Behavioral Ecology Vol. 14 No. 5: 668-678
© 2003 International Society for Behavioral Ecology
Major histocompatibility complex genes, symmetry, and body scent attractiveness in men and women
Departments of a Biology b Psychology, The University of New Mexico, Albuquerque, NM 87131-1091, USA
Address correspondence to R. Thornhill. E-mail: rthorn{at}unm.edu.
Received 16 January 2002; revised 25 September 2002; accepted 4 December 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Previous research indicates that the scent of developmental stability (low fluctuating asymmetry, FA) is attractive to women who are fertile (at high-conception risk points in their menstrual cycles), but not to other women or men. Prior research also indicates that the scent of dissimilarity in major histocompatibility complex (MHC) genes may play a role in human mate choice. We studied the scent attractiveness to the opposite sex of t-shirts worn for 2 nights' sleep. Our results indicate that the two olfactory systems are independent. We repeated previous results from studies of the scent of symmetry. We repeated previous results from MHC research in part; men, but not women, showed a preference for t-shirts with the scent of MHC dissimilarity. Women's scent ratings of t-shirts were uncorrelated with the wearer's MHC dissimilarity and allele frequency, but positively correlated with the wearer's MHC heterozygosity. Fertile women did not exhibit any MHC trait preferences. Women's preference for the scent of men who were heterozygous for MHC alleles may be stronger in women who are at infertile cycle points. Men preferred the scent of common MHC alleles, which may function to avoid mates with rare alleles that exhibit gestational drive. Men also preferred the scent of women at fertile cycle points. The scent of facially attractive women, but not men, was preferred. Neither FA nor facial attractiveness in either sex correlated with MHC dissimilarity to others, MHC heterozygosity, or MHC allelic rarity.
Key words: Homo sapiens, inbreeding, major histocompatibility complex, mate choice, parasites, pheromones, sexual selection.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The major histocompatibility complex (MHC) is a highly polymorphic gene complex (in humans often referred to as human leukocyte antigens, HLA genes), that encodes cell-surface receptors that play a critical role in the initiation of most immune responses. By binding and presenting peptides derived from either endogenous or foreign proteins, the MHC plays a central role in the discrimination of self from non-self, thereby regulating recognition of infectious diseases. Host–pathogen coevolution is now generally thought to explain much of MHC genetic polymorphism because of MHC's function in pathogen detection (Apanius et al., 1997
; Hughes and Hughes, 1995; Penn and Potts, 1999).
Increasing evidence indicates that MHC genes influence body odor and mate choice based on body odor attractiveness. Most investigations have involved house mice (Mus musculus) and humans. The mouse studies suggest a preference for the scent of MHC dissimilarity (Penn, 2002; Penn and Potts, 1999). In humans, Wedekind et al. (1995) found that normally ovulating women (e.g., not taking a contraceptive pill) preferred the scent of men (on t-shirts worn for 2 nights' sleep) who had MHC genotypes dissimilar to their own. A second study replicated the MHC-dissimilarity scent preference in women not on the pill and, furthermore, found the same scent preferences in men (Wedekind and Füri, 1997). Interestingly, participants in these two studies also claimed that the scents of MHC-dissimilar individuals were reminiscent of those of former or current mates. Indeed, Ober et al. (1997) found that Hutterite married couples tend to be more MHC dissimilar than expected by chance, an effect that may be at least partly mediated by body scent. In contrast, a study of South American Indians (Hedrick and Black, 1997) and a study of Japanese couples (Ihara et al., 2000) reported no evidence for MHC-dissimilar assortative pairing among married couples (for review, see Penn, 2002).
There are at least four adaptive explanations for MHC-dissimilarity scent preferences. First, these preferences may have the evolved function of inbreeding avoidance (see review of literature on this hypothesis in Penn and Potts, 1999). The three remaining potential explanations of scent preference for MHC-dissimilarity are founded upon parasite-mediated sexual-selection theory. The heterozygosity hypothesis posits that the preference has evolved because it produces MHC-heterozygous offspring that have sound immunocompetence against more parasite types (owing to a wider array of pathogenic antigens recognized by the immune system, given that MHC alleles show codominance; Brown, 1997; Wedekind et al., 1995). Penn et al. (2002) found that heterozygous house mice are more fit than homozygotes upon encountering multiple strains of pathogens. The rare-allele hypothesis claims that the preference functions to obtain rare MHC alleles for offspring, which give defense against rapidly evolving parasites that may escape recognition by immune systems containing common alleles or that may reduce the risk of autoimmunity possibly associated with common alleles (Penn and Potts, 1999). The rare-allele hypothesis implies preference for MHC dissimilarity because, to place the rare MHC alleles in offspring by mate choice, mate choosers (most of whom carry relatively common MHC alleles) must choose the infrequent individuals who are dissimilar because of their rare MHC genotype. Wedekind and Füri (1997) found no evidence that particular MHC genotypes were preferred, as the rare-allele hypothesis implies, but their study examined responses to only six people's scents. The diverse-genes hypothesis views the function of the preference as obtaining MHC alleles for offspring that are different from the parents' own (independent of both production of heterozygous offspring and offspring with rare alleles) as well as from one another because pathogens adapted to parents and other close kin will then be less adapted to offspring (Penn and Potts, 1999).
A prediction related to the the diverse-genes hypothesis is that individuals should avoid mating with MHC homozygotes. Homozygotes have offspring who vary less from each other and the mate, which increases the risk of a within-family disease epidemic. The diverse genes hypothesis is supported by a study of sticklebacks. Females preferred males with many MHC alleles over males with fewer (Reusch et al., 2001).
According to the first two parasite-mediated sexual selection hypotheses, the heterozygosity and rare-allele hypotheses, genotypic variation in MHC should be associated with interindividual phenotypic condition. Specifically, individuals with MHC heterozygosity or rare MHC alleles are predicted to be of relatively high phenotypic quality as a result of their greater resistance to parasites or, in the latter case, reduced autoimmunity. The outbreeding hypothesis gives no reason to expect such associations in the case of humans. In our largely outbred human sample (see below), MHC homozygosity should not be a reliable marker of being inbred and thus of low phenotypic quality.
A potentially important phenotypic marker of sound phenotypic condition and underlying high genetic quality is low fluctuating asymmetry (FA). FA is nondirectional deviation from perfect bilateral symmetry in traits that are, on average, bilaterally symmetrical. FA reflects ability to deal with stresses, both genetic and environmental, during ontogeny. Individual FA, then, is developmental maladaptation or instability owing to the individual's inability to achieve perfect bilateral symmetry given perturbations during traits' ontogeny (review in Møller and Swaddle, 1997; Polak, 2003).
Genetic perturbations known to generate FA in traits include mutations, homozygosity, and genetic disturbance due to incomplete coadaptation among genes (Møller and Swaddle, 1997; Polak, 2003). Variation in developmental instability should often reflect heritable differences, leading to selection for mate preferences based on traits that honestly advertise developmental instability (e.g., scents). The heritability of developmental instability is unknown, despite the many studies that have been performed, due to low power (Fuller and Houle, 2003; see also Møller and Thornhill, 1997a,b; Van Dongen, 2000). Based on these studies, Gangestad and Thornhill (2003) estimated that the heritability of developmental instability is, on average, 0.2–0.3.
In humans, FA may be involved in viability-based good-genes sexual selection. In both sexes, low FA appears to be associated with increased genetic, physical, and mental health, including cognitive skill and IQ (Furlow et al., 1997; Thornhill and Møller, 1997; Yeo et al., 2000). Compared to asymmetric men, symmetric men seem to be more muscular, vigorous, and socially dominant (Gangestad and Thornhill, 1997b), have lower basal metabolic rate (Manning et al., 1997), and may be larger in body size (Manning, 1995; Gangestad and Thornhill, 1997b) than asymmetric men. Low FA in men predicts components of mating success such as a relatively high number of sexual partners, quicker sexual access to a new romantic partner (Baker, 1997; Gangestad and Simpson, 2000; Gangestad and Thornhill, 1997b; Gangestad et al., 2001; Thornhill and Gangestad, 1994), facial attractiveness (Baker, 1997; Gangestad et al., 1994; Thornhill and Gangestad, 1994; but see also Gangestad and Thornhill, 1997a; Simpson et al., 1999; Thornhill and Gangestad, 1999b), and number of extrapair copulation (EPC) partners and number of times chosen as an EPC partner (Gangestad and Thornhill, 1997a). Moreover, men's symmetry predicts a relatively high frequency of their sexual partners' copulatory orgasms (Thornhill et al., 1995). Female copulatory orgasm may be a mechanism of cryptic female choice by selective ejaculate retention in contexts in which women mate with multiple partners (see Baker and Bellis, 1995; Thornhill et al., 1995).
The effect of symmetry on sexual attractiveness appears to be stronger in men than in women, and it is only in men that symmetry predicts the components of mating success mentioned above (Gangestad and Thornhill, 1997a). This difference is not surprising, as sexual selection has been stronger in males than females in human evolutionary history (Buss, 1994; Geary, 1998; Symons, 1979).
Several lines of evidence indicate that olfactory cues or pheromones may play a major role in the human sexual selection system. First, though adults of both sexes report that body scent of others significantly affects their sexual interest, women report a stronger effect than men (Franzoi and Herzog, 1987; Herz and Cahill, 1997; Regan and Berscheid, 1995). Second, evidence suggests that the importance of male scent in women's sexual interest and arousal varies across the menstrual cycle. Women's olfactory sensitivity to androstenol and related chemicals appears to be enhanced before ovulation, near the peak of fertility in the menstrual cycle (Doty, 1981; Vierling and Rock, 1967; but see Amoore et al., 1975; see also Pause et al., 1996). Androstenol, a chemical precursor of androstenone, is an important contributor to body odor, and its production is highly sexually dimorphic; men excrete three times more androstenol in urine than do women (Brooksbank, 1962; Brooksbank and Haslewood, 1961).
Women's sexual desire for men other than their primary partner and rate of EPCs also change across the menstrual cycle, peaking during the midfollicular to ovulatory phases (e.g., Baker and Bellis, 1995; Gangestad et al., 2002). When coupled with findings on cycle-related scent preferences, these shifts suggest a third connection between sex pheromones and human sexual selection. Specifically, the fact that EPCs and associated desire for nonpartner men and positive evaluation of sexually dimorphic substances in human sweat peak near maximal menstrual-cycle fertility suggests that selection has led to a preference in women for their offspring's sire and that the preference may include an olfactory component.
Based on this notion, Gangestad and Thornhill (1998b) proposed that olfactory stimuli pertaining to men's phenotypic and genetic quality, measured by degree of body FA, positively affects men's scent attractiveness to women, particularly during the fertile phase of the cycle when expressed preference for offspring's sire is critical. Three published studies now support this hypothesis (Gangestad and Thornhill, 1998b; Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999b). Specifically, all studies show that the scent of symmetric men is attractive to women who are cycling normally (i.e., not using hormone-based contraceptives) and are in the fertile phases of their menstrual cycles. Women using hormonal contraceptives and normally ovulating women in infertile phases of their cycles appear to show no preference for the body scent of either symmetrical or asymmetrical men. This research did not find that men rate the scent of symmetrical women more attractive than the scent of asymmetric women. If women's mid-cycle scent preference for low FA is caused by a preference for MHC dissimilarity, then men who, on average, are dissimilar to many others (for whatever reason) should have low FA.
The research on the scent of men's symmetry also gave results suggesting that facial attractiveness (as judged from photos) correlates positively with body scent attractiveness to the opposite sex in both sexes and that ovulating women's preference for the scent associated with men's facial attractiveness is maximal when their fertility is highest across the menstrual cycle (Thornhill and Gangestad, 1999b; also see Rikowski and Grammer, 1999). One theory of human physical attractiveness is that it reflects, in part, genetic differences in viability and, from the standpoint of sexual selection, is preferred partly for this reason (Gangestad, 1993; Thornhill and Gangestad, 1993). Facial attractiveness, in part, reflects judging facial hormone markers, which appear to be estrogen facilitated in women and androgen facilitated in men (Thornhill and Gangestad, 1999a). The male faces that women in the fertile phase of their cycle most prefer are more masculine than the faces they most prefer when in infertile phases ( Johnston et al., 2001; Penton-Voak and Perrett, 2001; Penton-Voak et al., 1999). These results raise the possibility that scent attractiveness may be related to the chemistry of sex hormones.
Overall, evidence indicates that pheromones are important in the human sexual selection system. Pheromones appear to be associated with MHC, men's FA, sexually differentiated odor substances, and traits of facial attractiveness. The specific chemicals that signal MHC, FA, or hormone status may or may not be shared. (For instance, the scent of low FA may or may not involve the chemistry of sex hormones; scents associated with rare or otherwise dissimilar MHC genotypes may or may not be those associated with the scent of low FA; etc.) Whether involving the same chemistry or not, the pheromone systems may or may not be interrelated in functional terms (e.g., they may or may not all be associated with parasite-mediated sexually selected signals). To further examine their roles, including the various hypotheses for MHC-based mate choice, and to explore their interrelationships, if any, we conducted a study of men's and women's attraction to the body scents of members of the opposite sex. The MHC of all participants was genotyped at the A, B, and DRß loci (as in Wedekind et al., 1995; Wedekind and Füri, 1997). In addition, we measured the body symmetry of all participants and had facial photographs of them rated for attractiveness. We recruited for the study women who did not use contraceptive hormones. The hypotheses for the MHC-dissimilarity preference all predict that the preference should be seen at fertile cycle times.
Singh and Bronstad (2001) reported that men prefer the scent of women who are in the fertile phase of their cycle, suggesting yet another pheromone-based system of preferences in humans. We examined Singh and Bronstad's finding in the current study.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
With the exception of obtaining a blood sample for MHC genetic analysis from each research participant and paying each participant $20, methods were identical to those in the study of the scent of symmetry by Thornhill and Gangestad (1999b)
. Research participants were 97 men and 100 women. Ages ranged from 18 to 54 for men (mean ± SD = 23.1 ± 6.1) and 17 to 44 for women (21.2 ± 5.5). Self-reported ethnicities of men were 62% Caucasian, 29% Hispanic, 2% African American, 5% Asian, 1% Native American, 1% other; of women, 48% Caucasian, 37% Hispanic, 2% African American, 1% Asian, 6% Native American, 5% other. Most participants received experimental course credit in an introductory psychology course in return for their participation in the research (see Gangestad and Thornhill, 1998b). Some received grade points in a biology course. Recruitment postings specifically asked for women who did not use hormone-based contraception (i.e., birth control pills, DepoProvera, or Norplant). We asked women whether they used a contraceptive of this sort and did not include the responses of those who did.
Wedekind and colleagues (Wedekind and Füri, 1997; Wedekind et al., 1995) found a different pattern of MHC scent preferences for women using contraceptive pills. These women prefer the scent of individuals with similar MHC genotypes. The pill is thought to generate a physiological state like that in pregnancy. Pill-users preference for MHC-similar scents is not a mate preference, but instead may be a preference for the smell of genetic relatives who may assist nepotistically.
The mechanisms of infertility suppression caused by the three types of hormone-based contraception used by some women in our study apparently differ (e.g., prevention of ovulation or implantation), making it difficult to predict how women using the various hormone-based contraceptives should behave in scent-based mate choice. Furthermore, sample sizes of each of the three hormone-based contraception users are too small for any meaningful estimates of effects, if any.
Measurement of FA
Participants reported in unisex groups of up to four for an initial session. After reading and signing an informed consent form, each participant was given a brief questionnaire on demographic and other information (e.g., age, height, weight, sexual orientation, socioeconomic status of family of origin, lifetime number of sexual partners). Participants were scheduled for a later date, at which time they provided a blood sample (see below). For each participant, the right and left sides of the following 10 characters were measured using a digital caliper, sensitive to 0.01 mm: ear length, ear width, elbow width, wrist width, ankle width, foot breadth, and lengths of all fingers excluding the thumb. In previous samples (n > 700), these characters have been shown to exhibit very little directional asymmetry and slight leptokurtosis, as expected of FA (Furlow et al., 1997; Gangestad and Thornhill, 1999). To assess and increase reliability, we measured each character twice. In addition, a facial photograph was taken with a 35-mm camera (50 mm lens), placed approximately 2.5 ft. from the participant's face (using 400 ASA color film).
After measurements were taken and the questionnaire was completed, each participant was given a clean, unworn, white, cotton Hanes-brand t-shirt and provided with explicit wearing instructions. Each was told that he/she should wear the t-shirt 2 particular nights (identical for each sex) while sleeping. Each was also instructed to wash his/her bed sheets with unscented laundry detergent (provided by us) before those two nights and, during the 2-day period, refrain from (1) using scented soaps, deodorant, or fragrance such as perfume, cologne, or aftershave, and instead use only unscented soap (which we provided); (2) eating garlic, onion, green chile, pepperoni, pungent spices, herbs, strong cheeses, cabbage, celery, asparagus, yogurt, and lamb; (3) drinking alcohol or using recreational drugs; (4) smoking tobacco; (5) engaging in sex with another person; or (6) sleeping with another person. Each participant was further instructed to place the t-shirt in a plastic bag (provided and identified with an arbitrary code number) during the day, when not worn, and return the shirt, in the bag, the morning following the second night at 0900 h. Almost all of male and female participants returned their shirts on time, allowing their use in the scent rating phase of the study (see below). When participants dropped off their shirts, each filled out a brief questionnaire about guidelines, if any, they had violated during the 2-day period when shirts were worn during sleep. They were told that they would lose no experimental course credit or grade points for violating any of our instructions, and that it was important that they be honest. The questionnaire also asked whether the participant had been ill, how often he or she had bathed, and whether scented soap or shampoo were used during the two-day period. Because we had told participants that, should they happen to use deodorant during the day, they should shower before putting on the shirt, we asked whether they had showered before wearing the shirt.
Scent attractiveness ratings
At 1000 h of the morning that participants of one sex returned shirts, the opposite-sex participants began reporting in groups of up to five. Following informed consent, each woman or man was placed in a separate room for rating the t-shirts. Shirts had been separated into groups of approximately 10 and each group placed in a box. In addition to the shirts worn by men and women, one unworn shirt was included in the sample. Boxes were circulated through the sample of raters present during a session. Though no attempt was made to fully randomize the order in which raters smelled shirts, it is likely that no two raters smelled them in precisely the same order. For each shirt, raters were asked to open the plastic bag and, without touching the shirt, smell it and rate the scent on three dimensions: (1) pleasantness, on a scale of 1 = very unpleasant to 10 = very pleasant; (2) sexiness, on a scale of 1 = very unsexy to 10 = very sexy; and (3) intensity, where 1 = not at all intense to 10 = very intense. They were instructed to roll the top of the bag shut before putting it back in its box and moving onto the next shirt. All researchers presenting shirts to participants for smelling were unaware of the symmetry scores of the participants who had worn them and of the MHC genotypes of t-shirt raters and wearers.
Women raters were also given a brief questionnaire to fill out, which assessed (1) whether the woman currently used a contraceptive pill or other hormone-based contraceptive; (2) the first day of the woman's last menstrual period (women were provided a calendar to assist with this task); and (3) the typical length (in days) of the woman's menstrual cycle. These data allowed our calculation of the women's probability of conception during shirt wearing and rating (see below).
In total, rating sessions lasted about 1 h and were conducted from 1000 until 1600 h the day of and from 0800 to 1500 h the day after the collection of the t-shirts.
Despite our instructions, the smell of fragrance (from perfume, soap or lotion) or smoke was evident on some shirts. Scent raters were asked to indicate if they smelled a nonhuman odor on any of the shirts. Guided by these comments, four researchers systematically smelled shirts and confirmed ones that had such odors. Of women's shirts, 14 had smell of fragrance and 4 had smell of smoke. Of men's shirts, 7 smelled of fragrance and 3 smelled of smoke. In addition, we asked men and women whether they had broken any rules. An additional 3 men said that they wore fragrance without showering, 4 claimed to smoke, and 3 said that they slept with someone; corresponding numbers for women were 4, 8, and 2. Finally, we asked whether participants had been sick over the preceding 2 days. An additional 9 men and 8 women reported sickness. The scent of men who had been sick tended to be less attractive than other men's scent (r = -.16, p <.10); sick women did not smell less attractive (r =.12, ns). As a conservative measure, all these participants' shirts were eliminated from analysis. A total of 56 men's shirts and 48 women's shirts were used for analysis. (Unlike in a previous study [Thornhill and Gangestad, 1999b], number of showers taken did not covary with scent attractiveness, and, hence, no adjustment was made for number of showers.)
Of the 89 women who reported to smell shirts, 21 used the pill or another hormone-based contraceptive (Depoprovera, Norplant), 1 was postmenopausal, and 1 did not appropriately fill in the rating sheet. Also, one woman reported an exclusively homosexual orientation; because this study examined attractiveness of scent in a heterosexual situation, we excluded her from analyses. Of the 77 men who reported to smell shirts, none reported an exclusively homosexual orientation. In total, our primary analyses of shirt raters included 65 women who did not use hormone-based contraception and 77 men.
Facial attractiveness ratings
We recruited 14 women and 15 men unfamiliar with the research to rate participants' facial attractiveness based on facial photographs. Ratings were made on a 1 (least attractive) to 10 (most attractive) scale. Raters were instructed not to rate pictured individuals whom they recognized. All opposite-sex ratings were averaged to yield an index of facial attractiveness. Internal consistency of these indices was high: for male participants, 
=.90; for female participants, =.86. (We used opposite-sex ratings only, though mean ratings made by the two sexes were highly correlated: for male participants, r =.88; for female participants, r =.88; both p <.00001.)
HLA typing
Approximately 10 ml of whole blood from each subject was collected by a phlebotomist into Vacutainer tubes containing EDTA to prevent clotting. Genomic DNA was prepared for typing by the BioTest SSP System (BioTest Diagnostics Corporation, Denvile, NJ) following manufacturer's recommended protocols. DNA was typed by polymerase chain reaction (PCR) for the HLA-A, B, and DRß loci following the manufacturer's recommended protocols. Briefly, this technique uses allele-specific primers to amplify PCR products. We determined the presence or absence of a PCR product in each reaction by resolving the products on 1.5% agarose.
Data treatment: symmetry
Repeatabilities (rIC) for the 10 measures of signed asymmetries of individual traits ranged from.79 to.93 for men (all p <.00001, mean rIC =.86) and for women from.76 to.93 (all p <.00001; mean rIC =.84). For unsigned asymmetries, rIC for men ranged from.63 to.87 (all p <.00001; mean rIC =.73) and for women from.60 to.87 (all p <.00001; mean rIC =.74).
Previous analysis on a large sample (n > 700) demonstrated that the traits studied exhibit at most small directional asymmetry (see Furlow et al., 1997). As was true in that sample, only foot width in this study showed significant but small directional asymmetry (with Bonferroni correction for number of traits; t192 = 4.22, p <.001; left > right; the mean - 0 =.3 SDs). No signed asymmetry exhibited significant platykurtosis, indicative of antisymmetry. Mean g2 was 0.78 across the traits, and hence distributions were mildly leptokurtic, as expected of FA (see Gangestad and Thornhill, 1999).
To guard against the effects of large asymmetries due to injury, FA of traits that participants reported as injured by break or sprain were excluded in these analyses if they were greater than the mean (4.6% for men, 1.1% for women). For measurement purposes, the mean FA for the trait in the whole sample of participants was substituted in these instances (see Thornhill and Gangestad, 1994).
We calculated an aggregated FA score in two different ways. First, each trait's absolute asymmetry was divided by the mean trait size ([right + left]/2) for that participant, and the FA of all 10 characters was summed to yield an overall index (relative FA). Second, each trait's absolute asymmetry was standardized (divided) by the sample mean of trait size and summed to yield an overall index (absolute FA). These two measures correlated.97 with each other and yielded nearly identical results. Thus, we report results for just one measure (the first) below. The intraclass correlation across the two measurements for the summed index was.82 for men (F96,97 = 10.10, p <.000001) and.84 for women (F99,100 = 11.13, p <.000001). The mean composite FA was.174 (SD =.052), close to the mean observed in previous studies (Gangestad and Thornhill, 1999).
Data treatment: scent attractiveness
For women, we estimated fertility (probability of conception following sex) on the basis of their day of the cycle and conception values reported in the medical literature (Baker and Bellis, 1995; Jöchle, 1973). We did so in two ways, one based on a forward method, the other on a backward method (see Baker and Bellis, 1995). For the forward method, we simply took the day of the cycle that women were at on the day of smelling based on their reported first day of last menstruation and used the actuarial table to estimate probability of conception. For the backward method, we took into account women's reported cycle length (mean = 27.9 days, SD = 4.0; average of two reports, which correlated.82, p <.0001). Women who have longer cycles, on average, ovulate later in the cycle than women who have shorter cycles (Baker and Bellis, 1995). We assumed that the typical day of ovulation was about 15 days before the end of their typical cycle (e.g., day 14 in a 29-day cycle) and that the average day that actuarial tables are based on is day 14. Women's probability of conception was then based on how far they had come toward or after their assumed day of ovulation and the actuarial tabled values for women in general. The two methods yielded values that correlated.62 (p <.0001), and results based on each were highly similar. As each method may have some validity that the other lacks, we averaged the two sets of values for our main analyses. (Full results are available from the authors.) The minimum value was.01, the maximum.39 (approximately day 12 of the cycle). Fertility steeply rises from near-zero to near.2 shortly after the average end of menses (days 5–6) and sharply falls from >.2 to <.1 immediately after mean ovulation (day 14). Undoubtedly, our method of estimating conception risk is likely to be in error for some individual women (though, across all women, it should correlate highly with actual risk). As it is unlikely that error introduced by our method spuriously generates significant relationships, the associations we report probably underestimate rather than overestimate true relationships with conception risk.
The mean correlation between individual participants' "pleasantness" and "sexiness" ratings was.78 for women raters and.82 for men. Hence, these ratings were averaged for each sex into a total attractiveness index for reported analyses. Analyses on specific ratings yielded highly similar results.
Data treatment: MHC genetics
MHC similarity for individual male–female pairs was computed as the total number of alleles shared across the three loci. Hence, if both individuals were A1, A3 at the A locus, similarity for that locus was scored 2. If one individual was homozygous for A1 and the other was A1, A3, similarity was scored 1. If both individuals were homozygous for A1, similarity was scored 2. Similarity scores could thus range from 0 to 6. Actual similarity for pairs averaged 1.25 alleles (SD = 0.37). MHC heterozygosity was the number of loci at which the participant was heterozygous. Mean heterozygosity was 2.69 (SD = 0.54).
Because our multiethnic sample may have unique characteristics, MHC allele frequencies were calculated within the sample. Total number of allele observations within the sample was 352, 348, and 348 for the A, B, and DRß loci, at which we detected 20, 28, and 14 different alleles, respectively. Relative frequencies of all alleles are given in Figure 1. We computed the commonness of a participant's alleles as the proportion of alleles within the sample of the type possessed by the participant, averaged across the participant's six alleles. Commonness scores averaged 0.113 (SD = 0.032, range = 0.04–0.21).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
For all scent effects that have been found in prior research on humans (preference for the scent of MHC dissimilarity, women's preference for the scent of symmetrical men when fertile, preference for the scents of facially attractive individuals, men's preference for the scent of fertile women), we used directed tests (Rice and Gaines, 1994
), which use a p value of.04 for a predicted relationship and.01 for a relationship opposite of prediction (rather than.025 for each, as with a two-tailed test). All directed tests are labeled as such. In all other instances, we used two-tailed tests.
Although unsigned (i.e., absolute) FA tends to be non-normally distributed (Swaddle et al., 1994), Monte Carlo analyses reveal that significance tests on parametric correlations involving FA are robust (Gangestad and Thornhill, 1998b), and hence we performed standard parametric analyses on this variable.
Descriptive statistics
On average, men's scents were rated 4.54 (SD = 0.95) on the 1–10 attractiveness scale. Women's shirts received a mean rating of 4.71 (SD = 0.63). There was no significant sex difference in mean scent attractiveness (t102 = 1.08). Men's shirts tended to receive more variable ratings than did women's shirts, though this difference was only marginally significant (F1,102 = 2.94, p =.089; Levene's test for equality of variances). On average, men rated the unworn t-shirt's scent attractiveness 3.82 and women rated it 4.02. On average, women's and men's scents were rated by the other sex as more attractive than the blank shirt's scent (t76 = 4.97 and t64 = 3.38, respectively, p <.002).
Preferences for MHC dissimilarity
Analyses that treat the t-shirt wearer as the unit of analysis (and examine whether raters who possess dissimilar MHC genotypes rate their scents more favorably) control for factors that contribute to overall differences in scent attractiveness and, hence, should offer the greatest power to detect preferences for MHC dissimilarity (Wedekind and Füri, 1997). Using the procedures of Wedekind and Füri (1997), we computed for each t-shirt wearer a correlation between individuals' ratings of scent attractiveness and the number of MHC alleles shared between the t-shirt wearer and the rater. We then asked whether these correlations differed significantly from zero in the predicted negative direction.
We found no evidence that men's scents were preferred by women with dissimilar MHC genotypes (mean r =.025, t54 = 1.39, directed). In contrast, we did find women's scents to be preferred by men with dissimilar MHC genotypes (mean r = -.033, t47 = 1.74, directed p =.040). The mean correlations significantly differed from one another (t101 = 2.30, p =.024), revealing a sex difference in the preference for MHC dissimilarity. Specifically, the attractiveness of women's scents to men appears to be positively influenced by MHC dissimilarity to a degree greater than the attractiveness of men's scents to women.
Analyses that treated the scent rater as the unit of analysis (and examined whether raters prefer the scents of those with dissimilar MHC genotypes) were also conducted. We computed raters' preference for dissimilar MHC as the slope of their ratings regressed on MHC dissimilarity (b). For neither men nor women raters did we find evidence of preference for MHC dissimilarity: For men, mean b =.007, t69 =.42, directed ns; for women, mean b = -.051, t61 = -1.41, directed ns. The mean preferences for the two sexes did not differ (t130 = 1.17). Again, these analyses are less powerful than those treating the t-shirt wearer as the unit of analysis, which may account for the difference in results.
Despite no evidence for an overall preference in women for the scent of MHC-dissimilar men, women's preference for MHC dissimilarity could be moderated by their fertility status across the menstrual cycle. However, we found no evidence of a role for fertility status. Women's probability of conception across the cycle had a near-zero correlation with their preference for MHC dissimilarity (r =.029, directed).
Possibly, preference for dissimilarity is not a linear function. Perhaps, for instance, a moderate degree of dissimilarity is preferred, with high levels of similarity and dissimilarity disfavored (see Penn, 2002). If that is the case, then targets' mean level of MHC sharing with others should predict, for individual targets, the correlation between the target's MHC sharing with a rater and the rater's attraction to the target's scent: those targets who, on average, share relatively many MHC alleles with others (i.e., because they have common alleles) should be particularly preferred by those who are dissimilar, whereas targets who, on average, share few MHC alleles with others (i.e., because they have rare alleles) should be more preferred by those who are relatively similar (e.g., share even a single allele). This hypothesis was not supported. For the 48 female targets, the association ran in the direction anticipated if there were a preference for the scent of those with moderately similar MHC genotypes, but fell short of significance (r = -.19). For the 55 male targets, the association between their mean MHC sharing with others and the correlation between raters' attraction to their scent and MHC sharing with the target ran in the opposite direction (r =.38, p <.01). (Hence, the scent of men who had rare MHC alleles and therefore shared relatively few alleles with men were more likely to be preferred by women with dissimilar MHC genotypes.) These correlations significantly differ from one another (z = 2.91, p <.01).
Preferences for symmetry
As in previous research (Gangestad and Thornhill, 1998b; Thornhill and Gangestad, 1999b), we calculated a preference for the scent of symmetry for each woman, which is the slope of the woman's scent attractiveness ratings regressed on men's symmetry (with positive values reflecting preference for symmetry and negative values preference for asymmetry). We correlated these preferences with women's probability of conception. Consistent with previous studies, women's preference for symmetry was predicted by their fertility status (r =.269, directed p =.019). Regression analyses revealed that, at zero probability of conception, women had a significant preference for symmetry (t63 = 2.45, p <.017), a pattern unique to this study. Moreover, this preference for symmetry increased as a function of their conception risk (see Figure 2). Mean scent-attractiveness ratings made by women (n = 22) who had conception risks of at least 0.15 (day 6–day 14) correlated (-.29; directed p =.02) with men's FA, whereas mean ratings made by all other women (n = 27) were more loosely correlated (-.11; ns) with men's FA; these correlations significantly differed from one another (t53 = 2.57, p =.008). Women's conception risk did not significantly correlate with the mean rating they gave all scents (r = -.04, ns) or the rating they gave the unworn shirt's scent (r =.06, ns). As also found by Thornhill and Gangestad (1999b), women's mean scent attractiveness to men did not correlate with women's FA (r = -.063, ns).
|
Preferences for MHC heterozygosity
We examined preferences for MHC heterozygosity by correlating the number of MHC loci at which individuals were heterozygotes with their mean scent attractiveness to the opposite sex. Male heterozygosity was positively correlated with scent attractiveness (r =.275, p =.049). Follow-up analyses showed that this effect was driven by heterozygosity at the B locus (r =.329, p =.017); no effect was observed for either the A or DRß locus (r = -.025 and.161, respectively, ns). As these correlations do not significantly differ from one another, however, these locus-specific differences may not be robust and must be interpreted cautiously. Female heterozygosity did not predict scent attractiveness (r =.004, ns). Sex did not interact with heterozygosity to predict scent attractiveness (t93 = 1.56, p =.12).
We examined whether women's preferences for heterozygosity change across the menstrual cycle. These preferences were calculated as the slope of women's ratings regressed on men's MHC heterozygosity. They correlated with women's probability of conception at a level just short of statistical significance(r = -.239, t63 = 1.96, p =. 055), such that women outside of the fertile phase tended to have the strongest scent preferences for heterozygosity. The correlations of women's preference for symmetry and preference for heterozygosity with conception risk differed significantly from one another (t62 = 2.87, p =.006), indicating that these female preferences at least partly invoke different systems.
Preferences for rare MHC alleles
We examined preferences for rare MHC alleles by correlating the mean frequency of MHC alleles within this sample across the six alleles an individual possesses at the three loci with mean scent attractiveness. For women, commonness of alleles significantly predicted the attractiveness of their scents, such that those with more common alleles tended to have more attractive scents (r = .309, p =.032). Men's scent attractiveness was not associated with the commonness of their alleles (r =.025, ns). Sex did not significantly moderate the relationship between commonness of alleles and scent attractiveness (t96 =.980). Women's preferences for rarity of MHC alleles (calculated as the slope of their ratings regressed on men's MHC allelic rarity) did not significantly covary with their fertility status (r = -.132).
Preference for the scent of facial attractiveness
As found in previous studies (Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999b), women's facial attractiveness predicted their scent attractiveness (r = .268, directed p = .041). Unlike in the previous studies, men's facial attractiveness did not predict their scent attractiveness (r = -.053, directed ns). Nonetheless, sex did not significantly moderate the relationship between facial and scent attractiveness (t100 = 1.30).
Previously, Thornhill and Gangestad (1999b) found that women's probability of conception predicted their preference for the scent of facially attractive men. In this sample, though the correlation was in the same direction, no significant association emerged (r =.096, directed). In this sample, FA did not predict men's or women's facial attractiveness (r = -.086 and -.029, respectively, ns).
Male preferences for fertile women
Women's estimated probability of conception at the time of wearing the t-shirt was correlated with their mean scent attractiveness. As found by Singh and Bronstad (2001), men were more attracted to the scents of women of high fertility status than those of low fertility status (r =.334, directed p =.021).
Associations of FA and facial attractiveness with MHC dissimilarity, heterozygosity, and commonness of alleles
Neither FA nor facial attractiveness was significantly predicted by mean MHC dissimilarity to others, MHC heterozygosity, or commonness of MHC alleles within either sex (see Table 1). These findings bolster the conclusion that, to the extent that the scent of MHC dissimilar individuals is favored, the phenomenon is unrelated to women's preference for the scent of symmetrical men when fertile. They also add further evidence that women's scent preference for men who are heterozygous at MHC sites is distinct from their scent preference for symmetrical men.
|
Additional analyses
For exploratory purposes, we correlated men's mean scent attractiveness rating with their age, weight, and self-reported socioeconomic status of family of origin (on a 5-point scale, where 5 = upper class, 3 = middle class, 1 = lower class). These correlations were not significant (r =.04,.21, and.08, respectively). For women, mean scent attractiveness correlated at a marginally significant level with age (r = -.28, p =.058 and socioeconomic status, r =.26, p =.077, but not with weight, r = -.10, ns). Analyses comparing scent attractiveness of men and women across ethnicities revealed no significant differences (F5,50 =.44 and F4,47 =.38).
We conducted additional exploratory analyses. We correlated preference for the scent of symmetry with the judge's own scent attractiveness, age, FA, physical attractiveness, and number of self-reported sex partners. Younger women particularly preferred the scent of symmetrical men (r =.34, p =.005), even with women's conception risk controlled (r =.30, p =.015). In contrast, older men particularly preferred the scent of symmetrical women (r =.24, p =.045). These correlations significantly differ(z = 3.42, p <.001). Men whose scent was rated as less attractive also preferred the scent of symmetrical women (r =.38, p =.009; n = 48 for this analysis). Women's scent attractiveness did not predict their preference for the scent of symmetry (r = -.14, ns). Again, the correlations significantly differed between the sexes (z = 2.18, p <.03). No other correlations were significant. We correlated women's preference for MHC dissimilarity, MHC heterozygosity, and MHC allele frequency with the same variables. Only a single correlation was significant (men's age negatively predicted their preference for heterozygosity), one that could easily be due to chance.
Older and more facially attractive men particularly preferred the scent of women near ovulation (r =.23 and.25, p <.05). Men's preferences tended to covary with one another. Men's preference for the scent of symmetrical women was negatively associated with their preference for MHC heterozygotic women (r = -.32, p =.004). Their preference for women near peak fertility also negatively predicted their preference for heterozygotic women as well as their preference for women with rare MHC alleles (r = -.24 and -.26, p <.04). Finally, men's preference for the scent of MHC-dissimilar women was negatively associated with their preference for facially attractive women (r = -.36, p =.003). (Preference for MHC dissimilarity also predicted preference for rare MHC alleles [r =.32, p =.006], an association that is not surprising, as women with rare alleles are relatively dissimilar to men at MHC.) No female preferences significantly correlated with one another.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This study used t-shirts worn by each sex for 2 night's sleep and then smelled by the opposite sex to examine the relationships between heterosexual attractiveness of one's body odor and one's body fluctuating asymmetry (FA), facial attractiveness, MHC dissimilarity, MHC heterozygosity, and possession of rare MHC alleles. Our results address four adaptive hypotheses that have been proposed for scent-based, MHC disassortative mate preferences: outbreeding, production of MHC-heterozygous offspring or of offspring with rare MHC alleles, and increased within-brood MHC diversity. Additionally, our results address issues concerning the prior finding that women, when fertile in their menstrual cycle, particularly prefer the scent of symmetrical men.
MHC allele frequencies may differ among ethnic groups. Our sample was multiethnic, and we based allele frequencies on the whole sample. Our finding (based on the whole sample) that ethnicity has no effect on scent attractiveness in either sex suggests that our results on MHC scent effects are meaningful. However, our ethnic categorization was self-reported using broad categories.
The scent of symmetry
Compared to the body scent of relatively asymmetric men, relatively symmetrical men's scent is most attractive to normally ovulating women (women not using hormonal contraception) during their period of high fertility in the menstrual cycle. We have found little evidence that women show scent preference related to male body symmetry during cycle times of low fertility. In this study only, and not in the studies of Gangestad and Thornhill (1998b) and Thornhill and Gangestad (1999b), women with zero conception risk exhibited a significant preference for the scent of symmetry. That women at high conception risk in their menstrual cycles prefer the scent of male symmetry has now been found in four separate studies (also see Gangestad and Thornhill, 1998b; Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999b). Although not examined in this study, prior studies have consistently found that women using hormone-based contraception do not show the preference for the scent of men's symmetry. Moreover, the effect of the scent of symmetry appears to be sex-specific. We found no evidence that men find the scent of symmetrical women more attractive, a result that replicates that of Thornhill and Gangestad (1999b).
Apparently, the attractiveness of the symmetry scent to women does not arise from hygienic differences between men in relation to their symmetry (e.g., number of times they showered; see Thornhill and Gangestad, 1999b). Nor does it appear to arise from not following our research guidelines for diet, alcohol and drug use, or sleeping behavior over the period in question (see also Thornhill and Gangestad, 1999b). The symmetry of men was unknown to the women rating scents and to the researchers involved in collecting the scent ratings; the only identifying information on the plastic bag containing each shirt was an arbitrary participant number. Although women exhibit increased olfactory sensitivity during the fertile aspect of the menstrual cycle (Doty, 1981; Kohl and Francoeur, 1995; Vroon, 1997), the odor preference for symmetry shown by normally ovulating women cannot readily be explained by an increased general olfactory sensitivity or response with high fertility. There is no significant relationship between women's intensity ratings of the shirts and male symmetry (Gangestad and Thornhill, 1998a; Thornhill and Gangestad, 1999b). In sum, we know of no explanation for our results other than that normally ovulating women use a chemical(s) in men's sweat or skin as a basis for discriminating men who have and have not experienced perturbations that generate FA. Thornhill and Gangestad (1999b) provided a number of hypotheses about the chemistry of the scent of men's symmetry.
Fertile women's preference for the scent of symmetry may, in part, account for the pattern of pair-bonded women tending to choose symmetric men as EPC partners. Given that symmetric men appear to invest less in their romantic relationships than asymmetric men, these choices are probably for genetic benefits rather than for increased male investment (Gangestad and Thornhill, 1997a,b).
Conceivably, the scent of symmetric men is preferred by women at mid-cycle because of a preference for adequate sperm. Manning et al. (1998; also Baker, 1997) have shown a positive correlation between ejaculate size, sperm quality, and body symmetry in men. This, however, is not necessarily an alternative to good-genes choice.
Preferences for MHC dissimilarity
The results from our study of the scent attractiveness of MHC dissimilarity repeated, in part, earlier studies. Our study reveals that men find the scent of MHC dissimilarity attractive, but we detected no clear MHC-dissimilarity preference in women. The sex difference in preference for MHC dissimilarity was statistically significant. Overall, then, there is positive evidence of MHC-dissimilarity preference exhibited by men from both of the t-shirt studies that have examined it (Wedekind and Füri, 1997; present study). Women have shown an MHC-dissimilarity preference in two of three t-shirt studies that tested for it (positive in Wedekind et al., 1995; Wedekind and Füri, 1997; negative in present study). The negative assortative pairing in Hutterite couples reported by Ober et al. (1997) could be the result of an MHC dissimilarity preference in only one sex or both sexes and hence does not shine any light on the issue of the consistency of the effect in each sex. Given that our study had a larger sample of women wearing shirts and more raters than either study showing a positive effect for women's scent and detected an MHC-dissimilarity effect for men, the absence of an effect for women may be meaningful. However, the two t-shirt studies finding an effect for women involved an ethnically homogeneous sample that may have reduced individual differences in odor preferences unrelated to MHC; these two studies also differed from our study in some other ways.
We did not find any evidence of a relationship between women's conception risk across the menstrual cycle and their preference for the scent of MHC dissimilarity. Thus our results indicate that women at high risk of conceiving do prefer the scent of symmetry but not the scent of MHC dissimilarity. It should be kept in mind, however, that Wedekind and colleagues found scent preference for MHC dissimilarity in women at relatively high conception risk, but none for women who were infertile due to contraceptive-pill use.
Based on our results, we conclude that men's scent of symmetry and men's scent of MHC dissimilarity involve different pheromonal systems, at least in part. This conclusion is bolstered by the fact that FA and mean MHC dissimilarity to others does not covary across individuals (of both sexes). Either different chemicals activate one receptor adaptation or multiple receptors are involved, each with chemical specificity. The receptor responsible for detecting the scent of symmetry would appear to be sexually dimorphic in function or present only in females. Furthermore, the sensitivity of this detector may be affected by menstrual cycle hormones. At least two different organs appear to mediate scent perception in humans: the vomeronasal organ (e.g., Wysocki, 1989) and the main olfactory epithelium (Engen, 1982).
Because women's attraction to the scent of men's symmetry is only seen in normally ovulating women at high conception risk, it has been hypothesized that this attraction is an adaptation that functions to obtain for offspring genes that encode developmental stability and related overall fitness, particularly through EPC (Gangestad and Thornhill, 1998b; Thornhill and Gangestad, 1999b, 2003). The lack of a detectable association between women's preference for MHC dissimilarity and conception risk suggests that the MHC preference does not involve good-genes mate choice because the benefit of choosing good genes—placement of a mate's genes in offspring at conception—cannot be gained in the absence of preference during times of high-conception risk. This interpretation also applies to what some investigators would label compatible genes (Penn, 2002; Zeh and Zeh, 2001). The interpretation that good-genes (including compatible-genes) mate choice is not the basis of women's MHC-dissimilarity preference is further supported by our finding that men have stronger preference for MHC dissimilarity than do women. If MHC-disassortative mating functions to obtain good genes for offspring, women are expected to exhibit a stronger preference than men. These lines of evidence, then, weaken the three parasite-mediated sexual selection hypotheses as explanations for MHC preferences.
Nonetheless, a good-genes–choice hypothesis could still apply if women's preference for MHC dissimilarity at midcycle is only manifested under conditions in which the benefits of good-genes mate choice can be realized (e.g., when diseases are prevalent). Such conditionality in the preference may have been selected in the context of EPC due to the concomitant high potential costs (e.g., divestment of pair-bond mate). There is evidence for conditionality in the MHC-odor system of mice. Mice infected with mouse hepatitis virus produced more MHC-heterozygous offspring than sham-infected mice, which may result from the expression of the MHC-dissimilarity preference only when a virus threatens offspring (Rülicke et al., 1998).
A stronger scent preference in males for MHC dissimilarity, as well as the absence of a link between conception risk and this preference in women, is also inconsistent with the inbreeding avoidance hypothesis. Females are expected to have a higher cost of inbreeding than males as a result of more female than male parental investment.
One possible explanation for why MHC preferences appear to be stronger in men than women is that they partly function in men to increase ability to invest discriminatively in their own offspring. Both sexes may partly use odors of offspring to detect kinship (Gall and Weisfeld, 2001). Potentially, a male could most reliably detect by olfaction his own MHC alleles in offspring if he shares few alleles with his mate.
Preferences for MHC heterozygosity
MHC heterozygosity was not correlated with either facial attractiveness or with FA in either sex. Heterozygosity is sometimes positively associated with relatively low FA across species, and there are data suggesting this association from some studies of humans (reviewed in Møller and Swaddle, 1997; Thornhill and Møller, 1997).
Women appear to prefer the scent of MHC-heterozygous men. Furthermore, a trend in the data suggests that women may prefer the scent of MHC-heterozygous men particularly when outside of the fertile phase of the cycle. Women's scent preference for MHC heterozygosity may be analogous to the female stickleback's scent preference for males with high MHC diversity found by Reusch et al. (2001). We found no evidence that men have a scent preference for MHC heterozygosity.
If future research confirms a greater female preference for the scent of MHC-heterozygous men outside of the fertile phase, two alternative explanations appear plausible. First, the function of choosing a heterozygous mate could be to diversify a brood's MHC, thereby reducing the chances of a within-family epidemic caused by a pathogen. This preference would be more important for choosing a long-term mate (i.e., one with whom a female would have multiple offspring), and selection for the preference may have been stronger on females than on males. The preference may be attenuated midcycle because other preferences, such as preferences for intrinsic genetic quality, increase in potency at that time. Second, infertile women's preference for heterozygous mates may reflect adaptation for securing male material benefits for self and/or offspring. If heterozygosity increases survival or vigor, it might promote male investment capability. Although we found no evidence that MHC heterozygosity is associated with developmental instability, it could be associated with disease resistance (see Penn et al., 2002; Penn and Potts, 1999). Given the substantial MHC diversity that exists, MHC heterozygosity should be only weakly heritable and hence not be the basis of mate choice for producing heterozygous offspring. In either case, as the correlation between women's conception risk across the cycle and women's preference for the scent of symmetry differed significantly from the correlation between conception risk and preference for heterozygosity, the two preference systems may involve distinct pheromone systems, at least in part.
Preferences for rare alleles
We found no evidence that either sex prefers scents associated with rare MHC alleles. Indeed, men in our study preferred scents associated with common MHC alleles. No association between women's conception risk and preference for the scent of rare MHC alleles was detected. A rare-allele preference is predicted from the hypothesis that MHC-dependent mating preferences function to generate offspring that have reduced autoimmune problems (associated with common MHC alleles) or that cannot be invaded by parasites as a result of possessing alleles to which parasites are not well adapted.
Gestational drive (Haig, 1997), in which a maternal allele disfavors offspring during gestation that do not inherit it, could account for a sex difference in preference for the scent of common alleles. Gestational drive is delayed meiotic drive and, like meiotic drive, is advantageous to the alleles involved, but disadvantageous to the remainder of the maternal genome. Gestational drive may especially be a property of rare female alleles because any driving effects of common alleles are likely to be successfully countered by the evolution of genes that prevent driving. A mate preference in males for females who possess common MHC alleles may function to avoid mates with alleles that show gestational drive and thereby reduce the likelihood of abortion of offspring. Females do not face this problem in mate choice.
MHC allele rarity did not covary with symmetry in either sex. It was previously hypothesized that the scent attractiveness of men's symmetry could arise if symmetric men possess rare MHC alleles and women prefer rare or dissimilar MHC alleles (Gangestad and Thornhill, 1998b; Thornhill and Gangestad, 1999b). This study provides no support for this hypothesis.
Preferences for scents associated with facial attractiveness
Women's facial attractiveness, but not men's, significantly correlated with body scent attractiveness. Three studies to date have reported this association for women (also see Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999b), indicating that women's facial attractiveness and scent attractiveness convey overlapping information.
Our study did not replicate the finding from two previous studies that women with high conception risk prefer the scent of facially attractive men (Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999b). The association between men's facial attractiveness and their masculinity varies across populations and samples (for a review, see Thornhill and Gangestad, 1999a), which may partly account for the instability of this finding. Possibly, women prefer the scents of more androgenized men midcycle (due to androgen-derived substances in sweat; e.g., Grammer, 1993), but in some samples these men are not the most attractive in terms of scent.
Facial attractiveness was not significantly correlated with MHC dissimilarity, MHC heterozygosity, or commonness of MHC alleles in either sex and hence may be independent of these MHC traits.
Men's preferences for women at high conception risk
We also found that men exhibit significant scent preferences for women who are at high conception risk in the menstrual cycle. Using a within-subject design, Singh and Bronstad (2001) previously found evidence for this preference. Using a between-subject design similar to the current study, we previously did not find this preference (Thornhill and Gangestad, 1999b). The within-subject design is statistically more powerful (as it controls for individual variation in scent across women), and we suspect that the variation in this result is due to the lower power of our between-subject design. The chemical basis for this effect is unknown. The scent of chemicals similar to estrogen evoke hypothalamic responses in men, but not in women (Savic et al., 2001), which may mediate this effect. Alternatively, the effect could involve variation in copulins present in women's vaginal secretions. There is some evidence that men show an olfactory, sexual response to copulins (Grammer and Jutte, 1997).
|
ACKNOWLEDGEMENTS |
|---|
We thank P. Andrews, T. Armijo-Prewitt, M. Aronov, C. Fincher, K. Ganster, K. Schancer, C. Schwenke, S. Singh, N. Terpstra, and J. Vigil for their assistance in measuring participants, collecting other data, and data entry. C. Copelin's extra effort and professional attitude allowed blood samples to be obtained from almost all our research participants. The research was supported by a grant from the Sense of Smell Institute (formerly known as the Olfactory Research Fund). C. Fincher, D. Penn, C. Wedekind, R. Ydenberg, and two anonymous reviewers provided useful comments on the manuscript.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Amoore JE, Pelosi P, Forrester LJ, 1975. Specific anosmias to 5 alpha-androst-16-en-3-one and alpha-pentadecalactone: the urinous and musky primary odours. Chem Sens Flavour 2: 401-425.
Apanius V, Penn D, Slev P, Ruff LR, Potts WK, 1997. The nature of selection on the major histocompatibility complex. Crit Rev Immunol 17: 179-224. [ISI][Medline]
Baker RR, 1997. Copulation, masturbation and infidelity: state-of-the-art. In: New aspects of human ethology (Schmitt A, Atzwanger K, Grammer K, Schafer K, eds). New York: Publisher; 163–188. <