; Sell, Hone, & Pound, 2012). Conversely, women have substantially higher fat stores than do men, much of it around the gluteofemoral region and breasts. This offsets much of the dimorphism in overall body mass, and may be one reason that Marlowe (2012) found the human operational sex ratio high compared to that expected by overall body mass.
The importance of these traits is reflected in adaptations for assessing men's relative strength and fighting ability. In both U.S. undergraduates and Tsimane forager horticulturalists, perceived strength based on body, face and vocal characteristics are all highly correlated with actual strength, controlling for height and weight (Sell, Cosmides, et al., 2009; Sell et al., 2010). Strength cues in men include upper body musculature, v-shaped torso, facial width to height (fWHR) associated with width of the bizygomatic structure (e.g., Windhager, Schaefer, & Fink, 2011; Zilioli et al., 2014), and low fundamental frequency and other characteristics of the voice (e.g., Hodges-Simeon, Gaulin, & Puts, 2011; Hodges-Simeon, Gurven, Puts, & Gaulin, 2014; Puts, Apicella, & Cárdenas, 2011). Controlling for age, marital status, and body mass index (BMI), fat-free mass (FFM) and limb muscle volume (LMV) also predict men's number of self-reported sex partners, and lower age at first intercourse among men in the NHANES sample (Lassek & Gaulin 2009; see also Gallup, White, & Gallup, 2007). Strength is associated with fighting ability, and better fighters more readily anger, use aggression, and feel entitled to getting better treatment from others (Archer & Thanzami, 2007; Hess, Helfrecht, Hagen, Sell, & Hewlett, 2010; Petersen, Sznycer, Sell, Cosmides, & Tooby, 2013; Sell et al., 2010; Sell et al., 2012; Sell, Tooby, et al., 2009).
Although among male foragers, strength peaks earlier in the lifespan than hunting return rates, body size and strength appear to play a role in productive ability. For example, among the Ache, Hadza, and Tsimane, body size was related to bow and arrow shooting accuracy (Blurton Jones & Marlowe, 2002; Gurven et al., 2006; Walker et al., 2002). Among the Hadza, upper body strength is the most consistent predictor of hunting ability, and men with stronger bodies had highest reproductive success (Apicella, 2014; see also Gurven & von Rueden, 2006).
Because caloric needs increase significantly with muscle mass (Lassek & Gaulin, 2009), too much musculature can be overly costly. As predicted there is an inverted U shaped relationship between musculature and sexual attractiveness among a U.S. sample of women (e.g., Fredrick & Haselton, 2007). In humans, higher testosterone levels scaffolding development and maintenance of greater muscle mass appear to trade off against immunity; however, men in better phenotypic condition may be better able to afford both, so this phenotypic correlation may mask the trade-off. For example, Rentala et al. (2012) found higher testosterone associated with greater facial attractiveness and greater response to hepatitis B vaccination (mediated by cortisol levels). In the NHANES III, more muscular men had lower CRP levels and white blood cell counts, suggesting a trade-off of greater testosterone (Lassek & Gaulin, 2009). This can be interpreted as support for the testosterone-as-immunological-handicap hypothesis (Hamilton & Zuk, 1982). However, CRP is an acute phase reactant that, in the absence of infection, should be near zero in the blood; thus, because no chronic elevation in CRP is found under evolutionarily relevant high-pathogen conditions (Blackwell et al., 2010; McDade et al., 2012), it is possible that males with greater Fat FM and VFM in the NHANES sample were either less prone to be infected at the time of measurement, or less likely to have chronic inflammation, reflecting better health. Similarly, white blood count increases in response to infection. Because patterns of pathogen immune development and activation and energy availability differ in foraging populations, more direct research is needed to disentangle relationships between investment in muscle mass and immunity.
In primates with multi-male/multi-female groups, males may form coalitions to prevent solitary males or other coalitions from gaining sexual access to group females. For humans, having larger, stronger, more aggressively formidable allies is likely beneficial in coalitional aggression and cooperative work. However, coalition and cooperative partners are also potential rivals in the contest to mate with female group members: The more formidable one's allies, the more formidable one's potential intrasexual competitors. More muscular men are not only more likely to have extra-pair mates, and more of them, but also are more likely to be the extra-pair mates of women, and to be found sexually attractive (e.g., Fredrick & Haselton, 2007). And, stronger males are more likely to have deployed physical force and to approve of its use in the service of their interests (Sell et al., 2009). In response to this problem, men appear to have evolved adaptations regulating competition between close allies. Men's testosterone rises in response to competitive wins against outside coalitions, but not in response to wins against friends. Further, men's testosterone usually increases in the presence of fertile women, but declines when the woman is one's friend's mate (Flinn et al., 2012).
Females face the risk that males will use their size and strength advantage coercively. For men, the costs of short-term mating can sometimes be reduced by choosing mates who exhibit cues of exploitability (Buss & Duntly, 2008; Goetz, Easton, Lewis, & Buss, 2012). This may help explain male attraction to cues of helplessness or low power in females. One solution to this problem is for females to obtain physical protection from other males (e.g., Buss & Schmitt, 1993; Scalise Sugiyama, 2014). Women who preferred larger, stronger, more dominant men as sires for their offspring would gain both indirect and direct benefits of alliance with those men, including sons and daughters (Cashdan, 2008) who inherited these qualities. On the other hand, females who preferred males exhibiting ability and willingness to invest in their offspring would tend to rear more offspring to maturity. The costs entailed by each of these preferences are affected by local resource constraints, intensity of intra- versus intergroup conflict, and operational sex ratio, and are mediated by the risks of domestic violence and desertion. In humans, adaptations reducing mating conflict within male coalitions could reduce the intensity of intragroup mating conflict, particularly when intergroup conflict is high (Flinn et al., 2012). Formal modeling of these trade-offs is necessary to predict evolutionarily stable mixes of strategies within specific constraints, including relative importance of testosterone-mediated traits in intrasexual competition vs. female mate choice (e.g., Puts et al., 2011). Increasingly, evidence suggests the nuanced distinctions among the constellation and degree of male androgen-linked traits found dominant and aggressive and those found sexually attractive may differ somewhat (e.g., Blackwell & Sugiyama, 2008; Hodges-Simeon et al., 2011; Windhager et al., 2011), and may differ in degree to which they are preferred across cultures (Scott et al., 2014).
Throughout the juvenile period, individuals face a number of trade-offs, including among basal metabolism, activity, immune function, and growth. Adult height is partially heritable, but nutrition, pathogen exposure, and immune function affect how much energy is allocated for growth. In subsistence societies larger males are those who had better nutrition, fewer parasites and illness, less psychosocial stress, and/or more efficient metabolism than smaller males. As noted though, weight (primarily muscle in males and fat in females) is prioritized over height (e.g., Blackwell et al., 2009; Urlacher et al., 2014).
Further, for females, human growth is determinate: Longitudinal growth ends when reproduction begins because the energetic costs of doing both simultaneously are too high (e.g., Walker et al., 2006). For women, the fitness benefit of additional growth prior to reproduction includes accumulation of somatic resources for later reproductive effort (Jousilahti, Tuomilehto, Vartiainen, Eriksson, & Puska, 2000), lower offspring mortality (Allal, Sear, Prentice, & Mace, 2004), and lower maternal and infant mortality. Women with more gluteofemoral fat have higher fertility, and taller women tend to have wider pelvises, easier births, and higher infant birth weights (Kirchengast, Hartmann, Schweppe, & Husslein, 1998; Martorell, Delgado, Valverde, & Klein, 1981; Rosenberg, 1992; Rosenberg & Trevathan, 2002). The potential benefits of earlier reproduction include lower prereproductive mortality risk and a potentially longer timespan in which to reproduce, so females in high mortality, resource-constrained populations show rapid growth for adult body size (Hill & Hurtado, 1996; Walker et al., 2006).
Adult height is affected by heritable factors and life history trade-offs associated with level and timing of developmental resource access, metabolic efficiency, mortality risk and pathogen exposure (e.g., Blackwell et al., 2010; Urlacher et al., 2014, Walker et al., 2006). Greater height offers various biomechanical advantages (Samaras, 2007). In sports, for example, taller elite athletes have an advantage in middle-distance running, swimming, and jumping. Disadvantages of greater height are apparent in the total energy required by taller people, aerobic activity in which maximal performance occurs for 30 seconds or more, and heat regulation and dehydration under heavy work load. Depending on local ecology, greater height can also make travel much more difficult: For example, tall individuals face more obstacles (e.g., low vines, branches) than shorter individuals in tropical rainforest foraging (e.g., Hill & Hurtado, 1996).
Although male strength has a larger effect, height also increases perceived fighting ability, which may underlie ability to get one's way (Sell et al., 2009). Height is positively associated with strength and reach, and may be correlated with actual fighting ability, although results are mixed (e.g., Carrier, 2011; von Rueden et al., 2008; Sell et al., 2012). Taller males report engaging in more frequent aggressive acts (Archer & Thanzami, 2007). Taller individuals self-report higher self-esteem, engage in and are perceived to engage in more dominant behavior, are less sensitive to dominance cues in other men, are perceived as being more intelligent, and were more influential in a negotiation experiment (e.g., Gawley, Perks, & Curtis, 2009; Judge & Cable, 2004; Watkins et al., 2010). Taller men also report less jealousy in response to dominant rivals than do shorter men (Buunk et al., 2008). However, because strength is a more powerful predictor of perceived fighting ability, entitlement, positive attitude toward and actual use of force, and deployment of anger (Sell et al., 2012), associations between these outcomes and height must ultimately be revisited to control for effects of strength.
In mates, both men and women generally prefer that the man be taller than the woman, and men tend to have partners who are shorter than themselves (e.g., Courtiol, Raymond, Godelle, & Ferdy, 2010; Salska et al., 2008). Height is associated with rated attractiveness of men, and women more strongly prefer relatively taller men to themselves during the fertile (follicular) phase of their menstrual cycle (e.g., Pawlowski & Jasienska, 2005). Taller-than-average men are preferred to men of short or average stature as dates and mating partners in questionnaire studies (e.g., Buss & Schmitt, 1993; Fink, Neave, Brewer, & Pawlowski, 2007), have more attractive mates (Feingold, 1982), and are more likely to be married (Pawlowski et al., 2000). In analyses of personal ads, 80% of women who stated height preferences wanted men 6 feet tall or taller (Salska et al., 2008). Ads placed by taller men receive more responses (Pawlowski & Koziel, 2002), and taller men were rated more desirable in the context of speed-dating (Kurzban & Weeden, 2005). Women even seem to take height into consideration in sperm donors (Scheib, 1997; Scheib, Kristiansen, & Wara, 1997). Sear and Marlowe (2009), however, report no male height bias in marriage partners among Hadza foragers. However, both extreme shortness and tallness may be associated with health problems in both sexes (Mueller & Mazur, 2001; Nettle, 2002a).
Associations between male height and reproductive success are mixed. A review by Stulp, Pollet, Verhulst, and Buunk (2012) found studies that reported no effect, positive effects, and negative effects of height on RS; however, the authors note that many of those studies were not based on men's entire reproductive career and did not test for curvilinear effects of height on RS. Curvilinear effects have been found in several studies of male height and RS. Stulp et al. (2012) found that average-height men married earlier, which likely accounted for their greater reproductive success. However, they note that studies (including their own) do not account for reproduction in the context of extra-pair or nonmarital relationships. This may account for additional reproductive success of taller men in particular. Relatively taller men also have numerical advantages in number of potentially accepting mates in the mating market (Pawlowski et al., 2000; Pawlowski & Koziel, 2002).
Height preferences are not limited to the sphere of mate selection. Coalitionary leadership and height appear to be associated in both small-scale (Brown, 1991) and state societies (Stulp, Buunk, Verhulst, & Pollet, 2013), although some of this effect is likely a product of strength. For example, in addition to other factors such as generosity, von Rueden et al. (2014) found strength but not height associated with leadership in two Tsimane communities. In U.S. presidential elections, the taller candidate is more likely to win, with the margin of victory positively correlated with height (McCann, 2001; Stulp et al., 2013). Senators and CEOs appear to be taller than the average American man (Etcoff, 1999; Keyes, 1980). Further, there appears to be a positive association between height and socioeconomic and social success in modern societies (e.g., Bielicki & Szklarska, 1999; Deaton & Arora, 2009). Interestingly, Mueller and Mazur (2001) found no relationship between height and either status (final military rank) or socioeconomic success among a sample of the West Point graduating class of 1950, even though they did find a significant indirect effect of stature on lifetime RS. However, the West Point cohort is more homogeneous, both in height and in determinants of success that may covary with height, than the general population.
The use of industrialized populations to study height preferences is problematic on several counts. For example, mate selection accounts for higher reproductive success among taller men in the West Point sample, who had higher probability of having a second family with a younger, fecund, wife. The authors conclude that directional mate selection for height appears to be unconstrained in this sample. However, military officers are extremely unlikely to have experienced the level of dietary and health constraints predicted to trade off against height that almost certainly affected our foraging ancestors. Another study found that taller-than-average British men had higher numbers of live-in partners, and lower chance of either being childless or having had no significant mating relationship (Nettle, 2002a), but no significant association between total number of offspring and height. However, the men had not yet completed fertility, and had ready access to contraceptives. In modern societies with ready access to birth control, number of sexual partners may be a better indicator of the links between preferences and reproductive success under ancestral conditions, than reproductive success per se.
If male size is positively associated with aggressive formidability yet involves costs, we may expect selection for a context-sensitive assessment mechanism functioning such that intensity of male height and strength preferences increase with increasing levels of intrasexual competition. Intensity of preference for taller males is also expected to vary with resource stress: although taller males are those who could better afford the costs of growing larger and relative height provides a signal of developmental phenotypic quality. However, at some level of resource constraints the energetic and mortality costs of maintaining large size may outweigh the benefits of signaling phenotypic quality. This accords with findings of an inverted U-shaped function between height and attractiveness. Since hominid evolution no doubt included periods of extreme resource scarcity, height assessment adaptations might well reflect this trade-off.
Findings on men's preferences for partner height run the gamut from below-average to above-average height (e.g., Grammer et al., 2002; Hensley, 1994; Swami et al., 2008). Again, this may be because there is an inverted U-shaped relationship between male preference and female height, with males preferring slightly taller than average females who are nevertheless shorter than the male assessor (Courtiol, Picq, Godelle, Raymond, & Ferdy, 2010; Courtiol, Raymond, et al., 2010). Male preference for partner height varies as a function of the assessor's own height (e.g., Fink et al., 2007; Salska et al., 2008; Swami et al., 2008). For example, in a large Polish sample, preferred difference between ego's height and partner height was affected in part by the rater's own height, with taller men and shorter women preferring greater height differential in their partners, thus increasing effective mating pool size (Pawlowski, 2003). However, in the fertile phase of their menstrual cycle, and for short-term mateships, women exhibited greater preference for taller men, independent of the rater's own height (Pawlowski & Jasienska, 2005). Below-average or average height women are reported to have greater RS than tall women generally (e.g., Mueller, 1979; Nettle, 2002b), although there is variation across populations (e.g., Stulp et al., 2012).
In a natural fertility population of Gambian women, Sear (2010) found the expected trade-off between growth and age of sexual maturity: Taller women had later age at first birth, but their offspring exhibited lower mortality. The study found no evidence of a relationship between female height and marriage patterns, divorce, or spouse's height. Higher mortality was observed at both ends of the female adult height continuum, but not enough to negate the positive relationship between height and reproductive success. Nor was there evidence of positive associative mating for height, which suggests that female height was not a significant factor in men's marriage arrangements in this population. However, the authors note that men's preferences per se were not tested, and the relative benefit of choosing taller women as wives may be offset by desire for quantity of mates in this polygynous society.
Using data from Britain's National Child Development Study, Nettle (2002b) found a weak but highly significant inverted U-shaped relationship between relative female height at age 23 and reproductive success at age 42, controlling for own or husband's socioeconomic status. Highest reproductive success was for women between .7 and 1.7 standard deviations below the mean. Women of mean height had the highest number of marriages or long-term mates, and were least likely never to have had a long-term mating relationship. Nettle also found the expected trade-off between growth and age of sexual maturity, with taller women beginning to reproduce later. However, the sample population had ready access to hormonal contraceptives, and mean fertility was low for all heights observed, so later first reproduction of taller women cannot account for their lower reproductive success. As predicted, preferred female height appears to change with (mild) socioecological risk. Pettijohn and Jungeberg (2004) found a significant positive correlation between yearly indicators of economic stress (predicted to covary with perceived ecological risk) and the height of Playboy Playmates of the Year.
Singh (1993a, 1993b) noted that the ratio of waist to hip circumference (waist-to-hip ratio or WHR) provides a potential cue to female mate value. Estrogen during puberty stimulates fat deposition on the thighs, hips, and buttocks, and is associated with the widening of the female pelvis. Androgen profiles lead to male fat deposition in the abdominal region. The result is postpubertal sex differences in WHR, with a normal WHR for Western women of ∼.7 and a normal male WHR of ∼.9 (Singh, 1993a). Western women with normal WHR (.67–.80) are at reduced risk for primary infertility and certain health problems (e.g., cardiovascular disease, stroke, diabetes, female carcinoma), independent of overall level of body fat (Singh, 1993). However, except for primary infertility, most health risks associated with higher female WHR are probably evolutionarily novel (Lassek & Gaulin, 2008; Sugiyama, 1996, 2005). In clinical studies, women with low WHR have significantly higher fecundity but as Lassek and Gaulin (2008) point out, clinical studies are largely based on older women trying to conceive. In a well-nourished sample of Polish farmers, Jasienska, Ziomkiewicz, Ellison, Lipson, and Thune (2004) found that women with low WHR and large breasts were the most fecund quartile in their sample and had estradiol levels suggesting a conception probability 3 times that of the rest of the sample.
Singh (1993a, 1993b) proposed that selection shaped men's mating psychology to prefer female WHR of ∼.7, and women's mating psychology to prefer male WHR of .9 regardless of preferences for overall body fat. In females, WHR provides potential cues to sex, lifestage, parity, and pregnancy. Significant evidence for WHR-associated assessment and preference psychology is found in studies using line drawings, standardized body photos, eye direction detection, and archival materials, across both time and cultures.
Only three studies report no effect of WHR on attractiveness. Yu and Shepard (1998) and Wetsman and Marlowe (1999) used a subset of Singh's (1993) 12 line drawings of female figures varying in three levels of body weight (high, medium, and low), and four levels of WHR (.7, .8, .9, 1.0). This subset presented WHRs of .7 and .9 only. A follow-up Hadza study (Marlowe & Wetsman, 2001) used a wider range of WHRs from .4 to 1.0. In all cases men preferred heavier-weight figures, with no apparent effect of WHR. However, males in these societies showed clear preference for higher-body-weight figures, and potential stimulus confounds between WHR and body weight mean that effects of preference for higher body weight could have swamped WHR preferences. Further, in the Matsiguinga case, the highest WHR presented was average for the population. Marlowe, Apicella, and Reed (2005) later showed preference for lower WHR when figures were presented in side view, with buttocks extension visible. Sugiyama (1996, 2004b) also found that Shiwiar forager-horticulturalists preferred heavier-weight figures, but that a preference for lower WHR was apparent when weight was better controlled and high and low WHR were classified in relation to the female population average (see Sugiyama, 2005, for discussion).
Some researchers contend that WHR accounted for very little of the variance in bodily attractiveness, and was primarily a by-product of preferences for body mass index (BMI: weight kg/height meters2; e.g., Cornelissen, Tovée, & Bateson, 2009; Tovée, Maisey, Emery, & Cornelissen, 1999). However, the by-product view lacks surface validity and has not stood up to scrutiny. At 6′3″ and 185 pounds, Prince William has essentially the same BMI as UFC women's bantamweight champion and Sports Illustrated swimsuit model Ronda Rousey at 5′6″, 134 pounds (fighting weight): Look at their bodies and see if they are equally sexually attractive to you. For the vast majority of readers, I'll bet not. Studies using photos of actual people presented only a small section of the evolutionarily valid range of WHR relevant to mate value. Primarily, they presented figures of one sex, with primary and/or secondary sexual characteristics visible: For instance, females shown were primarily nulliparous or low-parity women with mean age around peak fertility. This greatly limits the WHR variation presented to subjects (Sugiyama, 2005). I predicted that because WHR varies across populations, instead of uniform cross-cultural preference for a specific WHR, lower WHR relative to the normal female range to which a man is exposed should be preferred (Sugiyama, 1996, 2004b). Further, men exposed to a higher range of healthy nubile female WHR should find higher WHR more acceptable than men exposed to a lower range of female WHR, and lowering the natural range of WHR to which men are exposed should predictably lower their expressed WHR preference, at least within the limits of the reaction norm for these adaptations (Sugiyama, 1996, 2004b, 2005).
Additional research now supports this general prediction, originally proposed by Symons (1979), that attractiveness assessment is calibrated by local range of variation in the cue in relation to local optima as one moves across different ecologies, groups, or subgroups (e.g., Kościński, 2008, 2012; Tovée, Swami, Furnham, & Mangalparsad, 2006). Although earlier studies showed that Playboy models and film actresses tended to have WHRs of ∼.68), there is variation in the absolute WHR preferred. For example, Voracek and Fisher (2002) show decrease in WHR of Playboy models through time, and women in Reubens paintings have average WHRs of .78 (Swami, Gray, & Furnham, 2007). For example, Pettijohn and Jungeberg (2004), show change in body shape correlated with times of socioeconomic hardship, when playmates of the year were heavier and had larger waists and WHR.
Preference is also regulated in relation to options. This same principle is illustrated in the “closing time effect” (whereby standards of attractiveness decrease but perceived attractiveness increases as bar closing time approaches, regardless of alcohol intake), and by the negative effects of viewing highly attractive females on male relationship satisfaction and on female body image, respectively (e.g., Kenrick, 2011). Nevertheless, preference for relatively lower female WHR is apparent even without past visual exposure: Both men born blind and those who developed blindness later in life preferred the shape of mannequins with lower WHR (Karremans et al., 2010).
I (2005) argued that because multiple anatomical features are associated with waist and hip circumference, the precise shape receptor, assessment, and preference mechanisms might not consist of a waist-to-hip circumference assessor per se. Rather, they are likely based on more complex shape-assessment mechanisms, such as curve detectors, angle detectors, or the ratio itself. WHR is composed of numerous shape dimensions, including factors associated with skeletal functional morphology and body fat deposition. I therefore predicted that WHR-associated attractiveness assessment should take as input the observable range of female WHR and body fat, based on adaptations that incorporate assessment of critical WHR subcomponents associated with sex differences in functional anatomy, including: (a) pelvic width, shape, and angle, (b) hip width and circumference, (c) hip shape, (d) buttocks extension, (e) buttocks shape, (f) waist width and circumference, (g) waist shape, (h) stomach shape, and (i) stomach extension in relation to (j) other aspects of skeletal structure—for example, shoulder and/or ribcage width, distance from pelvis to shoulder, and length of long bones (which provide reference points for assessing pelvic width and fat deposition)—in relation to overall growth, developmental health, and biomechanical efficiency (Sugiyama, 1996, 2004, 2005). This section updates this information, and discusses new findings in relation to predictions or issues I raised previously.
One major development comes from Lassek and Gaulin (2008), who note that gluteofemoral fat (GFF) deposition is prioritized in women, with most fat composed of GFF, a pattern not seen in other primates. GFF is richer in long-chain polyunsaturated fatty acids (LCPUFAs) than abdominal and visceral fat, and the primary source of LCPUFAs necessary for fetal and infant brain development. GFF is protected from use until peak infant brain growth late in pregnancy and during lactation, even when women are under conditions of food restriction. Conversely, abdominal fat is prioritized for mobilization for short-term energy use and may decrease availability of LCPUFAs. The primary LCPUFAs in our brains are arachidonic acid (ARA) and omega-3 docosahexaenoic acid (DHA), with about 20% of brain weight comprised of DHA. Studies of mothers' milk, and of DHA supplementation and dietary intake, show improved cognitive performance in humans and nonhuman animals (Cohen, Bellinger, Connor, & Shaywitz, 2005; Koletzko et al., 2008; Lassek & Gaulin, 2014; McCann & Ames 2005). Further, gluteofemoral fat stored early in life is not replenished, so declines with parity, as does blood-circulating DHA, and some studies show that cognitive performance of offspring declines with birth order. Conversely WHR increases with parity. Lassek and Gaulin (2008) therefore predicted that women's WHR would therefore be negatively associated with own and offsprings' cognitive abilities. They further predicted that, because women who reproduce while they are still developing face competing demands from their own brain development, gluteofemoral fat storage, and their fetus' brain development, teen mothers and their offspring would have impaired cognitive development, but that this would be buffered in women with low WHR (i.e., high LCPUFA stores). Analysis of data from the U.S. Centers for Disease Control and Prevention National Health and Nutrition Examination Survey (NHANES) supported each of these predictions. Their working hypothesis is that, as selection acted to increase brain size in humans, female adaptations to support the increasing costs of providing resources for neurodevelopment included the acquisition, storage, and allocation of LCPUFAS to offspring. Those resources came to be stored in the gluteofemoral region, and thus came to be the targets of male mate choice. WHR may be a cue to the LCPUFAS resources a woman has available for fetal and infant brain development. Interestingly, DHA is reported to be particularly concentrated in the prefrontal cortex, an area of the brain hypothesized to have been the area most enlarged in later hominin brain expansion (e.g., Crawford et al., 1999; Van Essen & Dierker, 2007) and important for short-term working memory and association.
A related development is research showing that, under dietary energy constraints more characteristic of our evolutionary past, women face more noticeable trade-offs between gluteofemoral fat deposition for reproduction and abdominal fat deposition for energy mobilization to buffer food shortages and other environmental stressors. Trade-offs are in part regulated by steroid hormones, including cortisol, estrogens, and androgens. Stress activates the HPA axis and cortisol production, which mobilizes energy stores to deal with sources of stress. It also shifts allocation to storage in central adipositity in preparation for future stressors, and is therefore associated with higher WHR (Cashdan, 2008; Flinn & Ward, Chapter 24, this volume).
A possible evolutionary scenario is that as increasingly dimorphic fat deposition arose with later hominin increases in brain size, it provided cues to multiple components of female mate value, leading to evolution of WHR assessment and preference mechanisms that generated or enhanced it as a target of male mate choice. Deposition of fat on the gluteofemoral region may have initially been driven by preexisting male primate sexual attention to and assessment of this region in response to proceptive female primate sexual displays, and/or the energetic and structural efficiency of storing fat in this region, particularly during pregnancy. Moreover, as selection for gluteofemoral fat deposition increased to support brain development, and trade-offs between somatic and reproductive investment directed the timing of this deposition toward puberty, WHR provided an important cue to sex, onset of female reproductive lifestage, and parity, which intensified selection pressure on WHR-assessment mechanisms.
As GFF deposition evolved, WHR increasingly provided a cue to developmental markers of reproductive value. In a sample of 329 Shuar and Shiwiar forager-horticulturalists, female WHR decreases linearly from a high of almost 1.05 at 2 years of age, to an average low of 8.5 by around 12 years of age (Sugiyama & Blackwell, 2008). Change over this period reflects early prioritization of body fat deposition in childhood to buffer energetic fluctuation and trade-offs between basal metabolic needs, growth, immune function, and activity, transitioning to a major increase in reproductive investment of gynoidal fat distribution in girls near and at puberty (e.g., Ellison, 2001; Lassek & Gaulin, 2008). Despite earlier studies suggesting that critical levels of body fat stimulated onset of menarche (e.g., Frisch & McArthur, 1974), human biologists now widely accept that menarche is related to skeletal maturation and not total body fat (Ellison, 2001). However, using NHANES data, Lassek and Gaulin (2007, 2008) found that odds of menarche increase with gluteofemoral fat more than with height or biiliac breadth, and decrease with larger waist circumference. Among Shuar, the intersection of low female WHR with increasing body fat converges on age of peak female reproductive value (Sugiyama & Blackwell, 2008). Further, among well-nourished Polish farmers, women with low WHR and large breasts had estradiol levels associated with a conception probability 3 times greater than that of other women (Jasienska et al., 2004).
WHR also provides a reliable indicator of sex post-pubertally. Even though the range of Shuar female WHR is significantly higher than that of western industrialized populations, Shuar show significant differences in WHR by sex (Sugiyama & Blackwell, 2008). Experimental studies in Western/industrialized populations using standardized photos of 18- to 42-year-old western women showed that BMI (and WHR co-vary, with BMI accounting for over 80% of the variance in attractiveness ratings, while WHR accounted for less than 2%). Tovée et al. (1999), concluded that WHR preference was a by-product of BMI preference (see, e.g., Tovée & Cornelissen, 2001; Tovée, Hancock, Mahmoodi, Singleton, & Cornelissen, 2002). However, Shuar BMI does not differ by sex, and shows similar age-related change for both sexes. Typical of humans, adolescent and post-adolescent females are shorter and have more body fat, while males are taller and have more muscle. This contributes to very different body shapes even though age-related BMI does not differ: Male and female preferences based only on BMI would find male and female body shapes equally sexually attractive. This is not the case. Further, Shuar WHR and BMI are not correlated for either sex, so WHR preferences cannot be a by-product of preferences for BMI, unless the Shuar (a natural fertility, subsistence population) rather than Westerners are an evolutionary anomaly. Cashdan (2008) shows that across 33 non-Western populations (including the Shuar), WHR is above 0.8 in almost all populations, with high variability across them. Further, in populations without obesity, there is no correlation between BMI and WHR. After reviewing a large (n = 32,000) international (19 countries) sample from the WHO MONICA study, Cashdan notes that BMI could possibly account for only 18% of variance in female WHR, and that only ∼30% of variance in female WHR could be explained even after taking into account height, age, BMI, and population. Most of the association between BMI and WHR was due to the contribution of evolutionarily invalid obese populations (about half the sample). While preference for body fat levels is an important contributor to attractiveness, varies cross culturally, and is functionally regulated by the probability of resource hardship, it cannot account for preferences in WHR. Moreover, even when one cue is a more important component in attractiveness assessment than another, this is not relevant to the argument that we have assessment adaptations for the latter cue.
In earlier experimental studies using line drawings, WHR was potentially confounded with body mass, because lower WHR was manipulated by reducing waist size. This did not affect interpretation of results in most studies, but among Matsiguenga and Hadza, where men preferred the highest weight figures, it was impossible to determine if there was no preference for low WHR, or whether preference for high body weight swamped preference for lower female WHR (Sugiyama, 2004a, 2005). When presented with figures that bracketed high and low WHR in the local population but represented only one weight category, Shiwiar exhibited preference for locally lower female WHR (Sugiyama, 2004b; see also Marlowe et al., 2005). Singh, Dixson, Jessop, Morgan, and Dixon (2010) had subjects rate photos of pre- and postoperative plastic surgery patients in which fat was removed from the abdomen and placed in the buttocks: Pre- and postoperative photos thus had identical body fat and BMI, but postoperative photos had lower WHR. Cross-culturally (including among subjects in Cameroon, Indonesia, and Samoa), the postoperative photos were judged more attractive. Additionally, fMRI scans showed distinct activation of neural reward centers when men viewed the postoperative photos (Platek & Singh, 2010).
Under natural conditions, WHR-associated preference mechanisms operate in a social context that includes people of all ages, both sexes, and different parity. Morphometric measures of WHR and its relation to mate value tend to encompass this evolutionarily relevant range, but experimental studies rarely do: Female stimuli usually represent nulliparous or low-parity women around the age of peak fertility. If low female WHR is attractive in part because it signals sex and reproductive value, then the limited range presented in most experimental stimuli may artificially reduce the relative effect size of WHR on attractiveness (Sugiyama, 2005). To address the lack of studies bracketing the critical range of variation associated with women's peak reproductive value, Blackwell and I examined effects of sex and reproductive value across a previously unexplored range of WHR variation associated with the transition from childhood to reproductive maturity. We had heterosexual male and female subjects rate images from Tanner and Whitehouse's (1982) Atlas of Children's Growth. The Atlas includes front and back naked images in standard pose, anthropometric measures, and developmental markers, taken at 1- or 2-year intervals from age 4–6 to 20. Male short- and long-term attractiveness ratings did not differ. As predicted, across this critical range of variation WHR accounted for much more of the variance in attractiveness than BMI or body fat, and body fat accounted for more variance than did BMI. Using nonlinear quadratic or cubic terms in the models eliminated significant effects of both BMI and body fat, while effects of WHR remained significant. Because previous studies only used images of reproductive-age females, we then repeated analysis for reproductive-aged females only. As predicted, this removed much of the relevant variation in WHR: Although effects of WHR on attractiveness remained significant, effects of body fat or BMI had relatively greater effects on attractiveness than when male body images and prereproductive female body images were included (although not nearly as much as in other reported studies) (Blackwell & Sugiyama, 2008).
Lower WHR was also strongly associated with figures being perceived as female, and less strongly but positively associated with perceived age. Conversely, body fat was strongly associated with perceived age, but less strongly associated with the figure being perceived as female. WHR was a strong predictor of perceived sex, while body fat was a stronger predictor of age, and both perceived sex and perceived age had stronger direct effects on attractiveness than either WHR or body fat (although the latter two also had significant direct effects in the model). Using all data from female subjects included in Tanner and Whitehouse (1982) showed that female WHR reaches its lowest point before body fat reaches adult levels, just before Tanner developmental stage 5 (breast and pubic hair development). This occurs at around 15 years of age, the age closest to peak reproductive value in our stimulus set, and the age of figures (early adolescence) ranked highest in sexual attractiveness by our male raters. For male raters, there was significantly and strikingly greater attractiveness for the early adolescence female photos, associated with WHR. This contrasted with ratings by female subjects, for whom there was greater inter- than intraindividual stimuli effects on attractiveness.
In short, when a greater range of evolutionarily relevant stimuli are presented, as predicted the effects of WHR on attractiveness are much greater than previously reported. Further, preference for peak residual reproductive value was apparent. Previous studies used questionnaires to ask preferred age of a mate, and subjects may have been reluctant to consider women deemed inappropriate as mates by their respective cultures. Our study found a strong effect of actual stimulus figure age even though subjects' age estimates for the figures were not accurate. Of course, there is no reason to believe selection could produce adaptations to assess actual chronological age per se (Symons, 1979), and our subjects assumed that all figures were over the age of consent. Self-reported age preferences do not necessarily coincide with behavior. Consider the preferred age range advertised by male OkCupid members. The median 31-year-old male advertises a 22-year-old minimum and a 35-year-old maximum mate preference; however, the average 30-year-old man sends as many messages to 18-year-old women as he does to women his own age (http://blog.okcupid.com/index.php/the-case-for-an-older-woman/). Thus, stated age preferences are relatively poor indicators of actual strength of male preferences for youth. This may explain why, to date, questionnaire data have generally not found preferences for age of female peak reproductive value.
To examine how perceptions besides perceived sex and age contribute to perceived attractiveness, and to identify additional relevant variations in shape, a new set of subjects rated the Tanner and Whitehouse photo set for masculinity/femininity, physical dominance, health, and social status, as well as sexual attractiveness and perceived age. Principal Components Analysis showed complex shape dimensions involved in attractiveness assessments that are not easily described by simple anthropometrics such as WHR, BMI, and shoulder-to-stature ratio, even though they are sometimes related. For example, one factor accounting for 34% of the variance in shape preferences included high WHR with broad stance, and was associated positively with sitting height and shoulder to stature ratio, but negatively related to BMI. This factor was strongly positively associated with perceived masculinity, and with dominance and age, but only moderately attractive to female raters. It was also negatively related to perceived health and social status. Males found it sexually unattractive. Another factor included high WHR and square shoulder shape, but was unrelated to other dimensions previously hypothesized to affect attractiveness, such as BMI or shoulder to stature ratio. It was moderately sexually attractive to females, unattractive to males, and moderately associated with perceived masculinity and dominance (Blackwell & Sugiyama, 2008). Other studies using more sophisticated stimuli also show preferences for complex aspects of body shape (e.g., Brooks, Shelly, Fan, Zhai, & Chau, 2010).
Women's WHR also provides a cue of pregnancy: Pregnant women have higher WHR and, in the later stages of pregnancy, a distinctive body shape. Women's WHR is also related to parity and lactation. For example, from its lowest point around the age of peak residual reproductive value, Shuar female WHR increases with age and parity. NHANES data also shows that gluteofemoral fat stores are diminished by pregnancy and lactation, even among American women not facing major constraints in energy availability (Lassek & Gaulin, 2006). One avenue of inquiry that remains to be explored is the effect of pregnancy on attractiveness in long-term and short-term mating contexts. In the latter, we would expect pregnancy to reduce bodily attractiveness. In the former, the body shape of pregnancy may not enhance sexual attractiveness per se, but may increase attractiveness associated with bonding and investment in the woman by her long-term mate.
Trade-offs between maternal maintenance, pregnancy, immunity, basal metabolism and activity, maternal energy, and nutrient balance can have intergenerational fitness effects: Maternal nutrition and stress have significant epigenetic effects on offspring life history trajectories and health (Worthman & Kuzara, 2005). These trade-offs may explain, in part, why women in non-Western societies have higher average WHR (Cashdan, 2008). Relative energy allocation to gluteofemoral fat deposition early in the female reproductive lifespan reflects a life history bet on the future probable value of that fat for reproduction, regulated by trade-offs among uses of that energy for current reproduction, growth, maintenance, activity, and immunity. This trade-off has fitness effects. For example, the cross-sectional Shuar data suggests that women with higher WHR have slightly higher reproduction early in the lifespan than women with low WHR, but that women with low WHR have greater lifetime reproduction. Longitudinal data are needed to see if this observation holds up. If a male is following a shorter-term mating strategy, then favoring women with faster reproductive life history may yield preference for (slightly) higher WHR. Long-term mating strategy should favor lower WHR in this context.
Evidence of trade-offs and maternal depletion can also be seen in the cross-sectional Shuar data. Even though body fat is positively related to both waist and hip circumference, lower WHR is positively related to live births, whereas overall body fat is negatively associated with total live births (Sugiyama & Blackwell, 2008). Nenko and Jasienska (2009), however, found no evidence for maternal depletion among a sample of 296 well-nourished Polish women, perhaps because they had high dietary intake of LCPUFAs. Further, Jasienska et al. (2004) show optimal sex hormone profiles associated with low WHR, but only in women with large breasts. Cashdan (2008) notes another trade-off that may influence body shape and variance in male attraction. Women's ability to influence others, gain status, and get others to favor their interests is linked to hormonal regulation of behavioral correlates of assertiveness, such that women may face trade-offs in balancing androgen and estrogen profiles. Thus, in contexts where female status is relatively more important for mate value, males may show preference for slightly higher WHR.
Mates and kin are often cooperative and coalitional allies; thus, some cues of mate, offspring, kin, and coalitional value may overlap. However, others may not—for example, one may desire kindness in a mate but ruthlessness in a war ally. We must therefore understand how adaptations generating our perceptions of attractiveness are organized, and why we see cross-cultural and individual variability in assessments of attractiveness.
Complex information-processing adaptations are often expected to embody context-sensitive rules. These rules generate different psychological and behavioral outputs in response to different conditions within the range of those to which the adaptation is designed to respond (a.k.a. reaction norms). Hypotheses regarding such adaptations must ask how the mind processes local environmental cues to produce a given effect (e.g., Buss, 2000; Tooby & Cosmides, 1992). This means that hypotheses regarding the design/function of human attractiveness-assessment mechanisms must delineate specific psychological properties (or their by-products) that process local environmental cues to generate the intra- and intercultural similarities and differences found in attractiveness standards (Sugiyama, 2005). Simply documenting the variation is not enough.
A critical variable in the deployment of many adaptations is the phenotypic state of the assessor. For mating, parenting, and alliance formation, this includes developmental stage and sex, as well as health, nutritional, reproductive, and mating status. Other variables these adaptations must assess include: (a) number of assessor's co-resident kin; (b) number of people who value the assessor, how much they value him/her, and for what; (c) whether the assessor's father and/or mother are still living (e.g., Hill & Hurtado, 1996; Sugiyama, in press); (d) how aggressively formidable the assessor is compared to others; (e) how attractive the assessor is to others as a mate (e.g., Buss, 2000); and (f) how attractive the assessor is as a friend or ally. Even though the underlying functional design of attractiveness-assessment adaptations is expected to be universal, we should expect to see strategic variation in their psychological output and behavioral expression at the population, group, and individual levels.
Certain cues are expected to be weighted differently in arriving at an assessment of overall physical attractiveness. Variance in these weightings will be based on: (a) which features are statistically more likely to be associated with a particular aspect of the social value in question; (b) local environmental features (e.g., famine, health risk) that reliably change the relative value of attractiveness cues; (c) ecologically variable cues most highly cross-correlated with each other in the local environment; and (d) the phenotypic condition of the assessor. Overall judgment may reflect trade-offs among the outputs of each of these components. Additionally, outputs of different assessment components may conflict with or enhance others in the production of a final perception of attractiveness.
Each assessment mechanism can vastly reduce the computational complexity of its task by processing only a minute set of the information available in its environment. Nevertheless, each mechanism must be deployed under the evolutionarily relevant conditions, and doing this requires information intake and analysis. This implies a hierarchically organized but parallel processing system of feedback loops that inform the system based on cues received and instantiated. For instance, a cue may be related to sex and/or relative age (or stage of lifespan), and the outcome of those analyses fed into emotional adaptations that affect attractiveness, rather than affecting attractiveness directly. This view of attractiveness-assessment cognition markedly differs from the view that attractiveness-assessment mechanisms will produce cross-culturally uniform standards, with some criteria always weighted more than others. Anomalous findings and individual and cross-cultural differences in attractiveness assessments may well resolve under this approach.
Research over the last decade has been catching up with theory that psychological hypotheses and testing must take into account that psychological adaptations are expected to be context sensitive, and involve trade-offs dependent on socioecological contexts and individual phenotypic state. Here, I have focused on just a few hypothesized cues to social value to illustrate some of these complexities of attractiveness assessment psychology. Similar research is examining context sensitive regulation and trade-offs in facial, olfactory, movement and vocal cues used in attractiveness assessment, but the literature is vast so trade-offs between depth and breadth of coverage had to be made. In 2005 I called for greater collaboration between psychologists and anthropologists, particularly behavioral ecologists with established fieldsites in non-Western, nonindustrialized contexts, and this is increasing. Oddly, integration with other branches of physical anthropology, particularly human biologists who should be natural allies in the endeavor to discover the functional biology of the mind has, with a few notable exceptions, been limited. The welcome increase in cross-cultural study of attractiveness assessment, and psychology generally, make clear the need to measure, test, and control for relevant contextual variables, and that generalization from undergraduates is no longer sufficient as the basis for conclusions. However, the time, energy and monetary costs of systematic cross-cultural field collection and processing of relevant comparative data are larger than that for university lab based studies, so the coming decade will require a shift in funding priority toward cross-cultural research integrating psychological, socio-ecological, and human biological data.