Steven W. Gangestad
Department of Psychology,
University of New Mexico,
NM 87131, USA
sgangest@unm.edu

Jeffry A. Simpson
Department of Psychology,
Texas A&M University,
College Station, TX 77843, USA
jas@psyc.tamu.edu

During human evolutionary history, there were "trade-offs" between expending time and energy on child-rearing and mating, so both men and women evolved conditional mating strategies guided by cues signalling the circumstances. For some men, many short-term matings might be successful, whereas others might try to find and keep a single mate, investing effort in rearing her offspring. Recent evidence suggests that men with features signalling genetic benefits to offspring should be preferred by women as short-term mates, but there are trade-offs between a mates genetic fitness and his willingness to help in child-rearing. It is circumstances and the cues that signal them that underlie the variation in short and long-term mating strategies between and
within the sexes.

Keywords: mating, reproductive strategies, conditional strategies, evolutionary psychology, sexual selection, fluctuating asymmetry

Evolutionary principles can integrate the findings on interpersonal relationships, especially concerning mating and parental behavior. In the evolutionary approach one tries to understand human psychological design—the nature, organization, and operation of domain-specific psychological mechanisms—by identifying plausible constraints from selection pressures during evolutionary history (see Buss, 1995). Human behavior is highly flexible and environmentally responsive but "[psychological] designs that produce ‘plasticity’ can be retained by selection only if they have features that guide behavior into the infinitesimally small regions of relatively successful performance with sufficient frequency" (Tooby & Cosmides, 1992, p. 101). To understand behavioral flexibility (i.e., the ability to adjust adaptively to specific environmental circumstances), especially in the form of cultural variation, one must understand the psychological architecture that guides social interactions.

Romantic relationships have several unique qualities that distinguish them from other types of relationships. Romantic love, for example, differs from other forms of love (Hendrick & Hendrick, 1986); sexual jealousy has features and consequences that differ from other types of jealousy (Daly, Wilson, & Weghorst, 1983; Daly & Wilson, 1988); specialized verbal and nonverbal courtship rituals are observed in virtually all cultures (Eibl-Eibesfeldt, 1989); certain personal attributes (e.g., physical attractiveness) tend to assume greater importance in mating relationships than in other types of relationships (Buss, 1989; Gangestad & Buss, 1993 [see also Kenrick & Keefe, 1992]); specific facial and body features predict the attractiveness of mates in nearly all cultures (Cunningham, Barbee, & Pike, 1990; Jones & Hill, 1993; Perrett, May, & Yoshikawa, 1994); and marriage is culturally universal (Daly & Wilson, 1988). This all suggests that a specialized psychological architecture may underlie and guide romantic interactions. This would make sense considering the importance of mating, reproduction, and parenting throughout evolutionary history. The fact that the ties between mating and reproduction can now be severed by contraception does not imply that evolved psychological mechanisms no longer influence human mating (see Symons, 1987, 1992; Thornhill, 1991; Tooby & Cosmides, 1992). [See also BBS multiple book review of Symons’s The Evolution of Human Sexuality, BBS 1979.]

Evolutionary theories of human mating have been heavily influenced by research on mating in other animals (see Campbell, Simpson, & Orina, 1999). These theories—especially Trivers’s (1972) theory of sexual selection and parental investment—clarified the major (and slightly different) barriers that males and females in most species must surmount in order to increase their inclusive fitness. Trivers’s work launched the strong theoretical and empirical focus on sex differences in human mating strategies, most of which has tried to explain why women, in comparison with men, tend to be more discriminating when choosing mates and more "restricted" in their sexual behavior. Recently, Buss and Schmitt (1993) have applied and extended many of Trivers’ ideas in developing Sexual Strategies Theory. The focus on sex differences in human mating, however, has been criticized for not explaining why there is more variation in mating-related behaviors within sexes than between (see Gangestad & Simpson, 1990) and for not considering how women’s control of resources may have influenced the mating strategies of both sexes (see Gowaty, 1992a,b; Hrdy, 1997).

In this target article, we show how evolutionary principles can extend and deepen our understanding of human mating, and how and why both sexes display both short-term and long-term mating tactics in certain contexts. Sexual Strategies Theory (SST: Buss & Schmitt, 1993) tries to explain why men tend to adopt short-term mating tactics more than women. SST also emphasizes that both men and women have evolved mixed strategies involving both long-term and short-term matings. After reviewing SST, we will argue that selection produced mixed strategies that are conditional on environmental circumstances and their cues. Men and women accordingly shift between short-term or long-term mating, with considerable variation within each sex.

The paper has six major sections. In the first, we discuss basic concepts in evolutionary biology that are relevant to mating, paying particular attention to trade-offs and their adaptive role. In section two, we summarize previous theoretical claims about sex differences in short-term mating in humans, including Sexual Strategies Theory. We discuss how existing theories do not fully incorporate trade-offs to explain within-sex variation in mating tactics. In section three, we discuss how and why genetic fitness and models of good-genes sexual selection should affect mating decisions and behavior. In the fourth section, we review recent evidence concerning good-genes sexual selection in humans. In section five, we discuss how good-genes and good-provider sexual selection should affect how men and women make trade-offs and, hence, why individuals engage in different mating tactics. We also discuss how different environmental factors should affect the expression of short-term and long-term mating tactics within each sex. Section six is a short summary of the paper.

1.1. Sexual Selection

Sexual selection refers to differential reproduction among individuals due to differential "advantages" in mating, independent of advantages resulting from differential survival. Evolutionary biologists have traditionally studied the effects of sexual selection on two kinds of adaptations: (a) intrasexual competitive abilities and (b) specialized signals that appeal to members of the opposite sex (Andersson, 1994). In many species, the number of different mates that one sex can obtain is related directly and strongly to reproductive success, whereas this is less true of the other sex. In most mammals, the former sex is male, and the latter female (whose reproductive output is limited by internal gestation and lactation). Thus, in most mammals, females are a limited reproductive resource for males, who compete to attract mates. Given this disparity, sexual selection pressures should have acted more strongly on the male intrasexual competitive abilities and the specialized signals appealing to female preferences than vice versa (see Cronin, 1991; Trivers, 1972). Empirical evidence supports these predictions in many different species (see Trivers, 1985).

Theories about the signals or cues that females prefer in mates have focused on two types: (a) attributes that tend to signal qualities of a "good parent" (or a "good provider") and (b) attributes suggesting that an individual may have "good genes" (Cronin, 1991; Gangestad, 1993; Miller, 1998). Theories of good parenting have been fairly uncontroversial. Those involving good genes, on the other hand, have been debated extensively (see below). Consequently, most applications of sexual selection to human mating have not seriously considered good-genes sexual selection. Recently, however, theoretical and empirical research has indicated that both good- parenting and good-genes selection could have operated on many species (Kirkpatrick, 1996; Møller, 1994), particularly those in which males provide substantial parental care (as is true of humans). In what follows, we will argue that these selectional processes are likely to have produced differential mating tactics within each sex.

Natural selection also has important implications for human mating. In many species, parental care by males might have evolved to increase offspring fitness directly and, thus, could have evolutionary benefits (Clutton-Brock, 1991), particularly in humans (Lancaster & Lancaster, 1985). Natural selection could also have led females to obtain the resources necessary to reproduce and compete for additional resources (Gowaty, 1992a; Hrdy, 1981). Moreover, mating can have benefits that are indirectly derived from reproduction and are not associated with sexual selection. For example, females may induce paternity uncertainty by having multiple mates, possibly leading to more protection or greater tolerance of offspring by different men (Hrdy, 1981).

1.2. Adaptations

Adaptations are traits that gave individuals a gene-transmitting advantage over other individuals who had different variants of the traits in the original environments in which the traits evolved (i.e., in the "environment of evolutionary adaptedness" or EEA; Alcock, 1993). They are identified by showing that specific traits or behaviors meet the requirements of "special design" (Williams, 1966), evidence that a trait shows specificity, efficiency, and economy for producing a particular beneficial effect (the "function" of the trait). In many instances, adaptive behavioral flexibility should be facilitated by domain-specific psychological mechanisms (Tooby & Cosmides, 1992). These mechanisms should operate according to specific decision rules that are activated by certain environmental cues and that produce efficient, stable, persistent, and finely tuned responses (e.g., specific perceptions, arousal, behavioral reactions) designed to solve specific adaptive problems (e.g., choosing or attracting mates). These decision rules do not have to be consciously formulated or under deliberate control. During puberty, for example, individuals' physiological adaptations "decide" to produce secondary sexual characteristics. Likewise, the basis of attraction to a potential mate, suspicions about a mate's infidelity, or interpretations of flirtatious glances are all responses to specific environmental cues that are likely to be governed by implicit, unconscious decision rules.

1.3. Evolutionary Functional Analysis and Trade-Offs

Evolutionary theory existed for more than a century before Williams’s (1966) groundbreaking work on adaptations. This lag occurred because "classical Darwinists" viewed adaptations almost exclusively in terms of the benefits they bestowed on organisms; associated costs were not fully considered (Cronin, 1991). Individuals must invest considerable time, effort, and energy to accomplish the major tasks that most directly affect their inclusive fitness: surviving to reproductive age, reproducing successfully, and rearing offspring to reproductive age. These factors are important not only because they put individuals at risk (and, hence, can decrease their future reproductive success, despite immediate benefits), but because individuals could have used these resources differently. Thus, adaptations have "opportunity costs" associated with them—costs that accrue from lost fitness benefits that might have been achieved by using resources in different ways. A major "modern Darwinian" insight was that one must weigh the benefits in relation to the costs in order to appreciate whether and how adaptations evolved during evolutionary history (using cost-benefit analyses: Parker & Maynard Smith, 1991). The fundamental goal of evolutionary analysis is to specify the cost-benefit "trade-offs" that led individuals to allocate their time, energy, and effort to activities in ways that increased their ancestors’ inclusive fitness.

1.3.1. Parenting effort versus mating effort. Trade-offs are illustrated by the differential allocation of effort to parenting versus mating activities. Trivers (1972) defined parental investment as "any investment by the parent in an individual offspring that increases the offspring's chance of surviving (and hence reproductive success) at the cost of the parent's ability to invest in other [including future] offspring" (p. 139, italics ours). This definition implies a basic trade-off. Parental investment can increase the probability offspring survive and subsequently reproducte, yielding greater fitness. However, parental investment also has costs. Indeed, according to Trivers (1972), the amount of investment is measured in units of lost benefits of alternative investment (i.e., lost opportunities to invest in other offspring). Individuals who engage in parental effort (the sum total of parental investment in all offspring; Low, 1978) could be doing things with their limited time and energy, such as devoting greater effort to mating (e.g., attempting to attract additional mates). Thus, costs include the lost benefits of potentially productive yet foregone activities (Gross & Sargent, 1985).

If fitness gains arising from mating effort (through increased access to more mates) are greater for one sex, selection should tilt the balance toward more mating than parental effort in that sex compared to the other (see Low, 1978). Because obtaining more mates (a component of mating effort) cannot produce large fitness gains for females in many species (e.g., Bateman, 1948), females in those species should devote more time and energy to parental investment than the average male. Conversely, most males in these species should devote a greater portion of their time, energy, and effort to finding and attracting mates compared to the average female. Hence, trade-offs are implicit in the prediction that males and females should, on average, differ in the total reproductive effort they devote to mating, a prediction that is central to many theories of human mating. These same notions about trade-offs, however, also lead to the expectation that males should differ from one another in the total amount of reproductive effort devoted to mating.

As an illustration, consider Burley’s (1986) innovative research on zebra finches. She found that when females mated with males who were experimentally manipulated to be more "attractive," females increased their parental investment in the resultant offspring. As a result, males actually decreased the amount of parental effort they exerted and, at the same time, experienced greater success in extra-pair matings. Thus, when a male’s mating effort was more productive, he exerted more mating effort at the expense of reduced parental effort. This notion has not been fully incorporated into existing theories of human mating.

2.1. The Concept of Sexual Strategies

Consider, for example, Trivers’s (1972) hypothesis that males in many biparentally-investing species may invest heavily in offspring, yet remain open to low-cost mating opportunities with other females. This is a mixed mating strategy involving multiple behavioral tactics in which each tactic is displayed with a certain frequency. If each tactic is cued by specific environmental stimuli (e.g., the prolonged absence of a mate, having short-term sex only with mates who have certain attributes, seeking short-term sex only when such efforts have paid off in the past), the strategy is conditional.

2.1.1 Conditional strategies. In the last two decades, evolutionary theorists have begun to acknowledge that selection pressures should not have produced a single "best" mating tactic (or mixture of tactics) for males and females in most species. Instead, selection should have fashioned considerable phenotypic diversity in mating (Gross, 1996). Guided by concepts from game theory (Maynard Smith, 1982 [see also BBS multiple book review]) and the theory of evolutionary stable strategies (Dawkins, 1980; Parker, 1984), evolutionary biologists are now documenting how having alternative mating tactics gives individuals of each sex differential reproductive fitness in various species. Although relatively little theory and research has focused on the mating behavior of human beings, hundreds of studies have confirmed that males and females in a wide range of species display alternative mating tactics that reflect conditional strategies (Gross, 1996). Burley’s (1986) finding that male zebra finches’ relative allocation of parental effort and extra-pair mating effort is contingent on their attractiveness is a good illustration of a conditional strategy.

Conditional strategies have five main properties (see Gross, 1996): (a) They involve different behavioral tactics that are consciously or unconsciously "chosen" by an individual, (b) the choices between tactics are "made" in response to specific features or cues in the environment, often an individual’s status or attractiveness relative to other individuals, (c) all individuals are genetically monomorphic (i.e., they are are genetically designed to enact the same tactics), (d) during their evolution, the average adaptive values of different tactics were not equal except at a "switchpoint" on a continuum of environmental input (e.g., individuals’ relative status) where the costs and benefits of each tactic balanced out, and (e) during their evolution, the chosen tactic tended to yield higher fitness for the individual than other tactics given current environmental conditions than other tactics. Thus, the environmental conditions moderate the fitness gains of pursuing different tactics (e.g., exerting parental effort, pursuing short-term matings), thereby affecting the optimal allocation of effort to different tactics.

If males differ in the conditions under which they engage in different tactics, they are enacting alternate conditional strategies. Although alternate strategies can be noncontinuously distributed in a population (e.g., if certain males never invest in offspring and always seek short-term mates), they are usually distributed continuously. This should occur if males differ in how long they tolerate their mate's absence before pursuing other mates, or if males differ in the degree to which they expect extra-pair mates to have certain desirable attributes (Dominey, 1984). Alternate strategies can reflect genetic polymorphisms (see Gross, 1996). Although such polymorphisms exist in nature and may underlie certain variations in human mating strategies (see Gangestad & Simpson, 1990; Wilson, 1984), we focus on conditional strategies in this article.

2.2. Sex Differences in Sexual Strategies

Most evolutionary theories of mating have focused on the different strategies purportedly adopted by men and women. Based on Trivers' (1972) theorizing, Wilson (1978) suggested that men and women differ along an assertiveness-coy dimension:

More recent theories of sex differences in mating have incorporated similar notions. Perhaps the most ambitious of these new theories is Buss and Schmitt’s (1993) Sexual Strategies Theory (SST; see also Buss, 1994, 1998a,b). Buss and Schmitt’s (1993) theory emphasizes that both men and women should have evolved tactics for obtaining long-term mates and investing in offspring. Similar to Wilson’s theory, however, SST predicts that men should have evolved to seek multiple mates in particular:

This prediction is based on an implicit cost-benefit analysis that involves trade-offs between different activities which were presumably optimal for men and women in evolutionary history. According to SST, men experienced, on average, a greater net benefit than women for attempting to attract multiple mates. This prediction has received empirical support in several recent studies. For example men report desiring four times as many sex partners in the next five years as women do (Buss & Schmitt, 1993).

2.3. A Sex-Specific Universal?

Evidence indicates that men do, on average, desire and seek short-term mates more than women. Wilson, however, makes an even stronger prediction about this sex difference. He claims that males should benefit by being "aggressive, hasty, fickle, and undiscriminating." This should hold true for all males in a species. Although Buss and Schmitt (1993) do not explicitly claim that all men have a "powerful desire for sexual access to a large number of women," they do not qualify their statement to indicate that some men may not. Clearly, a mean sex difference—even a large one—does not imply a near-universal propensity in each sex. Indeed, a mean sex difference could exist even when the modal response of each sex is similar (e.g., when the distribution for one sex is more variable and skewed than for the other sex).

Sex differences in the desire for multiple sex partners and interest in engaging in short-term mating are moderate in magnitude. Buss and Schmitt (1993) report g s (i.e., the number of within-sex standard deviations on which the means of the two sexes differ) for each effect they found. We converted this statistic to percentages of variance accounted for on each variable by sex. Sex accounts for 16% of the variance in seeking short-term mates, 9% of the variance in the number of sex partners desired within a specified time, and 20% of the variance in the probability of consenting to sex after knowing an attractive, opposite-sex person for a given length of time. In a recent meta-analysis, Oliver and Hyde (1996) found that the mean sex difference in interest in casual sex accounted for 25% of the variance. Across different college samples, we have found that sex accounts for 8%-20% of the variance in expressed interest and willingness to engage in sex without commitment (Simpson & Gangestad, 1991).

While these effect sizes are fairly large by conventional standards (Cohen, 1977), substantial amounts of within-sex variation and overlap exist between the distributions of women and men. For example, despite the mean sex difference in the desire to have sex without commitment (Simpson & Gangestad, 1991), approximately 30% of men express less favorable attitudes about casual sex compared to the median attitudes of women. Substantial overlap also exists for many other measures, including interest in sex with multiple partners and optimal time before having sex (see Miller & Fishkin, 1997). Generic claims that women desire certain mating arrangements while men desire others raise the questions "which men?" and "which women?" (Gowaty, 1992b). If the within-sex variation in mating strategies is adaptive, the environmental contingencies in the EEA that led men and women to adopt short-term versus long-term mating tactics must be specified.

Sexual Strategies Theory acknowledges that both sexes can and do exhibit short-term and long-term mating tactics (i.e., that human mating strategies are mixed); indeed, this is one of SST’s defining features. Buss and Schmitt (1993) also explicitly note that within-sex variation in mating tactics may be adaptive. Nonetheless, within-sex variation has received less theoretical attention than sex differences have and remains an "unresolved issue" (Buss & Schmitt, 1993) within SST (see also Buss, 1998a). Moreover, to state that "Men who lack mechanisms such as a desire for a variety of partners ... would have been out-reproduced by men who successfully solved [the problem of partner number] entailed by the pursuit of a short-term mating strategy" (Buss, 1998a, p. 24) is to imply that evolution would not have favored a male strategy entailing little desire for multiple mates. However, as we shall see, most men may have benefited reproductively by having little interest in pursuing multiple mates. Both the desire for multiple mates and the lack of this desire should have been beneficial under certain conditions.

2.4. Cost-Benefit Considerations

If aggressiveness and interest in short-term mating is the "most profitable" strategy for men, why don’t all men pursue short-term mates? Differential cost-benefit trade-offs should have led men to pursue more mates on average than women. Consequently, men should, on average, dedicate more time and energy to mating effort than women. These premises, however, do not lead to the conclusion that the best strategy for most males would involve high mating effort, whereas the best strategy for most females would entail high parental effort. A consideration of trade-offs suggests that the best strategy for men may not always—or even usually—be to devote effort to pursuing short-term mates (see Kitcher, 1985 [see also multiple book review]; Maynard Smith, 1982). Similarly, the best strategy for women may not always be to devote less effort to pursuing short-term mates than the average male.

From our perspective, the basic notion that men, on average, evolved to engage in greater mating effort than women correctly assumes that some ancestral males could have increased their fitness by trying to attract multiple mates. However, the more stringent notion that the best evolutionary strategy for most men should have been to pursue multiple mates assumes that most men were successful at short-term mating. This is a dubious assumption. Men’s short-term mating success should have depended partly on which attributes women preferred and desired in short-term mates. Relatively few men are likely to have satisfied these preferences. Thus, for some men, one cost of engaging in short-term mating would have been the loss of opportunities to engage in parental investment (which human males also evolved to do: see Alexander & Noonan, 1979; Lancaster & Lancaster, 1985). If short-term mating effort generated poor pay-offs for men who did not have the attributes most women desired in short-term mates, these costs should have outweighed the benefits of pursuing short-term mating. Hence, the most profitable mating strategy for the typical man may have been to devote most of his reproductive effort to enhancing the phenotypic quality of his offspring by investing heavily in a single mate’s offspring, despite the fact that a few men with certain desirable attributes were even more successful by adopting short-term tactics.

Consider a parallel example where the currency of profit is money rather than fitness. Suppose that brain surgeons make more money than everyone else does. Clearly, it is not "most profitable" money-making strategy for everyone to decide to become a brain surgeon. Only so many people can become brain surgeons; hence, most people would be bound to fail. Instead, the most profitable strategy for most people would be to select a career that provides them with the most money, given their abilities and, in light of a competitive market, their chances of success at their career options.

If the most evolutionarily profitable strategy for men depended in part on the attributes each man possessed (e.g., his ability to attract multiple, short-term mates), what sexual strategy should have evolved in men? The answer is a conditional mating strategy (see Gross, 1996; Trivers, 1972). Most men would probably have profited from substantial investment in a primary mateship (or small set of mateships), investing fairly heavily in subsequent offspring. When men could obtain short-term, opportunistic matings with other women (often other men’s primary mates), they could have profited by pursuing the sex-typical strategy outlined by Wilson (1978) and Buss and Schmitt (1993). However, when they could not, men should have shifted greater amounts of time, energy, and other resources to long-term mating (including mate guarding) and parental investment in order to attract and retain one mate and derive the benefits of parental investment. Although men may have "evolved over evolutionary history [the capacity for] a powerful desire for sexual access to a large number of women" (Buss & Schmitt, 1993, p. 208; our addition and emphasis), the desire for short-term mating should be expressed conditionally, and it should be observed infrequently under conditions when only a few men are able to attract short-term mates. The key adaptation should be a set of decision rules about when and how to allocate reproductive effort wisely and contingently, rather than a universal desire for short-term mating. As noted above, this is not a novel notion. Conditional strategies are a central topic in behavioral ecology (see Krebs & Davies, 1993, on "making the best of a bad job"; see also Dawkins, 1980; Gross, 1996). However, the possible role of conditional strategies in human mating has received relatively little attention (see Buss, 1998a; Buss & Schmitt, 1993). We accordingly propose that evolution generated conditional strategies in both sexes, resulting in a plurality of mating tactics.

Trivers (1972) also noted that "the optimal [mixed strategy] is likely to differ for different males" (p. 146). Trivers was in fact the first evolutionary theorist to acknowledge that selection could have favored adaptations that led some males to devote very little time and effort to short-term mating. It is ironic that his classic article is often cited as suggesting that the most profitable strategy for males should be to pursue short-term mating almost universally.

What attributes have women evolved to desire in short-term mates? It is important to identify these attributes because they should moderate strategy choice in men. In the next section, we suggest that good-genes sexual selection may be central to answering this question. As we shall see, these selection pressures may also have important implications for understanding variation in women’s interest in short-term mating.

In many species, males provide little or no parental investment. Yet even in these species, females are selective and consistently prefer certain males over others as mates. Good-genes sexual selection (GGSS) can explain female mate preferences in species that do not have much paternal investment (Cronin, 1991). According to models of "good genes" selection, females have evolved to prefer males who possess indicators of viability and good condition, that is, adaptive attributes that might be passed on to their offspring through genetic inheritance.

3.1. Genetic Variation in Fitness

For mate preferences based on indicators of heritable fitness to evolve, additive genetic variance must underlie fitness (i.e., fitness must be transmitted genetically across generations). Historically, GGSS has been controversial because population geneticists have assumed that selection typically reduces the heritability of fitness to nearly zero (e.g., see Kirkpatrick's [1985, 1986] arguments against the "sexy son" [Weatherhead & Robertson 1979]). Recent empirical and theoretical developments, however, have led even the harshest critic to acknowledge that GGSS can and probably has occurred in a variety of species (see Kirkpatrick, 1996).

The new empirical development is the ability to estimate the actual amount of genetic variation in fitness (or fitness components) in natural populations. The most relevant measure of genetic variation is the additive genetic coefficient of variation (CV_a), the genetic standard deviation of a trait, standardized by the trait mean and multiplied by 100 (Houle, 1992). Houle (1992) has argued that the CV_as of fitness traits (e.g., longevity, fecundity) in natural populations are typically greater than those of ordinary morphological traits (e.g., height). Human fecundity, for example, has a CV_a of 15 to 20 (Burt, 1995), 4 to 5 times greater than that of human height. Values exceeding 10 have been estimated for fitness traits in other organisms (Burt, 1995; Houle, 1992), compared to values around 5 for ordinary traits or traits under stabilizing selection (Houle, 1992; Pomiankowski & Møller, 1995). Fitness itself has been estimated to have a CV_a of between 10 and 30 in natural populations of many species (Burt, 1995).

New theoretical developments are based on the notion that the amount of genetic variance in fitness within a population is the result of two opposing forces: natural selection (which removes genetic variation), and forces that degrade an organism’s fitness, such as mutations and environmental change (both of which enhance genetic variation: Fisher, 1958). As long as deleterious mutations and environmental change are negligible, natural selection should keep genetic variance near zero. Recently, population geneticists have begun to model and estimate the amount of genetic variance in fitness that can be maintained by mutations alone. The genome-wide mutation rate (i.e., the number of new deleterious mutations per organism) is about 1 in Drosophila, a rate that explains at least half of its estimated genetic variance in fitness (a CV_a of 6-17; Charlesworth, 1990; Charlesworth & Hughes, 1998). The genome-wide mutation rate in humans is probably higher, given our larger genome (Charlesworth, 1990). Thus, mutations probably account for a human CV_a of fitness of 10 or more. Hence, mutations alone might produce more genetic variance in fitness than is observed in most ordinary traits.

Recent theory has also drawn attention to phenomena that induce rapid change in the selective environments of organisms. One set of strong selective forces operating on long-lived organisms such as humans are pathogens. Because parasites evolve in response to their hosts’ defenses against them, no set of host defenses is evolutionarily stable; they are always being challenged by changing parasites. These rapid shifts in the selective environments of hosts not only ensure that hosts will never be free of maladaptations caused by parasites; they also ensure that hosts will vary in their ability to resist parasites. Hence, host-parasite coevolution maintains genetic variance in host fitness (Anderson & May, 1982; Haldane, 1949; Hamilton, 1982; Tooby, 1982).

It is important to emphasize that alleles which are good genes at one point in time need not be intrinsically better than alternate alleles. Host-parasite coevolution imposes changing selection pressures on host genes, maintaining heritable fitness in individuals. Thus, an allele that is a good gene today might be selected against in future generations, and an allele that is currently selected against could become a "good gene" in the future. Because no gene is inherently better than alternate ones, the population does not become more fit through selection.

In combination, mutation and host-parasite coevolution can maintain a substantial amount of genetic variance; they explain why fitness traits contain relatively large amounts of genetic variance in natural populations. Moreover, recent theoretical models indicate that large amounts of genetic variance in fitness not only can but do produce GGSS (Kirkpatrick, 1996; see also Andersson, 1986; Charlesworth, 1988; Grafen, 1990; Heywood, 1989; Iwasa, Pomiankowski, & Nee, 1991; Pomiankowski, 1987).

3.2. Markers of Heritable Fitness

The degree to which individuals harbor mildly deleterious alleles or are less pathogen-resistant must be inferred from phenotypic markers (e.g., "advertisements": Zahavi, 1975). GGSS must operate through honest signaling, which explains why only certain features tend to be valid indicators of individuals’ underlying condition and, potentially, their genetic fitness (Grafen, 1990; Zahavi, 1975). If an attribute is a marker of heritable fitness, sexual selection should favor new genes that simulate the valued feature, even in individuals who do not actually have high fitness. Thus, "cheaters"—individuals who display the selected phenotypic traits or behaviors but do not have high fitness—may invade the population. Over time, selection should eliminate these attributes as valid cues of heritable fitness, and preferences for them should diminish. Under certain conditions, however, an attribute's link with heritable fitness can remain honest and stable. In particular, an attribute can remain an "honest" advertisement when individuals who have deleterious alleles or who are less pathogen-resistant cannot develop or maintain the attribute without incurring substantial costs. One set of attributes that meet this criterion are those that conditionally "handicap" individuals who have mutations or are less pathogen-resistant.

Both mutations (Pomiankowski et al., 1991) and pathogens (Hamilton & Zuk, 1982) divert an individual’s energy and resources. Hence, honestly advertised traits tend to be energetically costly because individuals who have mutations or are less pathogen-resistant cannot develop such traits without diverting valuable resources from competing demands (e.g., sustaining their immune systems: Folstad & Karter, 1992). In peacocks and other birds with extravagant features, costly handicaps include exaggerated sexual ornaments and colorful plumage (Zuk, Thornhill, Ligon, & Johnson, 1990). In many mammals, they include large size and increased musculature resulting in sexual dimorphism for size in polygynous species, including primates (Alexander, Hoogland, Howard, Noonan, & Sherman, 1979).

This type of selection can be understood in terms of cost-benefit analysis. A costly trait confers certain benefits. For example, males who have the trait may be preferred as mates. However, having such a trait also involves costs because the resources used to develop the trait could have been used for other purposes. At some point, the marginal gains of investing in the costly trait begin to diminish, and individuals should no longer be selected to invest in the trait beyond that point. The benefits of sexual selection can be maintained if males with different levels of heritable fitness maximize their benefits relative to their costs (i.e., maximize their outcomes) at different levels of the costly trait (Grafen, 1990).

Trivers (1972) proposed that intrasexual competitive abilities may have evolved to be valid cues of heritable fitness. Successful intrasexual competition, such as winning physical fights, requires developing the potentially costly attributes used in competition (e.g., muscularity) and the expenditure of considerable energy during competition. Just as highly viable males should be more capable of enduring "handicapping" traits than less viable males, they should also be able to devote more energy to developing the physical tools needed for successful intrasexual competition. Females may in turn may have evolved to pay attention to the outcomes of intrasexual competition to assess male fitness (Andersson, 1994).

3.3. Tests of Good-Genes Sexual Selection

To test GGSS, researchers should show that direct indicators of individuals’ genetic fitness are associated with their attractiveness as a mate (particularly as a short-term mate). Unfortunately, perfect indicators of fitness do not exist. Tests of GGSS must therefore rely on indirect, fallible markers of genetic fitness. The best available measure may be fluctuating asymmetry.

Fluctuating asymmetry (FA) reflects the degree to which individuals deviate from absolute perfect symmetry on bilateral features (e.g., in humans, both ears, both feet; in other species, bilateral fins, bilateral tail feathers) for which the signed right versus left differences have a population mean close to zero and are nearly normally distributed (Van Valen, 1962). Asymmetry is believed to reflect deviations in developmental design due to the disruptive effects of environmental or genetic abnormalities encountered during the lifespan (i.e., developmental instability: see Lerner, 1954; Palmer & Strobeck, 1986; Parsons, 1990; Soule, 1982; Thoday, 1955; Waddington, 1957). Because these disturbances lower reproductive fitness, pronounced asymmetry reflects maladaptation (Møller & Swaddle, 1997). Recent meta-analyses and reviews have shown that greater asymmetry is associated with lower fecundity, slower growth, and poorer survival in many species (Leung & Forbes, 1996; Møller, 1997; Thornhill & Møller, 1997; see also Clarke, 1998, and a subsequent analysis by Møller, in press). FA should be a good marker of genetic fitness because an individual’s degree of asymmetry should be affected primarily by (1) mutations (that cause lower metabolic efficiency and imprecise development), and (2) diseases (that reflect an individual’s pathogen-resistance) (Møller, 1992). Consequently, FA probably reflects both genetic and nongenetic variation in fitness. Because sexual selection should lead females to mate with males who have phenotypic traits that signal greater fitness, the association between FA and mating success has been studied in many species. A recent meta-analysis of these studies reveals that, on average, more symmetrical individuals have greater mating success (Møller & Thornhill, 1998).

GGSS is of course not the only form of selection that could explain these findings. In some species, more symmetrical males may provide greater material benefits (e.g., more direct and better parental care or more physical protection of young). In addition, females may find less symmetrical males less attractive to avoid contracting infectious diseases from them. However, three novel sets of findings suggest that GGSS may have occurred in humans.

First, a recent meta-analysis indicates that, across many species, symmetry is partly heritable (Møller & Thornhill, 1997). Modeling the relationship between asymmetry and underlying developmental imprecision in these data, Gangestad and Thornhill (in press) have estimated that the additive genetic coefficient of variation of developmental imprecision is approximately 15 to 20, about the same size as other fitness traits and much greater than ordinary morphological traits (e.g., height: see Houle, 1992). If part of this genetic variance is associated with fitness, the favored status of symmetrical males may in part reflect the operation of GGSS (Kirkpatrick, 1996).

Second, in some species in which symmetry predicts male mating success, more symmetrical males are favored as extra-pair mates, even when they provide little or no material benefits. For example, although male and female European barn swallows mate seasonally and rear offspring together, their rate of extra-pair paternity is approximately 35% (Møller, 1994). More symmetrical male barn swallows are the main beneficiaries of extra-pair mating, yet they do not provide material benefits that enhance the reproductive success of their female mates (see Møller, 1994).

Third, in some species, more symmetrical males also provide fewer material benefits to their primary mates. For example, more symmetrical male barn swallows spend less time feeding their offspring than do less symmetrical males, and they do not compensate for their lack of time with greater feeding efficiency. In a recent review of 18 bird species, Møller and Thornhill (1997a) have documented an association between extra-pair paternity and the extent to which attractive males engage in direct parental care. Specifically, when the rate of extra-pair paternity is high (and, thus, when males can benefit more from trying to attract extra-pair mates), attractive males perform a smaller proportion of offspring feedings than do less attractive males. Exerting greater extra-pair mating effort should yield larger payoffs for more attractive males, and this is evident in the time they fail to spend engaging in a competing activity: providing direct parental care. We suggest that this type of trade-off also occurs in humans. Over evolutionary history, men who had indicators of genotypic quality should have experienced larger gains in fitness payoffs than men who lacked these indicators. Moreover, men should have evolved to conditionally "decide" to allocate more versus less effort to mating or parenting, depending on the degree to which they possess these features.

New evidence indicates that GGSS might have operated on ancestral humans. Most of this evidence indicates that fluctuating asymmetry (FA) is systematically associated with male mating success in humans. Although FA is not the only valid marker of heritable fitness, it currently appears to be the best marker. As we discuss below, other, more visible features (such as physicality, social dominance, and intrasexual competition tactics) are likely to be the proximate cues that "advertise" an individual’s genetic fitness. We focus on FA because rival models of selection do not capture the complex relations predicted from developmental imprecision.

4.1. FA and Sexual Behavior

If women evolved to prefer men who exhibited indicators of genetic fitness, more symmetrical men should have more lifetime sexual partners (see also Perusse, 1993, 1994). Thornhill and Gangestad (1994) measured the symmetry of men and women on seven bilateral features (foot width, ankle width, hand width, wrist width, elbow width, ear width, and ear length) using digital body calipers. They then aggregated these measures (each standardized by feature size) to form a global index of FA. With the effects of age partialed out, more symmetrical men reported more lifetime partners than did less symmetrical men, r = -.32. Controlling for potential artifacts (e.g., height, ethnicity, marital status) strengthened this effect, partial r = -.38. These effects have also been estimated in several other studies. The weighted correlation in a sample of over 500 men is -.21 (p < .0001), and the latent correlation between developmental imprecision and male partner number is approximately -.38 (Gangestad & Thornhill, in press). Waynforth (1998) has found a similar correlation (-.23) in a sample of Mayan men in Belize and has also documented that more symmetrical men have higher fertility. Because women should be less inclined to convert intrasexual competitive advantages into increased numbers of mates (see Trivers, 1972), no relation between women’s FA and their number of lifetime sex partners was predicted, and none has been consistently found in several studies (see Gangestad & Thornhill, 1997a).

Given that these studies are based on self-report data, the findings could be explained by a tendency for more symmetrical men to exaggerate their past sexual behavior. In one study, therefore, men’s scores on narcissism (a measure that reflects self-aggrandizement tendencies: John & Robins, 1994) were partialed out. Controlling for narcissism did not attenuate the relation between men’s FA and their number of lifetime sex partners, partial r = -.27 (Gangestad & Thornhill, 1997b).

4.2. Men’s Extra-pair Sex and FA

Extra-pair sex (i.e., sex outside a current, ongoing relationship) is one form of short-term, opportunistic mating. If women evolved to desire men with greater heritable fitness (independent of the investment and resources that men provided), more symmetrical men should have more extra-pair partners (see Benshoof & Thornhill, 1979; Smith, 1984). Gangestad and Thornhill (1997b) found that men’s FA does predict their number of extra-pair sex partners, r = -.17 (reflecting a latent correlation between extra-pair partners and developmental imprecision of about -.36; Gangestad & Thornhill, in press). Once potential artifacts were controlled (e.g., social status, SES in family of origin, anticipated future salary), the correlation between FA and extra-pair sex increased slightly. Neither men's social status nor their resources (as indexed by their SES and their anticipated future salary) predicted their frequency of extra-pair sex. Although Perusse (1993) has found that men with higher status report having more sex partners than men with lower status, this effect did not hold for married men in his sample. Thus, it is unclear whether men with higher status (e.g., wealthier men) have more extra-pair mates than men with lower status.

4.3. Women’s Extra-pair Sex and FA

More symmetrical men should also be preferred as women’s extra-pair mates (i.e., mates chosen by women who already have primary mates, regardless of whether or not the man has a primary mate). Gangestad and Thornhill (1997b) found that men’s FA predicted the number of times they were chosen as extra-pair partners by women who were simultaneously involved in other, ongoing relationships, r = -.26 (reflecting a latent correlation between developmental imprecision and these partners estimated to be -.60; Gangestad & Thornhill, in press).

4.4. FA and Associated Traits

What personal attributes should mediate the link between men's FA and their number of lifetime sex partners? One possibility is physical attractiveness (Feingold, 1990), which Buss and Schmitt (1993) have found is important to women when they evaluate short-term mates and Scheib (1999) found is particularly important to women choosing extra-pair mates. In fact, a man’s attractiveness in short-term mating contexts is just as important to women as a woman's attractiveness is to men when men evaluate long-term mates. Buss and Schmitt (1993) claim that this finding is consistent with the notion that attractiveness might have been an indicator of genetic fitness in ancestral environments, but they say little about this possibility.

If preferences for attractiveness evolved in response to GGSS, attractiveness should be correlated with markers of heritable fitness (such as FA). Gangestad et al. (1994) measured individuals’ FA and then had coders rate their physical attractiveness from facial photographs. After controlling for potential artifacts (e.g., age, height), FA significantly predicted men's facial attractiveness, r = -.33, but not women's, r = -.17, ns. That is, more symmetrical men were rated as more physically attractive. This sex difference has been replicated (Thornhill & Gangestad, 1994), but has not been found consistently (Gangestad & Thornhill, 1997a). Across several studies, the correlation between facial attractiveness and body FA is significant but rather small (see Gangestad & Thornhill, 1997a). Studies of the association between facial symmetry and facial attractiveness have similarly yielded mixed but positive results overall (e.g., Grammer & Thornhill, 1994; Kowner, 1996; Mealey, Bridgestock, & Townsend, 1999; Scheib, Gangestad, & Thornhill, 1999; see also Møller & Thornhill, 1998). Gangestad and Thornhill (in press) have estimated that the latent correlation between developmental imprecision and male facial attractiveness is about -.28.

Because facial attractiveness mediates only part of the relation between men’s FA and their sexual history, Gangestad and Thornhill (1997a, 1998a) tested three other potential mediators, namely three sexually dimorphic traits involved in intrasexual competition: (1) Body mass: Humans show moderate sexual dimorphism, consistent with their purported ancestral polygyny (Alexander et al., 1979); (2) Physicality: A measure of men's muscularity, robustness, and vigor, as rated by both men and their romantic partners; and (3) Social dominance: A measure based on the California Adult Q-Sort. Once again, a relation was found between men's FA and their number of lifetime partners, r = -.29 (estimated by causal modeling). In addition, all three traits associated with intrasexual competition were predicted by men's FA, estimated rs = -.31, -.39, and -.39, for body mass, physicality, and social dominance, respectively. The indirect effects mediated through these traits accounted for over 70% of the total effect of FA on lifetime number of partners. As expected, women's FA did not predict their number of lifetime partners, and it did not correlate with any of the three intrasexual competitive traits, average r = -.06.

The fact that men's FA covaries with traits that should have facilitated intrasexual competition in ancestral environments (i.e., physical intimidation of competitors) is consistent with other research showing that more symmetrical men engage in fights with other men more often, particularly fights they initiate (Furlow, Gangestad, & Armijo-Prewitt, 1998). Besides affecting the outcomes of intrasexual competition, these traits may serve as cues that women use to evaluate men as potential short-term mates. This would explain why all three traits mediate the link between men's FA and their sexual history.

Because men who have traits that facilitate intrasexual competitive success should benefit by directly comparing themselves with their competitors, Simpson, Gangestad, Christensen, and Leck (1999) predicted that, when competing for a date, more symmetrical men would use direct competition tactics (e.g., directly comparing themselves with, and derogating, their competitors). After measuring their FA, Simpson et al. had men compete with other men for a lunch date with an attractive woman. Each man was asked a series of questions over a video system by a female interviewer (actually a videotaped experimental assistant) located in another room. After the interview was finished, the female asked each man to tell the "competitor" (ostensibly located in a different room) why she should choose him instead of the competitor. Each interaction was videotaped and coded for specific intrasexual competition tactics. Relative to less symmetrical men, more symmetrical men were more likely to directly compare themselves with and belittle the competitor, r = .49. These results support the hypothesis that more symmetrical men should engage in direct intrasexual competition tactics. They also provide further evidence about the proximate cues that may "advertise" genotypic quality (indexed by men’s FA).

Finally, intellectual ability may be a marker of developmental precision and health. In two studies, Furlow, Armijo-Prewitt, Gangestad, and Thornhill (1997) found an association between a measure of fluid intelligence (Cattell’s Culture-Fair Intelligence Test) and FA, mean r = -.23 (the estimated latent correlation with developmental imprecision was -.56; Gangestad & Thornhill, in press). The size of this effect did not differ across the sexes.

4.5. Evidence for Women Preferring Symmetrical Men for Their Gametes

The fact that more symmetrical men have more mates might not be explained by GGSS exclusively. More symmetrical men could be advantaged because of the superior material benefits more viable males typically provided in the EEA or because they have success in intrasexual competition (Kirkpatrick & Ryan, 1991). However, additional evidence (reviewed below) suggests that these alternative selectional processes cannot fully explain the greater short-term mating success of more symmetrical men. This evidence suggests that women’s preferences for more symmetrical men may be specifically "designed" to favor their gametes.

4.5.1. Female orgasm and sperm retention. Baker and Bellis (1995) have conjectured that women's orgasms evolved to manipulate sperm competition via differential sperm retention. If male extra-pair mates in the EEA provided advantages to offspring above and beyond the investment and tangible resources they offered (through good genes or genetic diversification: see Smith, 1984), selection could have favored adaptations that biased sperm competition in favor of conceptions with extra-pair mates who had higher genetic fitness. Baker and Bellis (1995) report that women who have both an in-pair (primary) partner and an extra-pair partner have patterns of orgasms that facilitate the retention of sperm from the extra-pair partner.

If Baker and Bellis are right, indicators of a partner’s genetic fitness should be cues that lead women to have more frequent sperm-retaining orgasms during extra-pair sex. Thornhill, Gangestad, and Comer (1995) correlated women's frequency of orgasm during sexual intercourse with their current mate's FA, while controlling for his SES, anticipated future salary, age, and observer-rated physical attractiveness. Women experienced more orgasms if their partner was more symmetrical, r = -.27. More important, high sperm-retention orgasms (i.e., those occurring close to ejaculation; Baker & Bellis, 1995) were significantly correlated with the men’s degree of symmetry. Aside from their partner’s physical attractiveness, no other male feature correlated with the frequency of female orgasms. Analyses on a larger sample have replicated this effect (Møller, Gangestad, & Thornhill, in press). These findings are consistent with GGSS, and difficult to explain in terms of other theories. (For evidence that female orgasm favors dominant males in Japanese macaques, see Troisi & Carosi, 1998.)

4.5.2. Women’s olfactory preferences across their reproductive cycle. If women tend to choose extra-pair partners for their gametes, compared to sex with their primary partners, women's extra-pair sex should occur during the most fertile time of their reproductive cycle. Bellis & Baker (1990) found preceisely this pattern in a sample of British women. If women's patterns of extra-pair sex thus favor the sperm of extra-pair mates, this effect might be mediated by women’s preferences for specific attributes valued in extra-pair partners, especially those associated with symmetry.

Olfactory cues affect mate preferences strongly in many species (Alcock, 1993). Based on evidence that women’s olfactory sensitivities and preferences change across the menstrual cycle (Grammer, 1993), Gangestad and Thornhill (1998b) hypothesized that women should find the smell of more symmetrical men more appealing during ovulation. Men were measured for FA and then asked to wear a non-scented T-shirt for two nights. Women smelled each shirt (blind to all other characteristics of the men), and rated how attractive they found the odor of each shirt. Women’s fertility was estimated from their probability of conception when they participated in the study, based on medical data (Jöchle, 1973). Women taking the pill were excluded from the analyses. The shirts worn by more symmetrical men were rated as smelling better than those worn by less symmetrical men, but only among women who were in the fertile phase of their reproductive cycle, r = -.30. Indeed, women’s probability of fertility correlated .54 with their preferences for the scents of more symmetrical men. This effect has been replicated in a larger, separate sample (Thornhill & Gangestad, in press), where the correlation between fertility risk and preference for the scents of symmetrical men was .42. Statistically controlling for a number of factors (e.g., men’s number of showers) increased the effect size. These results confirm a very specific prediction that can be derived only from GGSS, and they provide further evidence about the proximate cues that may "advertise" male symmetry.

4.5.3. Women’s long-term and short-term mate preferences. Buss and Schmitt (1993) claim that the characteristics women prefer in long-term and short-term mates are quite similar. According to SST, women use short-term mating to evaluate men as potential long-term partners or for mate-switching. Women’s long-term and short-term mate preferences, however, are not identical. As discussed earlier, even Buss and Schmitt (1993) have found that women place greater emphasis on men’s physical attractiveness and physical prowess when evaluating them for possible short-term relationships. Scheib (1999) found that women place greater emphasis on physical attractiveness when evaluating men as extra-pair partners.

Gangestad, Simpson, Cousins, and Christensen (1998) examined women’s mate preferences by asking them to rate men who had been videotaped during an interview conducted by an attractive woman. After observing each man, women rated his attractiveness as a potential long-term mate and short-term mate (either as a one-time sex partner or as an extra-pair mate). Women’s sociosexual orientation (based on their SOI score; Simpson & Gangestad, 1991) was measured as well. Women with an unrestricted sociosexual orientation are more willing to have short-term relationships (i.e., they are more interested in short-term mating). Women with a restricted sociosexual orientation, in contrast, are less willing to have sex without commitment and emotional closeness (i.e., they are more interested in long-term mating; see Simpson & Gangestad, 1991, 1992).

Compared to restricted women raters, unrestricted women raters preferred more symmetrical men, particularly for short-term relationships. Indeed, unrestricted women’s short-term mate attractiveness ratings correlated .40 with men’s symmetry. Correlations between restricted and unrestricted women’s ratings of long-term mate attractiveness and men’s symmetry, as well as correlations between restricted women’s ratings of short-term mate attractiveness and men’s symmetry, were all non-significant. These findings indicate that more symmetrical men have (or display) features that are preferred in short-term mates by those women who are most willing and likely to engage in short-term mating—unrestricted women. They also clarify how women’s preferences may produce greater sexual success in more symmetrical men. These results would not be expected if more symmetrical men offered superior material benefits in long-term relationships.

According to the model we have presented, men’s allocation of effort to short-term mating during evolutionary history should have been contingent on their ability to satisfy the short-term mate preferences of women. Women’s short-term mate preferences, in turn, should have been influenced by GGSS. Thus, men’s tendency to engage in short-term mating should be a direct function of their genetic fitness (indexed by FA), while men’s propensity to invest in single, exclusive long-term relationships should be inversely related to their genetic fitness.

5.1. FA and Attitudes Toward Engaging in Casual Sex

Simpson and Gangestad (1991) developed the Sociosexual Orientation Inventory (SOI) to measure individual differences in willingness to engage in sex without closeness and commitment. If asymmetrical men have less desire to engage in short-term sex, as our model of conditional strategies predicts, men’s FA should correlate negatively with the SOI. As predicted, the mean correlation across several samples is -.20. Men’s SOI scores do not correlate with either the SES of their family of origin or their expected salary in 5 years.

5.2. FA and Investment in an Exclusive Relationship

In a sample of long-term dating couples, Gangestad and Thornhill (1998c) tested whether more symmetrical men tend to invest less in their ongoing relationships. Partners answered questions about their own and their partner's behavior in the relationship and then completed the Relationship-Specific Investment Inventory (RSI: Ellis, 1998). The RSI measures 10 sets of acts pertinent to the partner and relationship maintenance: Being Expressive/Nurturing, Being Committed, Giving Time, Being Sexually Proceptive, Investing Money, Being Honest, Providing Physical Protection, Being Attentive in Social Contexts, Having a Good Relationship with the Partner's Family, and Not Sexualizing Others. Some of these acts (e.g., giving time and attention to the partner) detract from mating effort outside the relationship. These acts, therefore, should correlate negatively with men’s symmetry.

Path analyses tested models in which both self-reports and partner-reports were treated as markers of men’s investment. When men's resource potential (indexed by their anticipated future salary), women's rated physical attractiveness, and women’s level of investment in the relationship were controlled, more symmetrical men provided less investment than did less symmetrical men, b = .26. More symmetrical men were particularly less honest with their partners, sexualized other women more, and spent less time with their partners. This pattern of lower investment should facilitate efforts to mate with women outside the current relationship.

The one component of investment that symmetrical men provided more of was physical protection, particularly their reported ability (as opposed to their reported willingness) to provide their current partner with greater physical protection. Because the ability to provide physical protection may involve the development of attributes that are also useful in intrasexual competition, providing this form of investment is not likely to interfere with men’s short-term mating. If men who possess markers of "good genes" do "compensate" their partners with some kind of material benefits, the most likely candidate is physical protection. Hrdy (1981) and Smuts (1985) have claimed that physical aggression from men other than fathers may have been one of the major dangers to offspring in the EEA . To reduce these threats, women may have either mated with multiple partners to confuse paternity or developed male-female friendships, particularly with socially and physically dominant men. Mesnick (1997) and Wilson and Mesnick (1997) have recently argued that women may have evolved to prefer protective males primarily to receive protection from other aggressive males, which could have enhanced women’s fitness directly.

5.3. Variation in Women’s Mating Strategies

Thus far we have concentrated on adaptive variation in men’s sexual tactics. However, women also vary considerably in their openness to, and willingness to engage in, short-term mating (Simpson & Gangestad, 1991). We now address adaptive reasons for this variation.

According to the model we have presented, the attributes that made men valuable as short-term mates in the EEA may have differed from those that made them valuable as long-term mates. Short-term or extra-pair mates should have offered genetic benefits that were often unavailable from long-term mates. Moreover, long-term mates who offered more genetic benefits may have provided fewer material benefits. Due to variation in their circumstances (discussed below), women should have differed in the extent to which they could have benefited from obtaining genetic versus material benefits. If some women could have appreciably improved their fitness by mating with men who offered better genetic benefits, these women should either have preferred long-term mates who had markers of genetic fitness or they should have engaged in short-term mating (especially opportunistic extra-pair matings) with males who had such indicators, even if it meant "trading-off" or risking the loss of material benefits they could have garnered from a long-term mate. Conversely, if other women could have enhanced their fitness by obtaining mates who provided (or could provide) superior material benefits, these women should have pursued long-term mates who were able and willing to provide the material benefits they most needed. Such women should have refrained from extra-pair mating, sacrificing some genetic benefits they might otherwise have obtained. The differential value of genetic versus material benefits across women should accordingly have produced adaptive variation in women’s mating tactics. (For a related discussion on avian mating, see Petrie & Kempanaers, 1998.)

5.4. Evidence that Differential Valuation of Men’s Attributes Predicts Women’s Mating

Earlier, we described a study examining women’s preferences for men’s symmetry in long-term and short-term mating contexts (Gangestad et al., 1998). It showed that women with a less restricted sociosexual orientation (i.e., more willing to have short-term sexual relationships) found symmetrical men more attractive, particularly as short-term mates. These results indicate that variation in women’s willingness to engage in short-term mating is associated with—and perhaps driven by—their stronger preference for indicators of genetic fitness in men.

Several additional lines of evidence also suggest that women who engage in short-term mating especially prefer men who display indicators of genetic fitness. First, women who are willing to engage in short-term mating (i.e., unrestricted women) care more about a man’s physical attractiveness than do women who are less willing to engage in short-term mating (Simpson & Gangestad, 1992). Second, when given a choice between dating a romantic partner who is very attractive but not highly reliable/loyal versus one who is highly reliable/loyal but only average in attractiveness, women who are more willing to engage in short-term mating (unrestricted women) tend to choose the more attractive/less unreliable male, whereas women who are less willing to engage in short-term mating (restricted women) tend to choose the less attractive/more reliable one (Simpson & Gangestad, 1992).

5.5. Environmental Factors And Sexual Strategies

Our model suggests that women should vary in their "exchange rate" between a partner's genetic fitness (indexed by FA) and his investment, whereas men should vary in their "exchange rate" between short-term mating effort and long-term parental investment. Environmental factors should influence women’s assessments of the exchange rate between a prospective male’s parenting qualities and his heritable fitness. In environments where biparental care was crucial to infant survival, male parenting qualities should have had more beneficial effects (Andersson, 1994) Conversely, in environments with prevalent pathogens, male genetic fitness may have had more beneficial effects (partly due to the importance of pathogen resistance and partly due to the decreased marginal value of heavy parental investment when mortality rates are high; see Kaplan, 1996). If ancestral women were repeatedly exposed to these contrasting environments, they should have evolved to make trade-offs between investment qualities and indicators of good genes contingent on specific environmental conditions. Factors that influenced this exchange rate may have differentiated (a) populations of individuals, producing differences in the mating systems across different groups of women (i.e., within different cultures; see Tooby & Cosmides, 1992, on "evoked culture"), and (b) women within a population, producing differences in the mating preferences and behaviors of women in a given group of people.

5.6. Environmental Factors Producing Differences Between Populations

5.6.1. Factors influencing the value of men’s genetic quality. Several factors may have affected the value of men’s genetic fitness. According to principles of host-parasite coevolution (Hamilton, 1982), pathogen prevalence should have been one critical factor. In environments where pathogens were prevalent, women should have benefited more from mating with men who had good genetic qualities that made them more pathogen resistant. Although additional work is needed, several lines of research are consistent with this notion.

First, besides conveying information about female fertility, physical attractiveness may also contain cues about a person’s health, pathogen resistance, and perhaps genetic fitness (Symons, 1979). Gangestad and Buss (1993) tested whether preferences for attractive mates might have evolved through parasite-driven sexual selection. Men and women from 29 countries around the world rated the importance of several mate attributes, including their preference for a "good-looking" mate (see Buss, 1989). Pathogen pressure at each geographical location was estimated from information about the prevalence of 8 different parasites (see Low, 1990a). Across the 29 countries, both men and women in regions containing more pathogens placed greater importance on a prospective mate’s attractiveness, even when latitude, geographical region, and mean income were controlled, partial r = .76. This finding does not of course demonstrate that host-parasite coevolution necessarily influenced sexual selection in humans. Attractiveness could have been more important in regions with more parasites because of direct selection against mating with diseased individuals (Kirkpatrick & Ryan, 1991).

Second, women in pathogen-prevalent environments should be more willing to trade off features associated with exclusive parental care. Indeed, across the 29 cultures, parasite prevalence correlated negatively with the mean ranked importance of four attributes relevant to direct and exclusive parental care: "dependable character," "pleasing disposition," "emotional stability and maturity," and "desire for home and children", r = -.41 (Gangestad, 1993). Thus, individuals place less weight on these attributes in environments containing more pathogens. Whether these findings reflect the increased value of physical attractiveness or an independent devaluation of these male attributes remains unclear.

Third, in environments where pathogens are more prevalent, women should trade off indicators of good genes for exclusive paternal investment. In other words, a higher degree of polygyny should be seen in pathogen-prevalent environments (see Gangestad, 1993). Low (1990a) correlated the degree of polygyny with parasite-prevalence in nearly 200 societies of the Standard Cross-Cultural Sample (Murdock & White, 1969; Low, 1988, 1990a). With latitude and geographical region held constant, polygyny was more common in societies where pathogens were more prevalent. Hence, variation in parasite prevalence might be partially responsible for the variation in desired mate attributes and associated mating strategies across cultures (see Tooby & Cosmides, 1992).

5.6.2. Factors influencing the value of men’s parental effort. Several factors should also have influenced the influence of men’s parental effort on infant mortality and their later reproductive success. For example, in environments where the primary causes of infant mortality were infectious diseases (rather than inadequate parental care), paternal effort should have had less impact on offspring fitness. In contrast, when infant viability was strongly tied to biparental care (e.g., when women could not provide all of the nutritional needs for their offspring), paternal care should have had greater impact on offspring fitness.

Women’s access to resources should also have influenced their need for—and the value of—male parental investment (Gowaty, 1992a,b). Because parental investment often has diminishing marginal returns (see Cashdan, 1993), men’s resources may have been less important when women had sufficient resources of their own. Hence, polygyny should be more prevalent in societies in which women have more access to resources. Low (1990b) correlated indicators of women's ability to care for both themselves and their infants independently with measures of polygyny across nearly 200 cultures in the Standard Cross-Cultural Sample. As predicted, polygyny is more common in cultures where women have more control over resources.

We do not know how women's control over resources in ancestral environments affected their mate preferences, but their participation in current economies does predict the importance women place on physical attractiveness in men. In the cultures surveyed by Buss (1989), women's mean preference for physical attractiveness in a mate was positively correlated with the proportion of women who were involved in the economy. However, women's preferences for qualities related to parental care did not correlate with their economic participation (Gangestad, 1993).

Across the cultures surveyed by Buss (1989), Eagly and Wood (in press) have found that women’s preference for men’s earning potential is predicted by a measure of women’s "empowerment" (United Nations Gender Empowerment Measure; United Nations Development Programme, 1995). As women’s empowerment (indexed by their earnings, their representation in legislative government, and their involvement in professional positions) increases relative to men in cultures, women place increasingly less value on the earnings of a mate. However, women do not place less emphasis on men’s physical attractiveness (nor do men place less emphasis on women’s attractiveness) as women’s empowerment increases. Although Eagly and Wood interpret these findings in terms of domain-general behavioral processes and responses to gender roles, they are also consistent with an ecologically-contingent, conditional mating strategy in which women moderate their mating tactics and preferences in response to specific environmental inputs (e.g., women’s control of resources; see Low, 1990b).

5.6.3. Effects on men’s tactic choice. How women evaluate men and how they make trade-offs should influence which mating tactics men adopt. In environments where male parenting qualities are needed and valued, women should be less likely to engage in short-term and extra-pair mating. In response to this, men should devote greater effort to parental investment, and variance in men's reproductive success should be reduced (i.e., a larger proportion of men should have offspring). On the other hand, in environments where men’s genetic fitness is needed and valued, women should be more willing to pursue short-term mating. Consequently, men should devote greater effort to short-term and extra-pair mating, and variance in men's reproductive success should increase (see Low, 1990a,b).

In environments where biparental care is not as crucial, even men with lower genetic fitness might benefit from channeling some of their effort to short-term and extra-pair mating (perhaps by attempting to "deceptively" advertise genetic quality). Under these circumstances, men's increased efforts to display their "quality" could result in an escalated war of attrition (Maynard Smith, 1982), which could be facilitated by testosterone-based somatic growth and aggressive behavioral displays. Thus, the fiercest intrasexual competition among men may not occur over resources relevant to parental investment, as SST might predict. Rather, it may occur when status displays that signal a male’s genetic fitness are contested (see Daly & Wilson, 1988, for a discussion of how men's face-saving tactics often lead to homicide).

5.7. Environmental Factors Producing Differences Within Populations

5.7.1. Differences due to differential exposure to cues signaling the value of parenting. If some women are exposed to environments that require paternal investment whereas others are not, within-population differences in mating preferences and tactics should emerge. Belsky, Steinberg, and Draper (1991) have proposed that patterns of parental care experienced during childhood may affect adult mating tactics (see also Chisholm, 1996; Simpson, 1999). Insufficient parental responsiveness to their needs as infants (possibly due to harsh, demanding environments) should lead individuals to adopt short-term mating tactics better suited for future environments in which stable pair-bonds are not needed or anticipated. Conversely, adequate parental responsiveness should result in long-term adult mating tactics more appropriate for environments in which biparental care and stable pair-bonds are needed and expected.

5.7.2. Differences due to women's phenotypic qualities. Women's personal characteristics may also lead them to value different qualities in mates. In many circumstances, parental investment has decreasing marginal returns: the more investment an offspring has received, the less adaptive the next "dose" (see, e.g., Cashdan, 1993; for important exceptions, see Kaplan, 1996). Especially when men’s and women’s investment in offspring is similar (rather than complementary) in form, the more investment a woman can provide on her own, the less need for additional paternal investment. Consequently, women may have evolved to focus more on a man’s genetic fitness when their own access to independent resources increased (Gowaty, 1992a), making them more willing to engage in short-term mating.

Correlations between women's sociosexual orientation and their personality traits provide indirect support for these conjectures. Unrestricted women (who are more likely to engage in short-term mating) tend to be more socially dominant, more extraverted, and less harm-avoidant (Gangestad & Simpson, 1990). Each of these traits should facilitate the acquisition and maintenance of independent resources. Hrdy (1981) and Gowaty (1992a) have suggested that women's access to resources through cooperation, competition, and social influence should have been an integral component of their fitness in the EEA. These personality traits may have helped unrestricted women to acquire their own resources and, thus, to pursue short-term mating tactics.

Based on these notions, one might expect that the value women ascribe to their mates’ resources to diminish as their own access to resources increases. This association has not been found: Women with high paying jobs tend to value resources as much as or more than women with lower paying jobs do (e.g., Buss, 1989; Townsend, 1989; Wiederman & Allgeier, 1992). These findings contrast with those of Eagly and Wood (in press), who found that, across cultures, women’s access to resources and power is associated with a lower rated importance of mates’ financial success. Because what women learn about the value of a mate’s resources is shared within a culture, the effects of women’s access to resources within a culture may differ from the effects of women’s access to resources across cultures (see Eagly & Wood, in press). Another possibility is that Eagly and Wood’s finding is not due to women’s access to monetary resources (only one component of the Gender Empowerment Measure they used) but to their access to power. Power and the ability to affect outcomes through social influence may have been important facets of women’s circumstances to which they evolved contingent strategies, whereas their ability to accumulate wealth in modern cash economies may not have been.

In sum, environmental factors should influence women’s mating tactics. Pathogen prevalence, for example, should increase the value women place on men’s genetic quality, whereas the need for biparental care should increase the value they place on men’s parental effort, particularly women who do not have independent access to resources. Depending on environmental conditions, therefore, women should make trade-offs between male genetic quality and parental investment in adaptive, ecologically contingent ways. Women should vary in the exchange rate between men’s genetic quality and their long-term investment, whereas men should vary in the exchange rate between short- and long-term mating. As a result, most men should adjust their mating strategies in response to what women value.

Mating tactics are highly variable in both men and women and evolved to be contingent on environmental factors. Complete theories of mating strategies must account for these individual differences and contextual effects. We have proposed that these phenomena cannot be fully understood without considering the nature of the trade-offs that underlie mating decisions in humans. We suggest that good genes sexual selection, in concert with good parenting sexual selection, may have generated the variation and contextual effects associated with the short- and long-term mating tactics witnessed in both sexes.

Given the demands of biparental care during evolutionary history, both men and women were selected to use long-term mating tactics and invest in offspring. However, they were also selected to use ecologically contingent, conditional mating strategies, dedicating some effort to short-term and extra-pair mating under specific conditions. Women may have evolved to trade-off evidence of a man’s genetic fitness for evidence of his ability and willingness to invest in offspring. The specific mating tactics and preferences women adopted, however, depended on the nature and quality of their local environment. If the local environment was difficult and demanded biparental care, women placed more weight on the investment potential of prospective mates and less weight on indicators of their genetic fitness. As a result, a larger proportion of women adopted long-term mating tactics almost exclusively. If, on the other hand, the local environment was prevalent with pathogens (or signaled the importance of the genetic fitness of offspring), women placed more weight on indicators of the genetic fitness of prospective mates. In such environments, a larger proportion of women were willing to engage in short-term, extra-pair matings, allowing them to gain genetic benefits from men who provided less parental investment at the risk of losing parental investment from their primary mates. The mating tactics and preferences of women accordingly reflected the nature and quality of the environments in which they lived.

Whereas women "tracked" their environment, men tracked and adjusted their mating tactics and preferences to the behavior of women (Thiessen, 1994). If most women expected heavy paternal investment, most men (especially those who displayed less fitness) offered more and perhaps exclusive parental investment, dedicating a greater portion of their effort to long-term mating tactics and parental investment. As a result, variance in men’s mating success was reduced. If women’s "demand" for genetic benefits increased, some men (especially those advertising such benefits) dedicated more effort to short-term, extra-pair mating tactics, thereby increasing variance in mating success among men. Only a small proportion of men (i.e., those who displayed the most fitness) were able to carry out short-term tactics successfully at all times, regardless of the environmental factors to which women were responding.

Many of the unique predictions derived from this model have been supported by recent empirical data. Although our notions extend our understanding of the strategic plurality of human mating in many ways, this account of the ties between evolutionary theory and human mating strategies remains far from complete.

1. Not all environment-response linkages are evolved. Many are learned. Learning, however, can reflect a set of implicit, evolved decision rules (see Mayr, 1974).

2. In certain species, males engage in more parental effort than females, in which case sexual selection pressures operate more strongly on females than males (Trivers, 1972). The selection pressures that lead to different amounts of parental effort for the sexes are not yet fully understood (for one treatment, see Parker, Baker, and Smith, 1972). Moreover, as we will discuss, there are circumstances in which mating with multiple mates can benefit females even in species in which they invest in offspring more than males do (see also Hrdy, 1981).

3. These effect sizes may be attenuated by unreliable measurement, particularly for single-item measures. Our 3-item measure of attitudes toward casual sex has a reliability of nearly .8 (Simpson & Gangestad, 1991). If sex accounts for 8%-20% of the variance in this measure, it should explain 10%-25% of the variance in an error-free measure. Single items in this domain may have as little as 50% reliable variance (estimated using the Spearman-Brown formula; Anastasi & Urbani, 1998). Hence, sex may account for 20%-40% of the reliable variance underlying single-item measures of optimal number of future sex partners and willingness to have sex with an unknown partner. Naturally, the reliable variance in these measures may not all be valid, as some may reflect differences in social desirability responding. The sex difference in Clark and Hatfield’s (1989) naturalistic field study of willingness to engage in unsolicited sex with a stranger (using a measure less susceptible to social desirability than lab measures) was one of the largest effects ever documented, accounting for about 60% of the variance. One possible reason this sex difference is so large is that men are willing to accept unsolicited sex even when they are not exerting effort to obtain short-term mates. The robustness of this result, however, needs to be replicated. Moreover, this sex difference may partly reflect differences between men and women in the fear of physical harm from opposite-sex strangers rather than differences in interest in short-term mating per se (see Hrdy, 1997). Overall, although sex accounts for a large amount of the variance in measures of interest in and willingness to actively pursue short-term mating, it appears to explain less variance—in some cases substantially less—than occurs within each sex. (For comparison purposes, sex accounts for about 50% of the variance in adult height.) Moreover, the distributions of men’s and women’s interest in pursuing short-term, opportunistic matings appear to overlap substantially.

4. The additive genetic CV is evolutionarily relevant because the potential rate of evolution ("evolvability" or the proportionate change in a character’s mean in the population per generation) is a function of the absolute amount of genetic variance (specifically, the square of the additive genetic CV). The sheer amount of environmental variance in a trait does not affect the rate of the trait’s evolution. The heritability of a trait is its genetic variance divided by total (genetic plus environmental) variance. Because this measure standardizes a trait’s genetic variance (relevant to its evolvability) in relation to its environmental variance (not relevant to its evolvability), heritability is a less evolutionarily relevant index than is the additive genetic CV. Fitness characters tend to have lower heritabilities (around .25 on average; Mousseau & Roff, 1987) than ordinary morphological traits do, but not because they have low genetic variance. Rather, they tend to have very high genetic variance as well as very high environmental variance. Because fitness characters have very high genetic variance, they have more potential to evolve than ordinary morphological traits do, despite lower heritabilities (Houle, 1992).

5. The fluctuating asymmetry of an individual trait (e.g., ear width) is often a very weak indicator of underlying developmental imprecision (Gangestad & Thornhill, 1999). Even the composite FA measure (which consists of 10 aggregated traits, and was used in many of the human studies we cite) is estimated to correlate only about .5 with underlying developmental imprecision. This makes the modest associations between FA and other traits (e.g., mating success; Møller & Thornhill, 1998) all the more impressive.

6. It could be argued that women find more symmetrical men more appealing because they are looking for long-term mates who will provide them (and subsequent offspring) with greater material benefits. Even though more symmetrical men may be more capable of providing certain benefits (e.g., physical protection), the reproductive cycle studies indicate that women find the smell of more symmetrical men more desirable primarily when they are ovulating (that is, when gaining material benefits is not an issue). From the standpoint of adaptive design, these data strongly suggest that women find symmetrical men more attractive than less symmetrical men for reasons beyond potential resource acquisition.

7. We are not suggesting that obtaining "good genes" from certain men is the only possible fitness benefit of short-term mating available to women. Buss and Schmitt (1993), for example, have proposed that some women may use short-term mating in order to evaluate and attract long-term mates. The evidence we present, however, provides support for the former function.

8. Because women also vary in their relative mate value, they should differ in the amount of benefit they receive from men. Certain highly valued women may be able to obtain both high material benefits and high genetic benefits. For most women, however, the genetic benefits and exclusive investment benefits that can be obtained from mates should correlate negatively within the most desirable set of mates a women can attract (Gangestad, 1993).

9. Previously, we suggested that women’s relative evaluation of genetic benefits and material benefits may partly reflect genetically polymorphic, alternate strategies (see Gangestad & Simpson, 1990). Although this possibility remains plausible, variation in women’s mating strategies may be more strongly governed by environmental conditions that moderate the relative value of genetic versus material benefits.

10. Low (1990a) suggests that healthy men in pathogen-prevalent regions may have multiple mates because they can provide better paternal care. We suggest that men who have better fitness indicators should invest less in their offspring than men with poorer indicators.