The Evolution of Male-Male Sexual Behavior in Humans

Frank Muscarella. Journal of Psychology & Human Sexuality. Volume 18, Issue 4. 2007.

The mother of Gilgamesh, Lady Ninsun, the wise, the all-knowing, said to her son, ‘Dearest child, this bright star from heaven, this huge boulder that you could not lift- it stands for a dear friend, a mighty hero. You will take him in your arms, embrace and caress him the way a man caresses his wife. He will be your double, your second half, a man who is loyal, who will stand by your side through the greatest dangers. Soon you will meet him the companion of your heart. Your dream has said so.’ Gilgamesh said, ‘May the dream come true. May the true friend appear, the true companion, who through every danger will stand at my side.’

—Mitchell, 2004, pp. 83-84

Difficulties in the Scientific Study of Homosexuality

Researchers often lament the lack of agreement in defining the con­structs they study, but “homosexuality” is an unusually ill-defined term. This lack of a consensus definition is a recurring impediment to the sci­entific study of homosexuality (Kauth, 2002, 2005; Muscarella, 1999; Stein, 1999). The term “sexual orientation” is also protean. Shively, Jones, and DeCecco (1984) analyzed the conceptual and operational definitions of sexual orientation in 228 articles from 47 different scien­tific journals and concluded that the great variation found in these defi­nitions reflected an underlying conceptual confusion. Similarly, Gonsiorek and Weinrich (1991) noted that the confusion surrounding the concept of sexual orientation reflects a lack of basic construct va­lidity. The difficulty in conceptualizing the phenomenon surely stems from the great variation in the sexual behavior and interests of people typically classified as “homosexual” (Bell & Weinberg, 1978; Shibley Hyde & DeLamater, 2006). The operational definition of sexual orien­tation is an ongoing problem in sex research (Kauth, 2005).

A second difficulty in the study of homosexuality is that the scientific community is not in agreement about the development of a homosexual orientation. Currently, scientists cannot predict the development of the sexual orientation for any given individual (Bancroft, 1989; Shibley Hyde & DeLamater, 2006). The best consensus among theorists since Freud (1962) has been that a homosexual orientation is the product of an interaction between genetic, cultural, developmental, and psychologi­cal factors (Bancroft, 1989; Greenberg, 1988; Kauth, 2000; McKnight, 1997; Shibley Hyde & DeLamater, 2006). The single-factor “gay gene” theory of homosexuality has been criticized as overly simplistic and insufficient (Stein, 1999).

Given the aforementioned difficulty with the construct of sexual orien­tation, it is ironic that the third problem in the scientific study of homosexu­ality is that same-sex sexual behavior not associated with an exclusively or predominantly homosexual orientation is considered not “real” homo­sexual behavior and excluded from study (Muscarella, 1999). A valid examination of homosexuality must include all of its manifestations. A number of researchers have argued that for most humans across his­tory some admixture of heterosexual and homosexual behavior has been the norm and that most human male homosexual behavior has been exhib­ited by males who are not exclusively homosexual (Greenberg, 1988; Kirkpatrick, 2000; Murray, 2000; Muscarella, 2000). Thus, the assumption that all humans can be neatly categorized as heterosexual and reproductive, or homosexual and nonreproductive, is not supported by the evidence (Kauth, 2002; Kirkpatrick, 2000; Muscarella, 2000).

Contemporary Western culture and Western scientific models of sexual orientation make the assumption that attraction to males is a feminine characteristic, not a masculine characteristic. This conceptual­ization ignores abundant evidence demonstrating that, in many cultures across historical periods, sexual attraction to and interaction with other males has been considered a masculine characteristic essential to mas­culine development (Dover, 1978; Greenberg, 1988; Herdt, 1981; Murray, 2000; Percy, 2005). The conceptualization of attraction to males as a uniquely feminine characteristic assumes that it is abnormal when it occurs in males. I argue that this assumption contributes to an incorrect understanding of the actual nature of human male sexual expression. Evolutionary psychologists hold that recurring and com­mon patterns of human behavior may have evolutionary significance (Tooby & Cosmides, 1992). Same-sex sexual behavior not associated with a predominantly homosexual orientation may be an important and evolutionary significant behavior pattern in human male sexuality.

In summary, homosexuality lacks a reliable operational definition. The development of a same-sex sexual orientation appears to be strongly influenced by non-genetic factors, and its occurrence cannot be reliably predicted in individuals. In addition, the historical record sug­gests that most same-sex sexual behavior among human males has been exhibited by those who are not exclusively homosexual. In view of these considerations, one might reasonably ask, “Is it possible to have a good evolutionary theory of homosexual orientation?” At present, the answer is, “No.” Consequently, theorists have argued that the evolu­tionary study of homosexuality is better served by a focus on same-sex sexual behavior and not homosexual orientation. Same-sex sexual be­havior is much more reliably operationalized and amenable to evolu­tionary analysis (Kirkpatrick, 2000; Muscarella, 2000; Muscarella, Fink, Grammer, & Kirk-Smith, 2001). An understanding of the evolu­tionary history of same-sex sexual behavior would be useful to theorists studying the development of homosexual orientation in contemporary individuals. Finally, in view of the ambiguities associated with the term homosexual and the varied and unspecified assumptions linked to it (Kauth, 2005), I will shift to the operationally simpler term male-male sexual behavior in my own treatment of the topic where possible in the rest of the text.

Revealing Assumptions

Human Sexuality

As Kauth (2005) has clearly indicated, the variation in the opera­tional definitions of sexuality-related terms and concepts and the unex­amined assumptions of researchers is dizzying. Further, this variation perpetuates confusion in sex research and an unreliable body of knowl­edge. In this paper, the operational definitions of the sexuality-related terms used demonstrate a focus on the behavioral and observable as­pects of sexuality and parsimony in interpretations of behavior and ap­plication of theory. Borrowing the description provided by Shibley Hyde and DeLamater (2006), sexual behavior is defined as “behavior that produces arousal and increases the chance of orgasm” (p. 3). Sexual attraction is defined as the psycho-emotional reaction to a stimulus which motivates sexual behavior. Borrowing from Kauth (2005), sex is not sexual behavior but instead refers to the biological and physiologi­cal characteristics that differentiate males and females. Sex is conceptu­alized as bimodal. This assumes that most people are clearly male or female, but some will be intermediate or indeterminate. Gender is conceptualized as the social characteristics and roles assigned to males and females by the culture.

Borrowing from Ellis and Mitchell (2000), sexual orientation is con­ceptualized essentially as “… a person’s relatively consistent and persis­tent directedness toward some thing or activity for sexual gratification” (p. 197). I also agree with their argument that individuals may have mul­tiple sexual orientations expressed either in behavior, fantasy, or both. I join the consensus among sex researchers that sexual orientation is the product of an interaction between genetic, cultural, developmental, and psychological factors. In view of this, I agree with Kauth’s (2005) argu­ment that classification of sexual orientation is almost exclusively linked to sex of the actors in contemporary Western culture and scien­tific theory and is an artifact of our culture. Kauth notes that in other his­torical periods and in other cultures diverse factors besides sex of the actors played a role in the development of sexual attraction and the con­ceptualization of sexual orientation. This is a very important point be­cause, as I will show later in this text, those other factors are not random. Rather, they are predictable and phylogenetically relevant and, inas­much, may provide insight into the evolutionary roots of male-male sexual behavior.

All said, I assert that our culture’s largely inescapable mandate that sexual orientation be linked to the sex of the actors results in a cultur­ally-shared and culturally-lived experience of sexual orientation. This sexual orientation is essentially bipolar, implying a continuum of attrac­tions between exclusive other-sex attraction and exclusive same-sex at­traction, and its measurement is best reflected by the Kinsey scale (Kauth, 2005). For most lay people and researchers, this results in a trichotomy of sexual orientations: heterosexual (attraction to the other sex), homosexual (attraction to the same sex), and bisexual (attraction to both sexes). My own reading of the literature suggests to me that al­though some researchers and lay people try to conceptualize sexual ori­entation out of these three categories, the cultural momentum of thinking and relating to others almost always pulls conceptualization back to these three discrete categories. And indeed, there appears to be documented evidence of this (Kauth, 2005). This bias undoubtedly con­tributes to the difficulty sex researchers have had in producing a reliable body of knowledge about sexual orientation. I believe that, in theory, humans can develop diverse sexual orientations, but in our cultural real­ity there are only three: other-sex oriented, same-sex oriented, and both-sexes oriented. In my own writing, I will try to avoid use of the terms homosexual, heterosexual, and bisexual because their meanings have been expanded to include a number of social and cultural issues that impede a cross-cultural study of sexual attraction. Consistent with other theorists (e.g., Herdt, 1997), I try to avoid the terms gay and lesbian because they are culturally-related terms that involve social as well as sexual concepts.

From my point of view, it seems that, for the average human male, sexual behavior is motivated by sexual attraction that derives from a va­riety of sources. These include biological contributions, such as testos­terone (Dabbs & Dabbs, 2000), but also psychological, emotional, and cognitive processes (Everaerd, Laan, & Spiering, 2000). All of these factors work together in a complex and, at this point, poorly understood manner. The object or objects of sexual interest are seen as components of the individual’s sexual orientation, which can be conceptualized more broadly as a major element of the individual’s love map, as de­scribed by Money (1986). It would also seem that for the average hu­man male the ideal partner or partners embodied by the love map would not be available. Consequently, he will engage in sexual behavior with whatever partner is available that best approximates his ideal. At the broadest level, he is likely to differentiate by sex of the partner and at­tempt to access the desired sex. I will argue later in the paper that when the partners are other males, other differentiations between partners are based on sociosexual dominance status, with three recurring, phylogenetically linked, sociosexual markers: age, sexual role, and social status. Currently, I think that in most cases the characteristics that most closely approximate the characteristics of the ideal partner are predominantly physical or behavioral signals of submissiveness or dominance, which trigger sexual interest. Finally, while fantasy may be unconstrained, sexual behavior will be affected by cultural, religious, societal, legal, and logistical factors. These essentially uncontrollable variables are what complicate the study, understanding, and prediction of human sexual behavior.

Evolutionary Psychology

Just as Kauth (2005) has indicated that there is confusion in the use of terminology in research in human sexuality, Buss, Haselton, Shackel­ford, Bleske, and Wakefield (1998) have indicated that there is similar confusion in terminology and its application in evolutionary psychol­ogy. According to Buss and colleagues, evolutionary psychology has attempted to integrate principles from evolutionary biology into the study of psychological phenomena. The central element is Darwin’s theory of evolution through natural selection. There are several prod­ucts of the evolutionary process: adaptations, by-products, noise, and exaptations. An adaptation is “an inherited and reliably developing characteristic that came into existence as a feature of a species through natural selection because it helped to directly or indirectly facilitate re­production during the period of its evolution” (p. 535). Adaptations are conceptualized as having solved some adaptive problem during the course of evolution: that is, problems affecting survival and, ultimately, reproduction. They have a functional design. A common example is the umbilical cord. By-products are characteristics that do not solve adap­tive problems and do not have functional design. Rather, they are car­ried along because they are closely linked to adaptations. An example of a by-product is the belly button. It serves no adaptive function but is the recurring by-product of an adaptation, the umbilical cord. Noise is de­fined as a random effect. For example, the shape of a belly button is due to a number of random effects including the manner in which the umbil­ical cord was cut, the manner in which the cut healed, the person’s weight, and the person’s age. Finally, exaptations are defined as charac­teristics that are adaptive, although they did not originate for their cur­rently identified function. They may have originated in one of two ways. They may have originated as neutral by-products that were co-opted by evolution because they helped to solve an adaptive prob­lem. Or, they may have originated as an adaptation for one problem and were later co-opted by evolution because they solved a second, separate problem.

When evolutionary psychologists analyze behavior, they are in es­sence arguing that the behavior under consideration is best conceptual­ized as one of the four products of the evolutionary process. Theorists disagree about the function of even fundamental human behavior, such as language (Buss et al., 1998). Buss et al. (1998) have indicated that disagreement may arise from a lack of understanding of the terms used and a lack of consistency in the way that terms are applied. In an attempt to foster better research in evolutionary psychology, Buss et al. have identified several criteria to assess the quality of a good evolutionary hypothesis. These are summarized as follows: (1) precision and internal consistency, (2) coordination with recognized processes in evolutionary biology, (3) ability to generate new and specific predictions, (4) greater parsimony than competing explanations for existing data and greater evi­dentiary power, and (5) associated with the psychological mechanism in­volved, computational capability for solving the hypothesized problem.

The evolutionary analysis of same-sex sexual behavior in humans is constrained by conceptual inadequacies and definitional confusion aris­ing from both the fields of human sexuality and evolutionary psychol­ogy. Confidence in the reliability of any one analysis may be improved by linking the analysis to standardized terms and criteria, as suggested by Buss et al. (1998) and Kauth (2005). In this volume, three authors take on the analysis of the evolution of same-sex sexual behavior in hu­mans and propose three different theories and come to two different conclusions. Vasey and Diamond argue that same-sex sexual behavior is best conceptualized as a by-product (but for different reasons), and I argue that it is best conceptualized as an exaptation. In view of the is­sues described by Buss et al. (1998), I will link my analysis to the crite­ria they set forth, in addition to those indicated by Kauth (2005).

A second issue related to the evolutionary analysis of human behav­ior involves the process of analysis itself. Buss (2004) describes two strategies of analysis. One is the theory-driven or top-down strategy: (1) Derive hypotheses from existing theory; (2) test predictions based on hypothesis; and (3) evaluate whether empirical results confirm pre­dictions. The other strategy is the observation driven or bottom-up strat­egy: (1) Develop hypotheses about adaptive function based on a known observation; (2) test predictions based on the hypothesis; and (3) evalu­ate whether empirical results confirm predictions. The bottom-up strat­egy can be used to develop theory about the evolution of specific human behaviors that can then lend itself to a top-down strategy. I argue that at this point in time the evolutionary analysis of same-sex sexual behavior is by default bottom-up, since the development of an evolutionary the­ory has just begun. The lack of theory appears to be related to the dis­comfort and censure caused by the taboo nature of same-sex sexuality, as well as its presumed non-adaptedness. There is documented preju­dice, distortion, and active suppression of data regarding same-sex sex­ual behavior in scientific fields that would contribute to such theory, including biology (Bagemihl, 1999), primatology (Vasey, 1995), cul­tural anthropology (Herdt, 1997), physical anthropology and history (Taylor, 1996), and evolutionary psychology (Muscarella, 1999).

In general, the evolutionary approach to human behavior focuses on why a behavior exists. As indicated earlier, a behavior may be consid­ered adaptive if it can be demonstrated that during evolution the behav­ior solved an adaptive problem by contributing to reproduction either directly or indirectly (Buss et al., 1998; Tooby & Cosmides, 1992). Consistent with the bottom-up approach, Tooby and Cosmides (1992) indicate that recurring patterns of behavior among humans suggest evo­lutionary significance. Thus, cross-cultural as well as cross-species data are used to generate and support evolutionary hypotheses (Buss, 2004; Diamond, 1992; Symons, 1979). The evolutionary approach assumes that behavior that was adaptive in the evolutionary past may be discour­aged or maladaptive in contemporary societies (Buss, 2004; Tooby & Cosmides, 1992) and local cultural forces (Tooby & Cosmides, 1992) as well as ontogenetic or personal developmental factors (Buss et al., 1998; Lickliter & Honeycutt, 2003) affect the manifestation of evolved behavioral dispositions.

Evolutionary Theories of Homosexuality

Theorists have assumed that traits that are biologically maladaptive rarely exceed an incidence of 1% in the population (Werner, 1998). However, the large National Health and Social Science Survey esti­mated that the prevalence of exclusive homosexuality among American males is about 2%, and the inclusion of those with a predominant same-sex attraction raises the number to almost 3% (Laumann, Gagnon, Michael, & Michaels, 1994). Further, after the advent of legalized gay marriage in Great Britain, the British government reported that 6% of its population identified as gay or lesbian (Campbell, 2005). These num­bers are quite high given that same-sex sexual behavior is not directly associated with reproduction. The high prevalence of predominant and exclusive same-sex sexual behavior supports the appropriateness of an evolutionary analysis, and there are several hypotheses about the evolution of homosexuality.

One theory of homosexuality by E. O. Wilson (1975, 1978) holds that male homosexuality may be related to altruism and male bonding. Wilson hypothesized that since exclusively homosexual individuals would not have reproduced they may have assisted siblings and other relatives who would have then reproduced more successfully than indi­viduals without homosexual relatives. Thus, genes for same-sex sexual behavior would have been propagated indirectly by relatives. Wilson’s theory has been criticized for a number of its underlying assumptions and for lack of supporting evidence (Bancroft, 1989; Bobrow & Bailey, 2001; Dizinno, 1984; Greenberg, 1988; Rahman & Hull, 2005).

A second theory by Miller (2000) holds that homosexuality is a polygenic trait that results from selection for a number of characteristics that each contribute to fitness. According to Miller, male-male attrac­tion results from a shift in male brain organization to a more feminized direction allowing males to exhibit characteristics like tenderness, kind­ness, and empathy that would have made ancestral males better fathers and more attractive mates. Miller has suggested that exclusive homo­sexuality results from random genetic distribution, such that some men inherit a high number of these genes that then cause brain organization to develop in an extremely feminine direction, ultimately giving rise to a homosexual orientation.

There are several problems with Miller’s theory. First, the theory ad­dresses only exclusively or predominantly homosexual males and does not explain the large amount of male-male sexual behavior exhibited by other males. Second, the underlying assumption is that attraction to other males is a feminine characteristic, not a masculine one. Third, Miller’s theory of homosexuality is not a parsimonious explanation of the evolution for caring, sensitive, and fathering behavior in human males. Kindness and empathy appear to be part of a general trend in pri­mate evolution, and caring behavior by adult males toward juvenile conspecifics is common among many nonhuman primates in the absence of exclusive same-sex sexual behavior (de Waal, 1996).

Finally, Miller’s theory is not consistent with current evolutionary theory. Older theory held that pair-bonding in humans co-evolved with infant dependency and the mother’s need for assistance (Fisher, 1989; Lovejoy, 1981) and supported the idea of selection for characteristics to promote fathering behavior. However, current theory suggests that hu­man males pair-bonded with females to prevent infanticide of their off­spring by other males (de Waal, 2001), suggesting, if anything, that selection for aggressiveness and jealousy should be associated with mate guarding because of the increased need for paternity certainty (Buss, 2000). Further, infant dependency in humans appears to have co-evolved with a more complex social organization that provided nu­merous individuals, not just the father, available to assist the mother (Ellison, 2001; Kauth, 2000; Parker, 2000). Thus, the human male’s ability to care for children and mates is a by-product of general human attributes, such as intelligence and compassion (cf., Parker, 2000).

Miller has further speculated on the purported relationship between birth order and homosexuality in men, citing a number of studies by Blanchard and colleagues (see Miller, 2000). These studies suggested an increasing probability of development of a homosexual orientation with each additional older brother. Miller presented an immunization theory originally proposed by Blanchard and Bogaert (1996b) and Blanchard and Klassen (1997) as the mechanism for this phenomenon. This immunization theory holds that mothers carry a “biological mem­ory” (in the form of an H-Y antigen) of the number of sons they have gestated, which leads to changes in the intrauterine environment and re­sults in the feminization of later born males. According to Miller, youn­ger sons, being more sensitive and flexible, would compete less with older brothers for resources. In this scenario, exclusive homosexuality is a rare by-product of genetic variability for the capacity to flexibly exploit various social niches, outweighing the cost of a few individuals who do not reproduce.

Bearman and Bruckner (2002) have noted that the studies that sup­port the birth order effect have been criticized for a number of method­ological problems, including reliance on nonrepresentative samples and indirect reports of siblings’ sexual orientation. They stated that Miller’s theory also lacks the parsimony that makes it a useful and likely explanation. They described the biological memory mechanism as “somewhat far-fetched.” Further, Bearman and Bruckner’s own study demonstrated that adolescent males with opposite-sex twins were more than twice as likely to report same-sex attraction than were adoles­cent males with older brothers-a finding that is inconsistent with the pu­tative birth order effect.

Another problem with the birth order effect is the absence of ances­tral conditions that were favorable for its natural selection. Ancestral populations of hominins (human ancestors) were very small (Gribbin & Cherfas, 2001), so it seems unlikely that the ancestral population pres­sures would have been great enough to generate the level of fraternal competition necessary for the selection of a birth order effect. In fact, the birth order effect could be harmful to parental fitness by reducing the likelihood of reproducing adult offspring. A hominin female had at most five viable offspring in her lifetime (Symons, 1979) and, on aver­age, only two to three children would have been male. Thus, most hominin males would have had only one or two maternally related brothers (cf., Western & Strum, 1983), again, suggesting a low level of fraternal competition. Further, older brothers probably helped to care for and mentor younger brothers (Parker, 2000). Adult hominin broth­ers, like chimpanzee brothers, rather than competing, probably formed strong cooperative alliances that allowed them to maintain and increase resources, such as territory and females (van der Dennen, 1995). Fi­nally, current evolutionary theory holds that there was strong selection pressure on all males, regardless of birth order, to be sensitive and flexi­ble in relationships with other males in the service of alliance formation, cooperation, and social strategizing (Cummins, 1996; Watanabe & Smuts, 2004).

Another line of research holds that maternally inherited factors are associated with both male homosexuality and the increased fecundity of their female relatives (also, see Vasey, this volume). Thus, exclu­sive homosexuality may be a by-product of a genetically based charac­teristic that increases reproduction in females. The evidence for this maternal line effect is mixed. McKnight and Malcolm (2000) have in­vestigated the maternal line effect and found no evidence to support it. Another study found that female relatives of homosexual men had sig­nificantly greater fecundity, although 79% of the variance in male sex­ual orientation was accounted for by unidentified factors and cultural effects accounting for the relationship could not be ruled out (Camperio-Ciani, Corna, & Capiluppi, 2004).

In summary, the evolutionary theories described above employ one or both of the questionable assumptions described earlier: (1) “real” same-sex sexual behavior only refers to exclusive or predominant ho­mosexuality, and (2) male-male sexual attraction is not a masculine characteristic. Additionally, the evolutionary theories of homosexuality noted above are often inconsistent with current theory on the broader phenomenon of human evolution.

In the following section, I present an alternative theory that focuses on the evolution of male-male sexual behavior, not homosexual orienta­tion. Same-sex sexual behavior is seen as behavior that all males may exhibit under certain conditions and, inasmuch, is conceptualized as a masculine characteristic. I argue that in humans male-male sexual be­havior can be considered an exaptation that evolved from sociosexual behavior that was already well-established in primates. Same-sex sex­ual behavior serves an important role in mediating primate relationships and alliances. Male-male alliances gained a uniquely important role in human evolution. Thus, I argue that male-male sexual behavior in hu­mans presented an indirect reproductive advantage for males because of its role in alliance formation. For this reason, natural selection favored male-male sexual behavior. I then incorporate this theory of male-male sexual behavior in humans into an explanation for the development of exclusive male-male sexual attraction.

The Evolution of Male-Male Sexual Behavior

Universal Patterns of Human Male-Male Sexual Behavior

Tooby and Cosmides (1992) have argued that a behavior may be con­sidered adaptive if it can be demonstrated that during evolutionary his­tory the behavior may have solved an adaptive problem by contributing directly or indirectly to reproductive success. They also indicated that recurring patterns of behavior observed in contemporary humans can provide clues to evolutionary origins, since those behaviors are likely to have had adaptive value for human ancestors. Male-male sexual behav­ior is common in humans and does not occur in a random fashion. Thus, the expression of male-male sexual behavior is an important clue to its evolutionary origin.

World-Wide Patterns

Anthropologists have found that same-sex sexual behavior in human males tends to occur in four major patterns: age-stratified (older male/younger male), gender-stratified (male active role/female passive role), egalitarian (Greenberg, 1988; Murray, 2000; Werner, 1998), and class-stratified (social superior/social inferior) (Greenberg, 1988). The age-stratified form is the most common pattern across cultures, fol­lowed by the gender-stratified form (Greenberg, 1988). The egalitarian form is usually seen only among juveniles and adolescents (Greenberg, 1988; Murray, 2000; Werner, 1998). Although egalitarian relationships between adult males are common in the gay subculture of contemporary Western countries, this pattern is historically rare and related to the spe­cific cultural and sociological features of these societies (Greenberg, 1988; Werner, 1998).

In all but the egalitarian form, there is a clear dominance-submission dynamic between the individuals. Some writers have suggested that the dominance-submission dynamic may also be present in some male-­male sexual relationships between peers, which have usually been inter­preted as consensual and equal (Hensley & Tewksbury, 2002; Niles, 1986). In their seminal study of male couples, McWhirter and Mattison (1984) found that couples with the longest relationships, those over 30 years, had the largest age differences, ranging from 5 to 16 years. The finding that age differences between partners are associated with rela­tionship longevity suggests a possible age-related dominance-submis­sion dynamic at work in long-term male-male egalitarian relationships. Studies on preferred partner characteristics have shown that self-identi­fied gay men, assumed to form egalitarian relationships, described ide­alized partners who are slightly but significantly more masculine than themselves on several characteristics, such as height, muscle mass, and sexual expression (Muscarella, 2002; Muscarella, Elias, & Szuchman, 2004).

Cross-culturally, patterns of male-male sexual attraction and behav­ior strongly suggest that the primary mechanism underlying these rela­tionships is one of dominance-submission. A dominance-submission mechanism is also thought to be fundamental to male-female sexual re­lationships (Eibl-Eibesfeldt, 1990; Feierman, 1990). The mechanism it­self and the fact that it appears to underlie both male-female and male-male sexual behavior will be elaborated upon later.

Prison Settings

Outside of the gay subculture, the two places in which male-male sexual behavior is common are prisons and segregated juvenile facili­ties, such as boys’ schools. The male-male sexual behavior that occurs within these settings is attributed largely to a lack of female sexual part­ners (Hensley & Tewksbury, 2002; Niles, 1986). Yet, within these set­tings, the expression of same-sex sexual behavior takes a particular form. In a review of empirical studies on the sexual behavior of prison inmates, Hensley and Tewksbury (2002) identified an overt domi­nance-submission dynamic, since much of the sexual behavior is co­erced. Studies have consistently shown that the inmates who are targeted and coerced into sex are perceived as weak and vulnerable, and the same individuals are targeted repeatedly. Factors contributing to risk of sexual coercion are small size, attractiveness, youth, feminine characteristics, first imprisonment, and display of fear. Older inmates often offer “protection” to younger inmates in exchange for sex. In a study of Brazilian prisoners, Silva (1998, cited in Cardoso & Werner, 2003) reported that men most concerned about their positions in the so­cial hierarchy spoke most favorably about raping other prisoners. Price (1984) concluded that coercive male-male sexual relationships in prison validate the masculinity of the dominant male.

Hensley and Tewksbury (2002) also found that consensual sex be­tween men in prison is common, with 12% to 65% of men surveyed re­porting consensual sex with men. Sixty-nine to 80% of these men identified themselves as heterosexual. Price (1984) suggested that sexual relationships in prison provide an avenue for release of sexual frustration. However, Hensley and Tewksbury noted that closer exami­nation of consensual sexual behavior among male inmates is likely to reveal elements of coercion and manipulation. Fear of sexual assault is pervasive among male inmates, and many men may engage in same-sex sexual behavior for protection, security, and financial support. Thus, the dominance-submission dynamic appears to predominate in prison set­tings. Further, some of the sexual behavior between the men can also be seen as reinforcing alliances that contribute to the safety and perhaps even the survival of the subordinate partner, who may ultimately repro­duce. Surveys suggest that 90% of inmates identify as heterosexual or bisexual (Hensley & Tewksbury, 2002).

Adolescent Settings

Developmental psychologists have long noted that same-sex sexual attraction and behavior is common among adolescents and more likely to occur in sex-segregated settings (Hurlock & Klein, 1934; Spurlock, 2002). This phenomenon exists across cultures and appears to be a pre­dictable part of human adolescence (Savin-Williams, 1988). Niles (1986) reviewed the limited literature on same-sex sexual behavior in adolescent residential settings and concluded that male-male sexual be­havior is common in these settings, attributed largely to a lack of female sexual partners, and rarely associated with a homosexual orientation in adulthood. Bartollas, Miller, and Dinitz (1976, cited in Niles, 1986) identified five major types of sexual interactions in male residential set­tings: (1) boys who willingly engaged in sexual behavior and who, un­der pressure, continued to do so in subsequent placements; (2) boys who offered sex for protection or goods; (3) boys who were passive and de­pendent and had sex out of fear; (4) boys who identified as gay who ac­tively sought receptive sexual experiences; and (5) boys who had sexual relationships that developed out of friendship and the proximity and opportunity posed by institutional living.

Initiation Rites

A number of authors have posited that male-male sexual attraction, which may (but does not necessarily) include consummated sexual acts, is an important psychological factor in contemporary literate and essentially opposite-sex sexual societies, for bonding, alliance forma­tion, and maintenance of dominance hierarchies in sports (Guttman, 1996), fraternities (Wingate, 1994), and the military and paramilitary (Henningsen, 1961; Poundstone, 1993; Shilts, 1993). There is a clear dominance-submission dynamic in initiation rites for both juvenile and adult males. Initiation rites serve to introduce males into the male social hierarchy, as well as to bond participating males to each other and rein­force the hierarchy itself (Weisfeld, 1997). Initiation rites often include physical trials and harsh treatment (Weisfeld, 1997) and real or sym­bolic sexual behavior. Greenberg (1988) has reported that sodomy in male initiation rites is often employed to promote group cohesion and bonding. Herdt (1981) has described societies in Melanesia in which boys are sexually initiated by older boys and men. Molina (1995) anecdotally described an initiation ritual in the Spanish army that in­cluded rigorously beating new recruits and stamping them on their na­ked buttocks with an official army seal. Poundstone (1993) described a ritual in many navies of the world, including the U.S. Navy, called the “shellback ceremony,” in which sailors who cross the equator for the first time are initiated. The ritual includes cross-dressing and playing the receptive role in mock sexual acts with senior crew members. U. S. sailors have reported that actual rapes sometimes have occurred in the ceremony.

Ecological Triggers of Male-Male Sexual Behavior

Based on the data presented above, there appear to be two conditions whose co-occurrence tends to trigger same-sex sexual behavior in hu­man males: the absence of females and a rigidly enforced male domi­nance hierarchy. Further, subordinate and dominant status in these hierarchical systems appears to elicit submission-related and domi­nance-related sexual behavior in men. Low status in the presence of sexually interested dominant men elicits submissive sexual behavior, although coercion also plays an important role. This pattern of sexual submissive behavior is evident among juveniles, adult males who adopt a stereotypical female subordinate role, and low status adult males such as initiates in all male groups and physically vulnerable men in prison.

Thus, dominant status among males appears to elicit a tendency to sexu­ally dominate subordinate males, as seen in prison, initiation rites and in societies that condone or encourage sexual interactions between adult and juvenile males.

Although the absence of female sexual partners may be a sufficient condition for same-sex sexual behavior in human males, it is not a nec­essary condition. Anthropological evidence indicates that men with high social dominance status and, thus, with sexual access to women will sometimes seek male partners who are submissive to them (e.g., Dover, 1978; Greenberg, 1988; Murray, 2000). This sexual behavioral pattern is consistent with Brooks’ (2004) speculation that highly domi­nant men can be sexually attracted to any target-female or male-who is deemed to be more passive and subordinate than them.

Male-Male Sexual Behavior and Dominance-Submission Relations

Several theorists have argued that since human male-male sexual be­havior has many manifestations, it must also have many causes (Bell & Weinberg, 1978; Cardoso & Werner, 2003; Eibl-Eibesfeldt, 1990; Shibley Hyde & DeLamater, 2006). Evidence reviewed here suggests that there are various proximate causes for male-male sexual behavior, such as profit, sexual frustration, anxiety, fear, aggression, coercion, af­fection, and love. However, as presented above, the recurring pattern of male-male sexual behavior across a variety of circumstances, cultures, and historical periods suggests that a dominance-submission dynamic underlies its expression. Thus, the dominance-submission dynamic is relevant to the evolutionary history of male-male sexual behavior in hu­mans and, in fact, a number of theorists have suggested it (Brooks, 2004; Eibl-Eibesfeldt, 1990; Feierman, 1990; Rahman & Wilson, 2003; Werner, 1998).

In light of the patterns presented above, I argue that same-sex sexual behavior in human males stems from the same dominance-submission relations from which male-female sexual behavior originated. Eibl-Eibesfeldt (1990) stated that the evolutionary history of sexual relation­ships moved from those characterized primarily by dominance-submis­sion relations, such as those seen in reptiles, to those characterized by affiliative relations, bonding, and love, such as those seen in mammals, including humans. Affiliative relationships developed with the advent of parental care and nurturing behavior, although the reptilian domi­nance-submission relations are still present to some degree in human sexual behaviors.

The use of varied sexual behavior as a method of communication for the regulation of social relationships is well-established in primates (de Waal, 1982; Hambright, 1995; Vasey, 1995; Wallen & Parsons, 1997). Primates exhibit expressive-behavior patterns linking male sexual be­havior to dominance expressed through ritualized mounting, with sub­mission expressed through female-type presenting (Eibl-Eibesfeldt, 1990). Hominins probably exhibited similar behaviors (Watanabe & Smuts, 2004). In his earlier work, Vasey (1995) postulated that same­sex sexual behavior in humans may have evolved as an exaptation from the general sociosexual behavior exhibited by primates, although his thinking on the topic has changed (see Vasey, this volume). Human male aggression and dominance is linked to sexual behavior, and phallic displays are common whether expressed indirectly through ges­tures and words or directly through actual sexual domination (Eibl-Eibesfeldt, 1990). In a similar vein, Werner (1998) proposed that the evolution of homosexuality is linked to dominance-submission rela­tions between males in association with the marking of territory and co­operation.

Dominance-submission dynamics are a fundamental element in male-female sexual relationships in which great affection and love are exchanged (Eibl-Eibesfeldt, 1990; Feierman, 1990), in spite of the cur­rent cultural value placed on egalitarianism and equality. The domi­nance-submission context of male-male sexual behavior, like that of male-female sexual behavior, also allows for the development of genu­ine affection and love. Further, the status differential does not exclude the experience of genuine sexual pleasure by participants. Feierman (1990) has argued that the sexual pleasure generated by difference in dominant-subordinate status between sexual partners has been under­estimated and understudied (also see Mercer, this volume). Eibl-Eibesfeldt (1990) has proposed that male sexual-dominance desire may be directed toward any subordinate partner. Scholars have noted that this was the context of much cited adult male sexual behavior in pre­Christian Greece and Rome (Cantarella, 1992; Verstraete & Provencal, 2005) and in most of the world throughout history (Greenberg, 1988). Sexual dominance complements sexual submission desire, which is usually expressed by females but also can be expressed by subordinate males (Eibl-Eibesfeldt, 1990; Feierman, 1990). Complementary domi­nant-submissive sexual desires, infused with affiliative sexuality and affection, play a significant role in most adult human sexual relation­ships regardless of the sex of the actors.

The Alliance Theory of Male-Male Sexual Behavior in Humans

Theorists from a variety of disciplines have proposed that male-male sexual behavior among early humans may have served as a mecha­nism to reinforce alliances and reduce aggression between males (Brooks, 2004; Dizinno, 1984; De Block & Adriaens, 2004; Kauth, 2000; Kirkpatrick, 2000; Mackey & Immerman, 2000; Muscarella, 2000; Ross & Wells, 2000; Rahman & Wilson, 2003; Schuiling, 2004; Van Winkle, 1984; Werner, 1998; Wilson, 1978). These theorists differ in the depth of their explanations and in terms of the processes involved. However, they all suggest that male-male sexual behavior may have functioned to reinforce alliances and decrease aggressive behavior be­tween males, contributing directly to the male participants’ survival and indirectly to their reproduction. Thus, male-male sexual behavior was adaptive.

Ross and Wells (2000) have proposed that homosexual behavior evolved from homosocial behavior. Male homosocial behavior may have contributed to male survival through increased social support and greater access to resources. Sexual behavior between males would have reinforced homosocial bonds and, thus, would have been acted upon by natural selection. Similarly, Kirkpatrick (2000) theorized that male-­male sexual behavior results from individual selection for reciprocal al­truism, which would have contributed to resource exchange and a reduction of inter-male aggression.

In an earlier work, I described an evolutionary scenario, consistent with the theories above, in which selection for male-male sexual behav­ior might have developed (Muscarella, 2000). Briefly, I argued that ad­olescent and young adult hominins were probably sexually segregated and socially peripheralized. Young males may have lived in all-male groups with their own dominance hierarchies that undoubtedly over­lapped with the central male hierarchy. Same-sex sexual behavior would have reinforced alliances with peers and with older males, con­tributing directly to survival. Males of all ages would have benefited from alliances. Older and higher status same-sex allies could have also helped younger males climb the male social hierarchy, ultimately giv­ing them access to females and reproductive opportunities. Sexual be­havior between adolescent male peers may have included playful experimentation with dominance-submission behavior and relations (Werner, 1998).

Theory holds that hominin life and social organization were very tur­bulent and strongly suggests that as a result adolescent and young adult hominin males lived in some degree of social segregation and vulnera­bility. There is general agreement that Homo erectus, the ancestor of modern humans, lived in loose family groups (Parker, 2000). There is no evidence to support the idea that hominins lived in secure and so­cially stable nuclear family groups of long-term mated parents and their offspring (Ryan, 2004). Nuclear families, if they existed at all, were probably short-lived. It is speculated that pair-bonds between mated male and female hominins lasted only about 4 years, long enough for a child to reach a point of maturity where the mother alone could have cared for it (Fisher, 1989). Most hominins probably died before reach­ing the age of 40 (Diamond, 1992). Predators and intra-species aggression contributed to the mortality of both adult males and females (Boaz & Ciochon, 2004). In addition, female mortality would have been increased by complications with childbirth, and male mortality would have been increased by the dangers associated with hunting (Megarry, 2000). Thus, on some regular basis adults were probably looking for new mates and children were left without one or both parents. Further­more, adult males engaged in polygyny (Megarry, 2000; Symons, 1979), and were a physical threat to the offspring of females, if the off­spring were not fathered by them (de Waal, 2001). Adolescent males were probably perceived by adult males, including biological fathers and stepfathers, as a sexual threat to the females in their groups. Conse­quently, adolescent males were probably separated from the females or even driven out of the community.

A contemporary parallel to this evolutionary scenario has recently come to light in several polygamous communities in Arizona and Utah. Hundreds of adolescent males, some as young as 13, were expelled from their homes and communities so that multiple wives would be available to adult males (Kelly, 2005). The boys became homeless, living on the streets of neighboring communities. The expulsions were initiated by both fathers and stepfathers. In view of the speculated evo­lutionary environment and evidence from contemporary societies, it seems likely that at least some adolescent hominin males drifted or were driven to the periphery of their communities, where they stayed until they worked their way back into the central male hierarchy.

The alliance theory meshes plausibly with general theory on the evo­lution of human behavior. Theorists have speculated that the determin­ing factor in the evolution of human intelligence, speech, and social behavior was the development of complex social organization based on affiliation, cooperation, and multiple alliances between related and non-related individuals (Corballis, 1999; Cummins, 1996; Parker, 2000; Watanabe & Smuts, 2004). Darwin also considered the possibil­ity that during evolution within-group alliances were even more important than the mother-child bond (Ryan, 2004).

Parker (2000) has hypothesized that the evolution of intelligence in Homo erectus required an extended juvenile period for learning adult skills and that male juveniles probably relied on prolonged associations with male mentors, most likely kin. However, Western and Strum (1983) have argued that during human evolution males had less cer­tainty of kin and fewer available kin and, thus, would not have benefited from kin as reliably as females. Western and Strum concluded that there was great selection pressure on males to develop social manipulation strategies in the creation of alliances. Mackey (1990) has reported evi­dence to support the hypothesis that humans evolved a mechanism of at­traction between adolescent and adult males. Among male primates reproductive success is robustly linked to dominance status (Cummins, 1996; Pusey, 2001), which is closely linked to strong male alliances (Cummins, 1996). Male alliances with non-kin are particularly impor­tant for successful within-group interactions (Pusey, 2001). Alliances between non-kin are dependent upon the formation of reciprocal obligations, such as social support and grooming (Cummins, 1996).

In summary, the alliance theory holds that male-male sexual behav­ior adaptively functioned to reinforce alliances that contributed directly to survival and indirectly to reproduction. Ecological conditions con­sisting of rigidly hierarchical male groups and an absence of females may predictably trigger same-sex sexual behavior, because such condi­tions are similar to the evolutionary environment in which this sexual behavior was adaptive. The alliance theory parsimoniously and plausi­bly explains why the most common form of male-male sexual behavior in humans is the cross-generational form and why the egalitarian form is developmentally linked to adolescence. Evolutionarily, male-male sex­ual behavior in humans appears to be related to dominance-submission relations associated with age-related differences in dominance status between males and to the playful experimentation with dominance-sub­mission behavior that develops between juveniles and adolescents. The gender-stratified and the class-stratified forms of male-male sexual be­havior can be seen as arising from cultures with rigid gender and class constraints on the expression of sexual behavior (Greenberg, 1988). The alliance theory, however, cannot predict individual cases of same-sex sexual behavior. Evolutionary theories can only suggest general tendencies within populations (Eibl-Eibesfeldt, 1990). Numerous factors associated with individual development interact with evolution­ary predispositions to affect the expression of a trait (Lickliter & Honeycutt, 2003).

The Evolutionary Development of Male-Male Sexual Behavior

Van der Dennen (1995) has suggested that human males, like chim­panzee males, evolved a coalitional reproductive strategy. That is, coali­tions of male allies take and defend territory and its resources, such as food, shelter, and females. Male chimpanzees employ grooming, which is often sexually arousing, to reinforce coalitions (de Waal, 1982; Taub, 1990). Dominant male chimpanzees sometimes allow their closest allies (those who groom them the most) to copulate with estrus females but deny other males sexual access; this phenomenon has been interpreted as “sexual bargaining” (de Waal, 1982). The average female chimpanzee produces only five viable offspring in her lifetime (Tutin, 1979; Symons, 1979). Thus, behavior that increases a male chimpanzee’s chance of fa­thering an infant contributes to his reproductive success.

Male-male sexual behavior in the service of alliance formation and sex­ual bargaining may have had a similar adaptive value for male hominins. Before the evolution of language, dominance-submission relations and rit­uals between hominins were probably physically communicated, as they are between nonhuman primates (Watanabe & Smuts, 2004). Churchill (1967) has argued that patterns of sexual behavior in nonhuman primates that lead to arousal could be expected to lead to consummated acts in early humans. Rawson (1973) has commented that the direct expression of sex­ual behavior as a means of communication in primitive groups of humans has been underestimated (also see Taylor in this volume).

Some theorists have argued that consummated sexual acts between males are maladaptive, because they do not lead directly to reproduction (Gallup & Suarez, 1983). However, this would only be true if every ejaculation had the same probability of impregnating a female, which is not the case. Evolutionary forces selected human males to have high and constant sex drives (Symons, 1979), but during their lifespan hominin males would have experienced many periods of time that of­fered little or no opportunity to impregnate females. Younger males were probably socially peripheralized (Muscarella, 2000) and, because some adult males were polygynous, other adult males would have had no female partners for part or all of their lives (Mealey, 1985; Symons, 1979). In fact, the contemporary study of sex differences in evolution­ary psychology is based upon the fundamental assumption that ances­tral human males had differential reproductive success due to the lack of mating opportunities for some (Buss, 2004). Even mated males would sometimes have had no sexual opportunities because of their mates’ lack of interest in sex during menstruation and pregnancy or following childbirth and menopause, as evident among contemporary human fe­males (Ellison, 2001; Shibley Hyde & DeLamater, 2006). In the evolu­tionary past, male-male sexual behavior that reinforced alliances and indirectly contributed to higher social status and future reproductive op­portunities would have been more adaptive than no male-male sexual behavior at all, particularly in the absence of heterosexual alternatives. Sexual activity between males may have stimulated the production of testosterone which contributed to dominance behavior, cooperative male-male alliances, and increased access to females (cf., Wrangham, 1976 cited in Crook, 1980). Kirkpatrick (2000) reported on studies that showed that sexual behavior with men did not lower a man’s reproduc­tive success, as long as he also engaged in sexual intercourse with women. Additionally, as suggested by contemporary studies of men in prison, it is possible that in rigidly hierarchical all-male environments an inability or resolute unwillingness to engage in sexual behavior with other males would have been maladaptive and selected against.

Proximate Causal Factors: A Neurological Mechanism

Many theorists have suggested that sexual orientation development in humans is partly determined by patterns of structural development and organization in the brain associated with the hormonally-medi­ated processes of feminization, defeminization, masculinization, and unmasculinization (Brooks, 2004; Ellis & Ames, 1987; Feierman, 1990; Kauth, 2000; Pillard & Weinrich, 1987; Rahman & Wilson, 2003). This is a complex issue and much research remains to be done. The simplistic hypothesis that the brains of homosexual males are “feminized” like those of heterosexual women is not supported by the evidence (Kauth, 2000). Woodson and Gorski (2000) have also argued that the human brain is not globally feminized or masculinized but that feminization and masculinization vary by structure. This pattern of masculinization or feminization by structure creates a continuum of sexual differentation in the brain, rather than a dichotomy (Kauth, 2000; Woodson & Gorski, 2000). This pattern may account for the complexity of expression of sexual orientations in humans. Consistent with this idea, Rahman and Wilson (2003) have noted that homosexuality is as­sociated with a constellation of predominantly sex-typical traits with some notable sex-atypical traits that vary unpredictably. For example, preferred partner characteristics of homosexual men differ in signifi­cant ways from characteristics preferred by heterosexual women, and this has been linked to speculated differences in brain organization (Muscarella et al., 2004). In this study, we found that that while both heterosexual women and homosexual men preferred partners who were heavier than themselves, heterosexual women preferred a weight differ­ence of 42.08 pounds, 7.5 times greater than the 5.58 pound difference preferred by homosexual men. Heterosexual women preferred partners 7.52 inches taller than themselves, while homosexual men preferred partners of the same height. We also found that while homosexual men perceived their own sexuality to be less stereotypically masculine than did heterosexual men, they also perceived it to be less stereotypically feminine than did heterosexual women.

There may be a link in human evolution between greater intelligence, complex social behavior, and same-sex sexual behavior arising from domi­nance-submission relations. Dominance hierarchies among primates are fluid and not static (Cummins, 1996). The need for greater behavioral and social flexibility, especially with regard to dominance- submission rela­tions, may have increased during human evolution given increased human dependence on social living and the complexity of human social organiza­tions, requiring social support from both males and females within the group (Bagemihl, 1999; Cummins, 1996; Hooks & Green, 1993).

Greater variation in the pattern of masculinization and feminization of the male brain may have allowed more flexibility in social behavior. Although Miller (2000) also made this argument, the theory presented here differs from Miller’s in two important ways. Miller’s theory holds that greater feminization of the male brain resulted in behaviors that ap­pealed only to females and that attraction to males is a uniquely femi­nine characteristic. I argue that greater variation in hormonally mediated organization of the male brain contributed to a flexibility of behavior that allowed males to appease or dominate both males and females, as the so­cial situation required. In addition, I argue that attraction to males is nota uniquely feminine characteristic. Brooks (2004) has also argued that at­traction to males should be conceptualized as a masculine characteristic. He noted, “Male homoeroticism is unquestionably a men’s thing, every bit as much as shaving in the morning and being a father” (p. 26).

Feierman (1990) has presented a model to explain the relationship between patterns of brain organization and patterns of sexual attraction based on physical signs of dominant and subordinate status such as size and age, and Brooks (2004) has proposed a similar model. Feierman stated that each individual evaluates others in terms of how much more dominant or subordinate they are to self. For example, a male whose brain is masculinized and defeminized should be attracted to females because females are less masculine and more feminine than self.

Feierman’s (1990) theory could be extended to conceptualize other social interactions. One may evaluate another individual relative to self on characteristics of masculinity and femininity, essentially domi­nant-subordinate characteristics. The other individual may be perceived as non-aggressive and weak, which may then elicit social and sexual dominance behaviors. Under the evolutionary conditions described ear­lier, an adult hominin male might find both females and younger males sexually attractive based on their relative subordinate status and sig­naled by physical and behavioral characteristics that are less masculine and more feminine than those of the adult male. However, the same young adult male when confronted with an older adult male who is more masculine and less feminine than himself may respond by being so­cially and sexually submissive or appeasing. This evaluative mecha­nism would result in a variety of dominance-submission behaviors and would allow for very flexible social interactions with various dominant and subordinate individuals in the social group.

In summary, variation in the patterns of masculinization and femini­zation of the brain may have evolved to allow human males to behave in a more complex and flexible way in groups of individuals with varying dominance status. This behavioral flexibility is evident among self­identified heterosexual men who engage in same-sex sexual behavior in a variety of sexually segregated settings and then resume opposite-sex sexual behavior when female partners are available. Increased femini­zation of the male brain may have increased flexibility of behavior al­lowing adult males to deal effectively with both males and females in complex and fluid relationships. Thus, male-male sexual attraction may be conceptualized as a masculine characteristic. Varied patterns of brain masculinization and feminization may account for the diverse sexual feelings that are evoked by varied dominance-submission interactions; the pattern of brain organization due to hormonal influence may contribute to development of varied sexual attractions.

Support for Male-Male Erotic Alliances in Ancient Literature

The flexibility of dominance-submission relations, the sexuality these relations can generate, and sexuality’s role in the maintenance of a male-male alliances are illustrated in the relationship between two men, Enkidu and Gilgamesh, in the world’s earliest known piece of literature, the epic Gilgamesh (Mitchell, 2004). The fact that the story’s central theme depicts these related psychological phenomena suggests that they are indeed fundamental to the nature of human males. The story of Gilgamesh was committed to writing almost 4,000 years ago but hails from an even older oral tradition.

Gilgamesh, the king of Uruk, is arguably a despotic (i.e., highly masculinized) dominant male. He fights with all the young men and “crushes” them and forces himself sexually on every bride in the city, disrespecting them and their husbands of all social ranks. The people cry out to the gods for help. In keeping with the concept of comple­mentarity, the gods send the wild man, Enkidu, equal in strength and courage, but just and kind (i.e., highly masculinized and somewhat feminized), to balance Gilgamesh’s aggressive nature.

A priestess is sent to civilize the wild Enkidu before presenting him to Gilgamesh. She transforms Enkidu through heterosexual intercourse into a rationale human being. When Enkidu learns about Gilgamesh’s aggressive, disrespectful behavior from the priestess, he responds with a stereotypic dominant male aggressive swagger:

I will challenge him. I will shout to his face:
I am the mightiest! I am the man
who can make the world tremble! I am supreme!
(pp. 80-81)

The priestess advises Enkidu to subordinate himself to Gilgamesh, the more dominant male, and the advice intimates the relationship between dominance-submission and sexual attraction:

You will stand before him and gaze with wonder, you will see how handsome, how virile he is, how his body pulses with erotic power.
He is even taller and stronger than you-
so full of life-force that he needs no sleep.
Enkidu, put aside your aggression.
(p. 82)

Eventually, Enkidu and Gilgamesh meet and fight. Gilgamesh wins, and Enkidu subordinates himself to Gilgamesh, acknowledging his dominance. From this dominance-submission relationship, love be­tween the two men emerges:

Finally, Gilgamesh threw the wild man
and with his right knee pinned him to the ground.
His anger left him. He turned away.
The contest was over. Enkidu said,
‘Gilgamesh, you are unique among humans.
Your mother, the goddess Ninsun, made you stronger and braver than any mortal, and rightly has Enlil granted you the kingship, since you are destined to rule over men.’
They embraced and kissed. They held hands like brothers.
They walked side by side. They became true friends.
(pp. 89-90)

The story of Gilgamesh and Enkidu is a parable for the alliance the­ory. It illustrates how even very dominant males like Enkidu may some­times need to subordinate themselves to even more dominant males like Gilgamesh. The dominance-submission mechanism generates sexual attraction and affection contributing to an alliance which benefits both partners. Gilgamesh gains a reliable ally in his future struggles for dom­inance with other individuals, and Enkidu gains the favor of the domi­nant male, giving him access to resources and females.

Exclusive Male-Male Sexual Attraction

In exclusive homosexuality, attraction to other males is constant, does not depend upon situational contexts or triggers, and excludes most heterosexual interest. However, exclusive male-male sexual be­havior, nonetheless, falls into one or more of the four major patterns of relationships identified earlier: age-stratified, gender-stratified, egali­tarian, and class-stratified. Most evolutionary theories have assumed that exclusive homosexual orientation is a by-product of selection for related characteristics, for example, altruism (Wilson, 1978), femi­nine traits (Miller, 2000), or female fecundity (Camperio-Ciani et al., 2004). Money (2004) has hypothesized that homosexuality is a by-product of the evolutionary loss of rigid biological programming of the brain’s map for courtship and mating (the lovemap) in humans. He speculated that this more flexible programming was a part of a gen­eral “derobotization” of brain maps, especially those related to the de­velopment of language and conceptual thinking.

The alliance theory holds that male-male sexual behavior itself was adaptive and that all males may exhibit same-sex sexual behavior under certain conditions. I have also speculated here that during hu­man evolution increased variation in the organization of the brain may have contributed to behavioral flexibility for males. Thus, genes for male-male sexual behavior would be present in the entire popula­tion. Through normal variation of genetic traits, some males may have a greater predisposition for same-sex sexual behavior, perhaps medi­ated by a specific brain organization, contributing to an exclusive male-male sexual attraction. In addition, because of cultural pres­sures to reproduce, most men with exclusively same-sex sexual inter­ests married and had children, which also would have contributed to the maintenance of related genes in the population (Money, 2004; Weinrich, 1987).

The consensus among theorists is that any genetic contribution to an exclusive male-male sexual attraction would interact with environmen­tal factors. Laumann et al. (1994) found that 2% of American men have exclusively same-sex sexual interests. However, this statistic cannot serve as the base rate for exclusive homosexuality in the human species. First, this statistic comes from one sample of one group of humans, in one country, at one point in history. Arguably, it is not representative of the species. Second, although traditional sex surveys in other countries such as Britain and France have reported similar percentages (Shibley Hyde & DeLamater, 2006), more recent economic actuarial studies in Britain have yielded a percentage of 6% (Campbell, 2005), which is a remarkably discrepant result.

The incidence of exclusive homosexuality in hominins cannot be known. A number of factors linked to culture and to high population densities associated with more developed societies may allow or even encourage a pattern of exclusive same-sex sexual attraction that would not have been possible in the evolutionary environment. Thus, at this time, in light of the current body of research, it seems that the best ex­planation for the development of an exclusively same-sex sexual attrac­tion is that held by most theorists: it results from an unpredictable interaction between genetic, cultural, developmental, and psychologi­cal factors.

Implications for Future Research

I have argued here that there is a link between male-male sexual be­havior, alliance formation, and complex social organization. The devel­opment of complex social organizations has been associated with the evolution of cognition and language in humans (Corballis, 1999; Cummins, 1996; Money, 2004; Watanabe & Smuts, 2004). Strong so­cial skills, high intelligence, and highly developed verbal skills are ste­reotypical elements of exclusively homosexual men and may contribute to profession preferences (Bailey, 2003; Miller, 2000). Thus, the bio­logical contribution to an exclusive male-male sexual attraction may be a by-product of evolution for the underlying social, cognitive, and lan­guage skills characteristic of the species. The characteristics of altruism (Wilson, 1978), empathy, and sensitivity (Miller, 2000), which have been attributed to exclusively same-sex attracted men, may be linked to their putative social skills. Research into the possible evolutionary link between male-male sexual behavior and social skills, intelligence, and language seems warranted.

The alliance theory has implications for the conceptualization of dominance-related behavior in human males. Miller (2000) has stated that there is ample evidence that homosexual men are more feminized and not as aggressive as heterosexual men. Although homosexual men may not be as aggressive as heterosexual men, they are just as assertive (LeVay, 1996). If homosexual men actually do have a lower level of ag­gression, this lower level may be linked to the evolution of male-male sexual behavior and may contribute to the ability to assume dominance and leadership roles. In primates, the individual’s physical size and ag­gressiveness are not highly correlated with dominance. Rather, social dominance is correlated with social skills, political alliances, and dyadic or triadic aggression resulting from these alliances (Bagemihl, 1999; Cummins, 1996; de Waal, 1982). Consistent with this notion, ho­mosexual men in the predominantly male hierarchy of the military reportedly function superlatively and have been viewed as excellent leaders (Shilts, 1993).

Finally, the alliance theory may prove to have good heuristic value. For example, colleagues and I have shown that same-sex sexual behav­ior that is not presented as “homosexual” can be seen as contributing to social status and reproductive success when it occurs under conditions posited by the alliance theory (Muscarella, Cevallos, Siler-Knogl, &Peterson, 2005). Male and female targets who engaged in same-sex sexual behavior were described in a scenario that had three outcomes. As a result of the same-sex sexual relationship, the outcome was socially nega­tive, socially neutral, or socially positive. In the scenario with the socially positive outcome, the same-sex sexual behavior reinforced an alliance that contributed to social opportunities. The targets in this con­dition received the highest scores related to future social status and re­productive opportunities. The compelling aspect of this study is that we were able to experimentally manipulate the powerfully negative per­ception of same-sex sexual behavior under the conditions indicated by the alliance theory. The study also sheds new light on the phenomenon of homonegativism. We argued that the term “homosexual” connotes low social status and, because males have been shaped by evolution and reinforced by culture to be exceptionally sensitive to social status, they are threatened by the word “homosexual.” However, when same-sex sexual behavior is presented without the “homosexual” label and asso­ciated with status-raising opportunities, men see it as less damaging, and this is predicted by the alliance theory.

Conclusion

Male-male sexual behavior is characteristic of the human species. Thus, the use of evolutionary theories is an appropriate method to study and understand it. I argue that the alliance theory of the evolution of male-male sexual behavior meets the first four criteria for a good evolu­tionary explanation set forth by Buss et al. (1998): (1) precision and internal consistency; (2) consistency with knowledge in evolution­ary biology; (3) ability to generate new and specific predictions; and (4) greater parsimony than competing explanations. It remains to be seen if the criterion regarding the computational capacity of the psycho­logical mechanism can be met. The theory holds that male-male sexual behavior in humans has the same dominance-submission basis as male-female sexual behavior and is more developed in humans relative to other primates because it had adaptive value by reinforcing male alli­ances that played a unique role in human evolution. I have speculated that male-male sexual behavior is an exaptation of the sociosexual be­havior that was well-established in primates and used to communicate and to regulate social interactions. The proximate mechanism for male-male sexual behavior may be variable and developmentally- linked to organization of the brain to allow greater flexibility in the dominance-submission social interactions found in human groups. The alliance theory provides a new framework for research in the area and may have significant heuristic value.

Issues for Further Consideration

A point for consideration regards the best evolutionary explanation for same-sex sexual behavior in humans. I have argued that there is enough evidence from the traditional sources to indicate that male-male sexual behavior is an exaptation and that this explanation meets most of the criteria set forth by Buss et al. (1998) for a good evolutionary expla­nation. However, Vasey (this volume) has argued just as persuasively that the by-product explanation is better, and Diamond (this volume) made a similar argument for female-female sexual behavior. Obvi­ously, this matter remains unsettled, but the challenging and exciting is­sue is that all authors have provided a basis for further theorizing and for the generation of testable hypotheses.