Consistency and Individual Differences in Facial Attractiveness Judgements: An Evolutionary Perspective

Consistency and Individual Differences in Facial Attractiveness Judgements: An Evolutionary Perspective

Ian Penton-Voak


DARWINIAN approaches to the study of facial attractiveness are based on the premise that attractive faces are a biological “ornament” that signals valuable information to a potential mate. Evolutionary approaches to interpersonal attraction have grown in numbers over recent years, and have provided a fruitful theoretical base from which testable hypotheses have been generated and, in many cases, supported. One potential criticism of this field of research, however, is that it is almost exclusively structuralist in nature. Perception of facial attractiveness is assumed to be data-driven; the properties of a particular set of facial features are the same irrespective of the perceiver. From this perspective, attractive faces are perceived as such independently of any of the characteristics of the observer, such as culture or upbringing. Systematic studies of the differences in facial attractiveness judgements between individuals, or of an individual’s changing judgements of facial attractiveness over time, are rare. Although individual differences in attractiveness judgements obviously exist, such behavioral variation is rarely given a biological interpretation.

In fact, evolutionary biologists studying the sexual behavior of other species provide a framework suggesting that individual differences in human attractiveness judgements may have parallels with the behavioral variation shown by other species. While individuals appear to share basic criteria of facial attractiveness, humans may learn the “fine grain” of the faces they find attractive in an analogous way to the learning or imprinting experiences of some other animals early in their lifetime. Furthermore, individuals may adopt different reproductive strategies as a result of life history factors–these differing strategies may lead to adaptive preferences for different face types. In this paper, we briefly review “structuralist” facial attractiveness research from an evolutionary perspective, before moving on to individual differences in the perception of attractiveness. We then speculate as to how such differences may derive from learning and life history factors.

Structuralist Approaches to Facial Attractiveness

While both males and females claim in self-report that physical attractiveness is not of primary importance when choosing a partner (Buss, 1989), the single best predictor of satisfaction with a “blind date” is facial attractiveness for both men and women (Walster et al., 1966). Given the apparent importance of the face in mate choice decisions, and the centrality of mate choice theories to evolutionary explanations of behavior, sexual selection may have acted on human facial characteristics. Many researchers have searched for evidence that the characteristics of attractive faces are as much biological as cultural.

The first goal of researchers proposing a biological predisposition to find certain faces attractive is to demonstrate that the learning of arbitrary cultural values has little influence on judgements of attractiveness–that attractive faces are considered so universally. Researchers have demonstrated high cross-cultural agreement in ratings of attractiveness of faces of many ethnicities, in many geographic locations (e.g., Cunningham et al., 1995; Perrett et al., 1994). Even Darwin (1871) acknowledged the possibility of the universality of facial attractiveness across cultures when he noted that explorers had remarked that indigenous peoples around the world had similar standards of beauty.

A second line of evidence in favor of a biological rather than an arbitrary cultural basis of physical attractiveness judgements comes from studies of infant preferences for face types. Langlois et al. (1987) showed pairs of female faces to infants between 2-3 months and 6-8 months old. The attractiveness of the faces had been judged by adults prior to presentation of the images to the infants. Infants preferred to look at the more attractive face of the pair, indicating that even at two months of age, adult-like preferences are demonstrated. A further study replicated and extended the finding to show that infants also expressed adult-like preferences for Caucasian male faces, Black female faces, and even the faces of other infants (Langlois et al., 1991). Importantly, Slater et al. (1998) demonstrated that newborn infants prefer attractive faces.

Infants have been demonstrated to form cognitive prototypes extremely rapidly: Exposure to exemplar faces in early infancy may allow infants to learn facial configurations. So, although infants demonstrate adult-like preferences very soon after birth (e.g., Slater et al., 1998), this doesn’t necessarily imply that children are born with “innate” knowledge of attractiveness. Nonetheless, the learning process occurs rapidly, without any obvious linguistic component and at a very young age. Such learning could be considered to be effectively “culture free” and hence possibly more analogous to an imprinting process than the learning of cultural norms.

Three Theories of Human Facial Attractiveness

The rapid learning of attractiveness judgements and impressive cross-cultural consistency indicate that individuals around the world respond fairly similarly to facial attributes. Evolutionary psychologists consider that attractive faces are considered so as they advertise some kind of heritable quality–an attractive face is a “health certificate” indicating an individual’s value as a mate (Thornhill & Gangestad, 1999a). Selection has favored preferences for healthy, fertile mates. Research has concentrated on three characteristics of faces that may honestly advertise health and viability: symmetry, averageness, and the size of sexually dimorphic traits. These three theories are briefly reviewed below.


Bodily symmetry is thought to be an indicator of developmental stability. Deviations from symmetry result from an organism’s failure to cope with various inclement environmental (e.g., climate, malnutrition, parasitization) and genetic (e.g., inbreeding) factors (Moller, 1997). Better “quality” individuals may resist environmental hazards more effectively than poorer quality individuals; some of this quality may be heritable. As such, a preference for symmetry in sexual partners may have been favored by natural selection. In non-human animals, symmetry appears to be correlated with reproductive success (Moller & Thornhill, 1998).

Several researchers have studied the effects of symmetry on human sexual selection. Surprisingly, several studies failed to find any preference for symmetry in manipulated human faces (see Perrett et al., 1999, for review). These studies employed chimaeric facial stimuli: The images presented were mirror reflections of either the left or right side of an individual face. Chimaeric techniques may introduce unnatural artifacts such as double blemishes (spots or moles that are mirrored in both sides of the face) and unnatural feature sizes (an individual whose mouth is offset to the left will have an extremely wide mouth in left-left chimaeric pairs, or an extremely narrow mouth in right-right images) that reduce the attractiveness of the resulting stimuli. Studies of asymmetry in natural faces (e.g., Grammer & Thornhill, 1994) and more recent studies using manipulated, but not chimaeric, stimuli (e.g., Perrett et al., 1999) indicate that facial symmetry does make some contribution to both male and female facial attractiveness in humans.


An influential theory of facial attractiveness is that average facial configurations are optimally attractive, as such faces are thought to indicate high levels of heterozygosity in their owners (Symons, 1979; Thornhill & Gangestad, 1999). Because genes in an individual lead to the development of proteins that form the environment for parasites, and because these pathogens are generally best adapted to proteins that are common in the host population, rare alleles in a host may result in rare proteins that pathogens are poorly adapted to. Average faces, indicating heterozygosity, may thus be preferred, as their owners may be more likely to possess proteins to which pathogens have not adapted. Additionally, extreme (non-average) genotypes are more likely to be homozygous for deleterious alleles (Thornhill & Gangestad, 1999). A preference for averageness is compatible with both cognitive theories of prototyping (see Langlois & Roggman, 1990) and work in theoretical biology that suggests a preference for “average” phenotypes would rapidly replace random mating (Koeslag, 1994). The averageness hypothesis, however, has received only mixed empirical support. While early studies favored the hypothesis (e.g., Langlois & Roggman, 1990), other studies find that although averageness is certainly attractive, it can be bettered. Consistent deviations from averageness, detailed in the next section below, lead to an increase in attractiveness in both male and females faces (e.g., Perrett et al., 1994, 1998; Grammer & Thornhill, 1994).

Average composite faces tend to have smooth skin and be symmetric; these factors, rather than averageness per se, may lead to the high attractiveness attributed to average faces (Alley & Cunningham, 1991). Dissociating symmetry and averageness is problematic, but a recent paper by Rhodes et al. (1999) varied averageness and symmetry independently and found that both contributed to attractiveness judgements. Pollard et al.’s 1999 study, however, used facial-metric measurements of feature size rather than composites to calculate averageness, and found that faces with features close to the mean size of the population studied were rated as average, rather than high, in attractiveness.

Secondary Sexual Characteristics

Despite the attractiveness of average faces, consistent deviations from average feature proportions result in increased attractiveness ratings. In other species, females exhibit preferences for exaggerated secondary sexual characteristics in males. The peacock’s tail is often cited as a classic example of “honest advertisement,” whereby males demonstrate their quality by displaying costly ornaments. In mammals, the growth of secondary sexual traits is linked to levels of androgens (Owens & Short, 1995), which depress immune system function (Folstad & Karter, 1992). Hence, only males in good condition can bear the “handicap” of large secondary sexual traits that represent an honest advertisement of male viability.

Differences in levels of male and female sex hormones at puberty are thought to largely account for sex differences in adult face shapes. High testosterone levels cause forward growth of the brow ridges, and an increase in the size of the bones of the jaw, lower face and cheekbones–“masculine” facial features (Thornhill & Gangestad, 1999a). Estrogen inhibits this growth, leading to “feminine” face shapes with high eyebrows, gracile jaws and fuller lips. A preference for sex typical traits may operate in females’ judgements of male facial attractiveness, and male’s preferences for female faces.

Considerable evidence suggests that extremely feminine female faces are considered attractive. A wide variety of techniques ranging from measurement of facial photographs of women (Grammer & Thornhill, 1994), through studies of facial composites (Perrett et al., 1994; 1998) to the generation of attractive female face shapes using genetic algorithms (Johnston & Franklin, 1993) indicate that feminine features indicating estrogenized female faces increases their attractiveness cross-culturally. Sex typical female features (small lower face, a relatively flat mid-face, full lips and high eyebrows associated with a lack of brow ridge prominence) may indicate youth (as estrogen levels decrease in adult females with age), and reproductive health (Symons, 1979, 1994; Thornhill & Gangestad, 1999a). A wealth of studies indicate that femininity (caused by an exaggeration of sex typical features) rather than averageness is attractive in female faces. Given the probable “signaling” properties of estrogenized female faces, a male preference for such features is potentially adaptive.

There is some evidence for female preferences for exaggerated male facial characteristics. Scheib et al. (1999) found a positive relationship between attractiveness and two markers of facial masculinity (cheekbone prominence and jaw size). Cunningham et al. (1990) and Grammer & Thornhill (1994) used facial-metric measurements, and found a female preference for large jaws in males. “Masculine” features, such as a large jaw and a prominent brow ridge are reliably associated with ratings of dominance in photographic, “indenti-kit” and composite stimuli (e.g Berry & Wero, 1993; McArthur & Berry, 1987; Perrett et al. 1998). Facial dominance appears to correlate with status in some human hierarchies (Mueller & Mazur, 1997) and facial dominance in adolescent males is associated with earlier age at first copulation (Mazur et al., 1994). Nonetheless, the relationship between facial dominance and attractiveness is unclear–some studies find a positive relationship (Keating, 1985) while others find preferences for “feminized” or “baby-faced” male faces (Perrett et al., 1998). The failure of researchers to agree on the characteristics of attractive male faces is especially interesting as the predictions from cross-species literature are clear: Females tend to prefer exaggerated male traits in most species (Andersson, 1994).

Female preferences for “feminine” or “baby” male faces can perhaps be explained in terms of the personality attributions that such faces attract. Although biological hypotheses suggest that masculinized faces should be preferred, such faces elicit negative personality attributions (coldness, dominance, and dishonesty, for example). Cross-culturally, personality factors are reported to the most important factor in mate choice by both sexes (Buss, 1989). It seems inconceivable that personality attributions have no effect on attractiveness judgements. Personality attributions, though stereotypic, may predict behavior. For example, ratings of perceived dishonesty from facial appearance correlate with the face owner’s willingness to participate in deceptive behavior (Bond, Berry & Omar, 1994). Less masculine male face shapes may increase in attractiveness because the features usually perceived to be associated with negative personality traits are “softened.”

Individual Differences

The approaches to facial attractiveness briefly described above have revealed considerable agreement on which faces are attractive across cultures and individuals. While cultural values play important roles in physical attractiveness judgements, biological signaling also seems to play a significant part. Nonetheless, individual differences in attractiveness judgements certainly exist, as demonstrated by both the less than perfect agreement within and across experimental studies and common sense. The rest of this chapter examines possible origins of such differences within a “biological” model of physical attractiveness. The first is the possibility that early childhood experience and learning may influence later sexual preferences, in a fashion somewhat analogous to “imprinting” mechanisms common in birds and, in a weaker form, mammals. Secondly, evidence that facial attractiveness judgements may reflect complex, context-dependent decisions based on the affordances offered by a potential partner’s face is reviewed.

Imprinting In Non-Human Animals

Many theorists have proposed that imprinting at an early age influences later behaviors including mating preferences in many species. Imprinting, famously studied by Lorenz, takes place during a specific period of an animal’s life known as a “sensitive period.” The length of this period varies between species and the behavior that is imprinted, but the important concept is that experiences can have more of an effect on an individual’s later behavior when they occur during one period of development than another. Secondly, imprinted behaviors cannot be forgotten. Finally, imprinting can be completed long before the behavior is implemented. Sexual imprinting is a good example of this phenomenon; preferences imprinted in infancy manifest themselves at sexual maturity, long after the end of the “sensitive period.” Below, we briefly review some of the characteristics of imprinting processes in non-human species, before looking for evidence for an analogous process in humans.

The power of imprinting and early life experiences in nonhuman species is clear: Inappropriate imprinting can result in cross-species mating preferences. Kendrick et al. (1998) cross-fostered newborn sheep and goats with nanny or ewe mothers. Male infants, on reaching maturity, strongly preferred to mate with females of their inappropriate maternal foster species as a result of their early life experience. Young cross-fostered female lambs and kids showed similar but weaker patterns of mate preference on reaching maturity, but unlike male offspring, these preferences were reversible. In a variety of rodents, infants cross-fostered with a parent of another species show later preferences for the maternal rather than genetic species (D’Udine & Alleva, 1983).

Studies of imprinting in birds demonstrate that as well as imprinting on maternal species, more subtle influences on adult mate choice can be demonstrated. For example, Vos (1995) exploited the fact that white zebra finches lack any sexual dimorphism except for a difference in bill color–female beaks are orange and male beaks are red (Vos 1995). By painting the bills of some mating pairs with either red or orange nail varnish just before their chicks opened their eyes, Vos generated two groups of test birds; one group raised by “natural” colored parents (i.e. mother’s bill painted orange, father’s red) and another group that had gender switched characteristics (father’s beak orange, mother’s red). Stimuli birds of both sexes with both beak colors were prepared for testing–the results demonstrated that male offspring preferred to mate with a bird with their mother’s bill color, irrespective of the sex of the test bird.

Bateson’s studies of imprinting in Japanese quail add a further complexity to the influence of learning in infancy on later mate choice. In a series of experiments, he demonstrated that quail develop sexual preferences for birds that are similar, but not identical, to those that they are raised with. Bateson (1988) proposed that the proximate mechanisms of such behavior are imprinting and habituation; quail imprint on family characteristics generating preferences for self-similar birds, but then habituate to individual family members, rendering them less arousing than non-familiar birds with similar characteristics. Imprinting provides individuals with experiences that influence later mate choice both positively (to learn appropriate individuals to mate with) and negatively (to avoid inbreeding with individuals who are likely to be very close relatives).

For the purposes of this paper, evidence that early experience can have predictable effects on future sexual preferences in many animals seems sufficient to stimulate a search for similar effects in humans. Some evidence has demonstrated that early life experience has effects on later sexual behavior in humans, reviewed below.

Imprinting In Humans: Westermarck’s Hypothesis

The clearest example of early experience influencing later sexual behavior in humans is as a mechanism to avoid sibling incest as proposed by Westermarck at the end of the 19th century. From his studies in Morocco, the anthropologist argued that children have an innate tendency to develop a sexual aversion to individuals with whom they live closely in infancy and early childhood, i.e., siblings and parents in the vast majority of human families (Westermarck, 1894). Although his theory was greeted favorably at the time, the leading scholars of the day (including Freud, Durkheim & Malinowski) soon rejected Westermarck’s hypothesis (Wolf, 1993), and proposed that a culturally imposed taboo was the major prevention of incest, and one of the defining characteristics of humanity. The “social science orthodoxy” that replaced Westermarck’s theory of incest avoidance was built on erroneous assumptions; critically, that incestuous mating is the “natural” choice throughout the animal kingdom (and would be in humans without the intervention of culture), and that incest taboos are universal in humans (van den Berghe, 1983). Incestuous matings are clearly not the norm in non-human animals, and are avoided by various mechanisms such as dispersal or possibly an imprinted aversion to co-socialized peers as sexual partners (Murray & Smith, 1983, see also van den Berghe, 1983, and Bateson’s work described briefly above). Excellent discussions of the strengths and weaknesses of Westermarckian and other theories of incest avoidance can be found in Wolf, 1993, and McCabe, 1983. Since the fifties, a series of ethnographic studies have formed “natural experiments” that seem to test, and support, Westermarck’s theory. These studies share the common feature that infants and children are placed in intimate contact with other infants in a sibling-like relationship, and yet they are encouraged (or in some cases not prevented) to interact sexually later in life with these co-socialized individuals.

The studies most cited in support of Westermarck have taken place in Kibbutzim in Israel (e.g., Shepher, 1971). The methods of child rearing in Kibbutzim provide an ideal test of the Westermarck hypothesis. Firstly, children are socialized from infancy in a mixed sex peer group (a kitah or kevutza), determined by age. They live separately from their parents in the kitah (although parents spend time with their children in the afternoons), in which group solidarity and cooperation between sabras (comrades) is encouraged. Thus, a situation is generated in which the relationship between sabras has much in common with siblings in other cultures (a lengthy period of intimate contact with other individuals during childhood) but also crucial differences (most importantly, the lack of a culturally imposed incest taboo). Westermarck’s hypothesis predicts that despite this lack of an incest taboo, the innate predisposition against sexual interaction with co-socialized peers will prevent sabras interacting sexually.

Overwhelming evidence appears to support Westermarck: In one detailed study there were no cases of heterosexual activity or marriage between members of the same kitah (Shepher, 1971). Shepher reports that the lack of sexual attraction between peers and the subsequent voluntary avoidance of sexual interaction is actually regretted by those involved. Shepher also summarizes impressive census data in support of Westermarck: Out of 2769 marriages involving individuals raised in kibbutzim “there is not a single case of true intra-peer group marriage to be found” (p. 297).

The second commonly reported body of data in support of the Westermarckian thesis is reported in Wolfs studies of sim-pua marriages in Taiwan (Wolf, 1993). Until the mid-1940s women in much of China and Taiwan gave away their infant daughters, and instead raised other women’s daughters as future wives for their sons. The future husband and wife are raised together, sharing sleeping mats in the same fashion as siblings. Westermarck would predict that, at sexual maturity, the husband and wife should have developed an aversion to consummating the marriage due to their intimate childhood association. An excellent natural control group is available to compare the relative success of these forced unions to marriages without co-socialization of partners; the alternative to sim-pua marriage was to marry a mature son to a post-pubescent girl from another family in what was referred to as “major marriage.” Such couples had often not met until the marriage ceremony, thus precluding any Westermarckian aversion to sexual activity.

Wolf reports that childhood associates are extremely reluctant to consummate their legal bond–the marriage ceremony is itself referred to as “pushing them together”. Husbands in sim-pua marriages were three times as likely to visit “dark rooms” (brothels) as men married in the major fashion, an observation that Wolf interpreted as a lack of satisfaction with the sexual aspect of their marriages. Later census studies of more than 14,000 marriages indicate that 2.5 sim-pua marriages end in divorce for every “major” marriage that does so, and the fertility of major marriages is 25% higher than sim-pua marriages (Wolf 1993).

The differences in fertility between marriage types are unlikely to be caused by “adoption trauma” or maltreatment of the adoptee. Girls raised for a sim-pua marriage that did not occur (often due to the death of the prospective husband), but instead married other men, had the same fertility and divorce rates in their marriages as women in major marriages. In addition, Wolf ruled out links between socioeconomic status and the low fertility of sim-pua marriages (poorer families tended to prefer such marriages), by considering the taxable value of the husband’s estate in his analysis.

Wolf also claims to have discovered a “critical period” in which the aversion between the individuals in sim-pua marriages develops. He notes that the age at which the bride moves into the husband’s home is crucial; “Brides adopted before age three display sharply reduced fertility and a markedly higher probability of divorce; those adopted after age three do not.” (Wolf, 1993). When he reanalyzed the data considering only minor marriages in which the brides moved into the husbands’ residences before three years of age, he found that the relative minor/major marriage divorce rate increased to 3:1, and that major marriages were 45% more fertile.

The Chinese data complement the findings of the kibbutz studies. Firstly, kibbutz children are never forced to marry peers–the aversion they feel to each other may be depthless, and easily overcome once sex has been made a part of the previously platonic relationship. In the Chinese case the childhood associates are compelled to marry by extremely strong social pressures. Studies of marriage patterns in Israel, Taiwan and elsewhere (e.g., the Lebanon, McCabe, 1983) seems to indicate that childhood association causes later sexual disinterest between peers. Interestingly, theorists seem to propose a “critical” or “sensitive” period of life during which co-socialization must occur to develop these aversive sexual feelings towards certain individuals later in life, analogous to imprinting processes in non-human animals (e.g., Shepher, 1971). Is there, however, a tendency for people to become attracted to individuals similar to those that they are raised with that could be the result of positive imprinting in humans?

Human Imprinting on Familial Characteristics

Evidence for an imprinting type process of learning leading to preferences for familial characteristics in human partners is limited. As family members tend to resemble one another, adult sexual preferences for mates who are similar to family members may lead to some similarity in physical appearance between partners. Large-scale studies of anthropometric characteristics indicate that married couples tend to be physically similar to each other, but the degree of similarity is small (see Roberts, 1977 for review). Nonetheless, imprinting or analogue process is a potential source of individual differences in attractiveness judgements.

Some studies have directly studied the perceived similarity of members of real married couples, usually by presenting subjects with a selection of images and asking them to match up pairs. Griffiths & Kunz, (1973), took photos of 30 actual married couples. Each member of the couple was photographed individually against the same background. The couples were split into six groups by the length of the marriage. The test involved presenting ten photos of subjects’ faces to each group. The members of the group were then asked to identify the couples. The results of the study are intriguing: Although couples married for less than ten years were matched at levels above chance, subjects failed to successfully match couples married for between ten and twenty years. Curiously, couples married for more than 20 years were once again successfully identified by the subjects. This is obviously a difficult result to interpret, although the small stimuli sets used (just five couples in each group) may go some way to explaining the curious patterns in the data.

Zajonc et al. (1987) performed a similar experiment investigating whether couples grow physically alike over time due to sharing similar diets, lifestyles and emotional experiences. To test this theory of environmental co-existence, Zajonc et al. obtained two photographs from each individual in twelve married couples. One of these photos came from the first year of the marriage, and the second was taken in the twenty-fifth year of the partnership. The photos were cropped and masked to reveal only the head and face of each person, and then split into “young” and “old” subsets. None of the subjects were exposed to the young and old versions of the same face. Zajonc et al.’s results indicated that for the “young” subset subjects did not perceive a level of similarity (or a likelihood of marriage) between the target and partner any greater than would be expected by chance. Conversely, real couples from the “old” subset were ranked as more similar and more likely to be married than chance would predict–a result that supported the environmental co-existence hypothesis. These results are somewhat contrary to the Griffiths & Kunz (1972) study. Although both studies indicate some homogeneity between couples married for twenty-five years or more, the Zajonc et al. study does not demonstrate any similarity between newly wed couples. An interesting observation that Zajonc does not stress is the lack of any difference between the “similarity” and “likelihood of marriage” rankings. This indicates that Zajonc’s subjects associated facial resemblance with the likelihood of marriage between individuals; an example of the common idea that people marry people who look like themselves.

Hinsz (1989) conducted a third study of facial similarity within real life partnerships. The stimuli in this experiment were photographs of the individuals in 30 engaged couples and 30 couples who had been married for around 25 years. The subjects were presented with opposite sex pairs of photos, and then asked to rate the similarity between the two faces. Half of the pairs presented were actual couples, and the other half were randomly generated couples. A significant real vs. random couple effect was discovered; real couples were rated as more similar than randomly generated couples. Unlike the Zajonc et al. study, couples that had been together for longer periods of time were not perceived as more similar than new couples in the Hinsz experiment.

Penton-Voak et al. (1999a) adopted a different approach in order to study similarities between partners, by studying preferences for faces, rather than looking at real couples. Individuals were photographed and the resulting image was “gender transformed” using computer graphics techniques to generate a hypothetical opposite sex “sibling.” Ratings demonstrated that these computer-generated images were perceptually similar to the original photograph, indicating that the transforms were successful and credible. When the original subjects were asked to rate a set of photographs including their opposite sex “twin,” they tended to rate faces similar to their own higher in attractiveness than those to which they were dissimilar. Unfortunately, whether this really represents a preference for self-similar faces is unclear as a preference for averageness could also generate this finding: Faces very far from average receive low attractiveness ratings and such atypical faces differ from the faces of most individuals more than average faces, possibly accounting for the result.

Collectively, the findings of these partner similarity studies could be attributed to the development of weak preferences for family-like (i.e., self-similar) facial characteristics in life, although this is by no means the only possibility. Indeed an imprinting based explanation fails to explain the idiosyncrasies of the data generated, such as Zajonc’s findings of a reported increase in partner similarity with age. Nevertheless, two other studies indicate that parental characteristics, and specifically opposite sex parental characteristics, could influence later choice of partner.

The first of these studies is an analysis of the tenth Italian census data (Zei, Astofli & Jayakar, 1981), which investigates the relationship between father’s age and husband’s age for a sample of over 350,000 women. Small but consistently positive correlations between these variables indicate that the daughters of older men subsequently tend to choose older husbands. Separate analyses were performed for two separate time periods (a fifteen-year prewar and fifteen-year post-war period) and for different locations around Italy. The age of the women at marriage and their educational level was also taken into statistical consideration. Correlations in all analyses were positive, but were found to be stronger in rural and less economically developed areas.

The overall positive correlation between father’s and husband’s age is consistent with the possibility of an imprinting type process; daughters may imprint on the visually salient characteristics of their father (including their age) as children and later find these characteristics preferable in their own partner.

A second study provides further evidence for a correlation between father’s age and partner’s age. Wilson & Barrett (1987) obtained responses to a questionnaire from 314 young British women, giving data on the ages and the eye color of the women, their partners and their parents. The age data showed a correlation of borderline significance between father’s age at the time of the daughter’s birth and the daughter’s partner’s age; a result consistent with Zei et al.’s findings. The eye color data collected showed a weak tendency for women to choose partners whose eye color resembles their fathers.

Although both the Italian census data and Wilson & Barrett’s work are encouraging to those looking for evidence of sexual imprinting in humans, it should be noted that the significant correlations and tests found in both studies have very small effect sizes, and account for very little of the variance in the data (the correlations between father’s and husband’s age in Zei et al.’s study are between 0.05 and 0.08; in Wilson’s study the same correlation is 0.11). The difficulties in detecting the influence of any of the multitude of factors that contribute to mate choice is clear–very large samples are needed to achieve statistical credibility when any one factor is considered.

Cyclic Preferences for Male Faces and Mixed Female Mating Strategies

Recent research into male facial attractiveness has revealed that female preferences for male faces may vary over time, and with personal circumstances. Three recent experiments (reported in Penton-Voak et al., 1999b, and Penton-Voak & Perrett, in press) demonstrate evidence that preference for male face shape changes across the menstrual cycle. Women’s preferences for feminine male face shapes (which appear contrary to most examples of sexual selection in non-human animals, as detailed above) are attenuated in the follicular phase of the menstrual cycle: When conception is most likely women prefer relatively masculine faces. Menstrual phase may be one variable that contributes to the difficulty of defining what females find attractive in male faces.

Furthermore, there were trends that women with and without partners had differing preferences: Women with partners preferred marginally more masculine faces overall, and tended to undergo a larger shift toward masculinity when their chance of conception was high (Penton-Voak et al., 1999b). The type of relationship a woman sought from a partner also influenced preferences. Women asked to judge male faces for a “long-term” partner showed no change in preference across the menstrual cycle, whereas women asked by the experimenter to judge the faces for a “short-term sexual partner” did. Women using oral contraception showed no cyclic shifts.

These results are consistent with earlier research into changing preferences across the menstrual cycle. For example, Frost (1994) reports that around ovulation, women prefer relatively darker male skin color. As skin color is sexually dimorphic within all cultures (with males having relatively darker skin than females when factors such as exposure to sunlight are eliminated), this result parallels the preferences for masculine face shape reported above. Thornhill & Gangestad report that female preferences for male odor change across the menstrual cycle. Women find the scent of symmetrical men more attractive than the odor of less symmetrical men during the follicular phase of their cycle, but not at other times (Thornhill & Gangestad, 1999b). There is a positive relationship between trait size and symmetry in many species including, perhaps, humans (Scheib et al., 1999), so again this can be considered as a cyclic shift favoring masculinity when conception is likely.

Both masculinity and femininity may be associated with behaviors that have costs and benefits to reproductive success (Perrett et al., 1998; Berry & Wero, 1983). Standard models of sexual selection predict preferences for males with exaggerated sexual traits (i.e., masculine faces), which may signal potential benefits for off-spring in terms of heritable fitness, but in humans imply costs due to potentially decreased paternal investment. For example, Gangestad & Thornhill (1997) show that symmetrical men may be less likely to invest in a relationship, and have more extra-pair sex. Femininity in male faces may be associated with the opposite collection of characteristics, such as warmth, honesty and a willingness to invest in offspring (Perrett et al., 1998).

In humans, paternity uncertainty results from internal fertilization, concealed ovulation, limited visual similarity between off-spring and their fathers, and the apparently cross-cultural finding that couples prefer to copulate clandestinely (Christenfeld & Hill, 1995; Pagel, 1997; Schroder, 1993). Rates of extra-pair paternity are certainly non-zero cross-culturally, although precise levels are unknown due to a lack of well-controlled studies (Macintyre & Sooman, 1991). What little is known about patterns of sexual behavior across the menstrual cycle indicates that sex outside of a pair bond is more likely to occur mid-cycle (Baker & Bellis, 1995). Females may subconsciously balance the costs and benefits of various reproductive strategies in the context of life history factors, preferring relatively feminine looking males when looking for long-term investment in offspring, or more masculine men when heritable benefits are judged to be more important. Alternatively, some women may adopt a mixed reproductive strategy and cuckold a primary partner whose relatively feminine appearance suggests cooperation in parental care by occasionally seeking extra-pair copulations with “masculine” males when conception is most likely, and hence benefit without cost. Similar patterns of sexual activity are found in non-human species (Graves et al., 1993; Andersson, 1994). Individual differences in the faces that females find attractive may reflect strategies to improve reproductive success that vary according to circumstance.

Condition Dependent Preferences

Pathogens, parasites and other stressors encountered by an organism have a negative effect on the condition of an individual. Condition has long been known to reflect an individual’s attractiveness to potential mates in may species (Andersson, 1994). More recent work has indicated that condition may influence an individual’s preferences as well as their perceived attractiveness. For example, the red coloration of male sticklebacks reduces in intensity with parasite load and female sticklebacks demonstrate a preference for intense male coloration. Females in poor condition, however, show an unexpected preference for less intensely colored (i.e., poorer condition) males, and this is not a result of competition (Bakker et al., 1999).

Recent work has shown somewhat similar processes may occur in humans. In one experiment, women who self-rated themselves as attractive preferred relatively more masculine facial composites than those women who rated themselves as low in attractiveness. In addition, the self-rated attractive women showed a preference for symmetric male face shapes that was not shown by women who rated themselves lower in attractiveness (Little et al., in review). As both facial masculinity (size of sexually dimorphic traits) and symmetry are thought to be indicators of condition in men, and hence are possibly cues to heritable genetic quality, these findings may represent a woman’s adaptive response to an assessment of her own “condition.”

One speculative interpretation of these preferences is that women who are of “high mate value” (i.e., whose phenotype displays cues to health and fecundity) may be more likely to attract and retain a “high mate value” male (who possesses cues to heritable quality such as masculinity and symmetry, but also may be less inclined to invest in parental care) than a lower mate value female. What is especially interesting about this finding is that the actual preferences for male faces are changed as a function of female self-rated attractiveness. So, the preferences of women who rate themselves as low in attractiveness do not represent conscious choices in response to past failure in competition for males (I find high quality male X attractive, but experience tells me that I will not succeed in competition for him, so I will settle for lower quality Y), but actual changes in criteria (I find Y more attractive than X). Preferences contingent on the chooser’s own condition may be adaptive in species where mate choice is costly and time consuming–as in humans.


Attractive faces appear to have some stable qualities that generalize across cultures and are salient to individuals very early in infancy, indicating that normative social information alone is unlikely to be responsible for attractiveness judgements. Averageness, symmetry and size of secondary sexual traits are proposed to be candidate features of facial attractiveness, as they may honestly indicate viability. People nonetheless differ in the faces that they find attractive. Tentatively, we have suggested here that at least some of this behavioral variation may reflect biologically adaptive processes similar to those documented in other species. Changes within individuals over time may reflect short-term changes in hormonal state (as demonstrated by cyclic changes in face preferences in females) or changes in circumstance such as the presence or absence of a partner. Imprinting on family characteristics may also contribute to differences in attractiveness judgements between individuals, and recent work indicates that women may alter their preferences for a partner in response to an assessment of their own attractiveness–both “condition” and imprinting are know to influence mate choice in some non-human species. Future research should investigate such factors to see if they affect facial attractiveness judgements in a predictable way.


Alley, T. R. and Cunningham, M. R. “Averaged Faces are Attractive, But Very Attractive Faces are Not Average.” Psychological Science 2:2 (1991):123-125.

Andersson, M. Sexual Selection. Princeton, NJ: Princeton University Press, 1994.

Baker, R. R., and Bellis, M. A. Human Sperm Competition: Copulation, Masturbation And Infidelity. (1 ed.). London: Chapman and Hall, 1995.

Bakker, T. C., Kunzler, R. and Mazzi, K. “Condition-Related Mate Choice In Sticklebacks.” Nature 401 (1999): 234.

Bateson, P. “Preferences For Close Relations In Japanese Quail.” Acta XIX Congressus Internationalis Ornithologici Vol. 1, pp. 961-972. Ed. H. Ouellet. Ottawa: University of Ottawa Press, 1988.

Berry, D. S., and Wero, J. L. F. “Accuracy In Face Perception: A View From Ecological Psychology.” Journal of Personality 61:4 (1993): 497-519.

Bond, C. F., Berry, D. S., and Omar, A. “The Kernel Of Truth In Judgements Of Deceptiveness.” Basic and Applied Social Psychology 15, (1994): 523-534.

Buss, D. M. “Sex Differences In Human Mate Preferences: Evolutionary Hypotheses Tested In 37 Cultures.” Behavioral and Brain Sciences 12, (1989): 1-49.

Christenfeld, N., and Hill, E. A. “Whose Baby Are You?” Nature 378 (1995): 669.

Cunningham, M. R., Barbee, A. P., and Pike, C. L. “What Do Women Want? Facialmetric Assessment Of Multiple Motives In The Perception Of Male Facial Physical Attractiveness.” Journal of Personality and Social Psychology 59:1 (1990): 61-72.

Cunningham, M. R., Roberts, A. R., Barbee, A. P., and Druen, P. B. “`Their Ideas Of Beauty Are, On The Whole, The Same As Ours’: Consistency And Variability In The Cross-Cultural Perception Of Female Attractiveness.” Journal of Personality and Social Psychology 68:2 (1995): 261-279.

Darwin, C. The Descent Of Man, And Selection In Relation To Sex. London: John Murray, 1871.

D’Udine, B., and Alleva, E. “Early Experience and Sexual Preferences in Rodents.” Mate Choice. Ed. P. Bateson. Cambridge: Cambridge University Press, 1983.

Folstad, I., and Karter, A. J. “Parasites, Bright Males and the Immunocompetence Handicap.” American Naturalist 139 (1992): 603-622.

Frost, P. “Preference For Darker Faces in Photographs at Different Phases of the Menstrual Cycle: Preliminary Assessment of Evidence for a Hormonal Relationship.” Perceptual and Motor Skills 79 (1994): 507-514.

Gangestad, S. W., and Thornhill, R.. “The Evolutionary Psychology Of Extrapair Sex: The Role Of Fluctuating Asymmetry.” Evolution and Human Behaviour 18:2 (1997): 69-88.

Grammer, K., and Thornhill, R. “Human (Homo Sapiens) Facial Attractiveness and Sexual Selection: the Role of Symmetry and Averageness” Journal of Comparative Psychology 108 (1994): 233-242.

Graves, J., Ortega-Ruano, J., and Slater, P. J. B. “Extra-Pair Copulations and Paternity and Shags: Do Females Choose Better Males?” Proceedings of the Royal Society of London, B 253 (1993): 3-7.

Griffiths, R. W., and Kunz, P. R.. “Assortative Mating: A Study Of Physiognomic Homogamy.” Social Biology 20:4 (1973): 448-453.

Hinsz, V. B. “Facial Resemblance in Engaged and Married Couples.” Journal of Social and Personal Relationships 6 (1989): 223-229.

Johnston, V. S., and Franklin, M. “Is Beauty In The Eye Of The Beholder?” Ethology and Sociobiology 14 (1993): 183-199.

Keating, C. F. “Gender and The Physiognomy of Dominance and Attractiveness.” Social Psychology Quarterly 48 (1985): 61-70.

Kendrick, K. M., Hinton, M. R., and Atkins, K. “Mothers Determine Male Sexual Preferences.” Nature 395 (1998): 229-230.

Koeslag, J. H. “Koinophilia Replaces Random Mating in Populations Subject to Mutations with Randomly Varying Fitnesses.” Journal Theoretical Biology 171 (1994): 341-345.

Langlois, J. H., Ritter, J. M., Roggman, L. A., and Vaughn, L. S. “Facial Diversity And Infant Preferences For Attractive Faces.” Developmental Psychology 27:1 (1991): 79-84.

Langlois, J. H., and Roggman, L. A. “Attractive Faces Are Only Average.” Psychological Science 1:1 (1990): 115-121.

Langlois, J. H., Roggman, L. A., Casey, R. J., Ritter, J. M., Riser-Danner, L. A., and Jenkins, V. Y. “Infant Preferences For Attractive Faces: Rudiments Of A Stereotype?” Developmental Psychology 23(3) [23:3], (1987): 363-369.

Macintyre, S., and Sooman, A. “Non-Paternity And Prenatal Genetic Screening.” The Lancet 338 (1991): 869.

Mazur, A., Halpern, C., and Udry, J. R. “Dominant Looking Males Copulate Earlier.” Ethology and Sociobiology 15 (1994): 87-94.

McArthur, L. Z., and Berry, D. S. “Cross-Cultural Agreement in Perceptions of Babyfaced Adults.” Journal of Cross-cultural Psychology 18:2 (1987): 165-192.

McCabe, J. “FDB Marriage: Further Support For The Westermarck Hypothesis of the Incest Taboo?” American Anthropologist 85 (1983): 50-69.

Moller, A. P. “Developmental Stability And Fitness: A Review.” American Naturalist 149, (1997): 916-942.

Moller, A. P., and Thornhill, R. “Bilateral Symmetry and Sexual Selection: A Meta-Analysis.” American Naturalist 151, (1998): 174-192.

Mueller, U., and Mazur, A. “Facial Dominance In Homo Sapiens As Honest Signaling Of Male Quality.” Behavioral Ecology 8(5) [8:5] (1997): 569-579.

Murray, R. D. and Smith, E. O. “The Role of Dominance and Intrafamilial Bonding in the Avoidance of Close Inbreeding.” Journal Human Evolution 12 (1983): 481-486.

Owens, I. P. F., and Short, R.V. “Hormonal Basis Of Sexual Dimorphism In Birds: Implications For New Theories Of Sexual Selection.” Trends in Ecology and Evolution 10:1 (1995): 44-47.

Pagel, M. “Desperately Concealing Father: A Theory Of Parent-Infant Resemblance.” Animal Behavior 53 (1997): 973-981.

Penton-Voak, I., Perrett, D., and Pierce, J. “Computer Graphic Studies of the Role of Facial Similarity In Attractiveness Judgements.” Current Psychology 18 (1999a): 104-117.

Penton-Voak, I. S., Perrett, D., Castles, D., Burt, M., Koyabashi, T., and Murray, L. K. “Female Preferences For Male Faces Change Cyclically.” Nature 399 (1999b): 741-742.

Penton-Voak, I. S., and Perrett, D.I. (in press). “Cyclic Female Preferences For Male Faces–Further Evidence.” Evolution and Human Behavior.

Perrett, D. I., May, K.A., and Yoshikawa, S. “Facial Shape And Judgements Of Female Attractiveness.” Nature 368 (1994): 239-242.

Perrett, D. I., Burt, D. M., Penton-Voak, I. S., Lee, K. J., Rowland, D. A., and R., E. “Symmetry and Human Facial Attractiveness.” Evolution and Human Behavior 20 (1999): 295-307.

Perrett, D. I., Lee, K. J., Penton-Voak, I. S., Rowland, D. R., Yoshikawa, S., Burt, D. M., Henzi, S. P., Castles, D. L., and Akamatsu, S. “Effects Of Sexual Dimorphism On Facial Attractiveness.” Nature 394 (1998): 884-887.

Pollard, J., Shepherd, J., and Shepherd, J. “Average Faces Are Average Faces.” Current Psychology 18:1 (1999): 98-103.

Rhodes, G., Sumich, A., and Byatt, G. “Are Average Facial Configurations Attractive Only Because Of Their Symmetry?” Psychological Science 10 (1999): 52-58.

Roberts, D. F. “Assortative Mating in Man: Husband/Wife Correlations in Physical Characteristics.” Supplement to the Bulletin of the Eugenics Society 2 (1977).

Scheib, J. E., Gangestad, S. W., and Thornhill, R. “Facial Attractiveness, Symmetry, and Cues to Good Genes.” Proceedings of the Royal Society of London B (1990.

Schroder, I. “Concealed Ovulation And Clandestine Copulation: A Female Contribution To Human Evolution.” Ethology and Sociobiology 14 (1993): 381-389.

Shepher, J.. “Mate Selection Among Second Generation Kibbutz Adolescents and Adults: Incest Avoidance and Negative Imprinting.” Archives of Sexual Behavior 1:4 (1971): 293-307.

Slater, A., Von der Schulenberg, C., Brown E., Badenoch M., Butterworth G., Parsons S., and Samuels C. “Newborn Babies Prefer Attractive Faces.” Infant Behavior and Development 21 (2) [21:2?] (1998): 345-354.

Symons, D. The Evolution Of Human Sexuality. Oxford: Oxford University Press, 1979.

Symons, D. “Beauty Is In The Adaptations Of The Beholder: The Evolutionary Psychology Of Human Female Sexual Attractiveness.” Sexual Nature/Sexual Culture. Eds. P. Abramson and S. Pinkerton. Chicago: University of Chicago Press, 1994.

Thornhill, R., and Gangestad, S. W. “Facial Attractiveness.” Trends in Cognitive Science 3 (1999a): 452-460.

Thornhill, R., and Gangestad, S. W. “The Scent Of Symmetry: A Human Sex Pheromone That Signals Fitness?” Evolution and Human Behavior (1999b).

Van Den Berghe, P. L. “Human Inbreeding Avoidance: Culture in Nature.” Behavioral and Brain Sciences 6 (1983): 91-123.

Vos, D. R. “Sexual Imprinting In Zebra-Finch Females: Do Females Develop A Preference For Males That Look Like Their Father?” Ethology 99 (1995): 252-262.

Walster, E., Aronson, V. Abrahams, D. and Rottman, L. “Importance Of Physical Attractiveness In Dating Behavior.” Journal of Personality and Social Psychology 4:5 (1966): 508-516.

Westermarck, E. The History Of Human Marriage. London: Macmillan, 1894.

Wilson, G. D., and Barrett, P.T. “Parental Characteristics and Partner Choice: Some Evidence for Oedipal Imprinting.” Journal of Biosocial Sciences 19 (1987): 157-161.

Wolf, A. P. “Westermarck Redivivus.” Annual Review Anthropology 22 (1993): 157-175.

Zajonc, R. B., Adelmann, P.K., Murphy, S.T. and Niendenthal, P.M. “Convergence in the Physical Appearance of Spouses.” Motivation and Emotion 11:4 (1987): 335-346.

Zei, G., Astofli, P., and Jayakar, S. D. “Correlation Between Father’s Age And Husband’s Age: A Case Of Imprinting?” Journal of Biosocial Science 13 (1981): 409-418.

Ian S. Penton-Voak is a lecturer in Psychology at the University of St. Andrews. He is principal author of “Female Preferences for Male Faces Change Cyclically” (Nature, 1999).

David I. Perrett is Professor of Psychology at the University of St. Andrews. He is principal author of “Effects of Sexual Dimorphism on Facial Attractiveness” (Nature, 1998).

COPYRIGHT 2000 New School for Social Research

COPYRIGHT 2001 Gale Group