Male facial attractiveness: Perceived personality and shifting female preferences for male traits across the menstrual cycle

ADVANCES IN THE STUDY OF BEHAVIOR, VOL. 30

Male Facial Attractiveness: Perceived Personality and Shifting Female Preferences for Male Traits across the Menstrual Cycle I A N S. P E N T O N - V o A K

AND DAVID I. PERRETT

SCHOOL OF PSYCHOLOGY U N I V E R S I T Y O F ST. A N D R E W S ST. A N D R E W S , FIFE, KY16 9JU UNITED KINGDOM

I. INTRODUCTION A review of the facial attractiveness literature describes the field as "among the success stories in cognitive science" (Thornhill and Gangestad, 1999). Certainly, studies of female attractiveness have achieved remarkable consensus as to the facial features that are preferred by males, and the biological signals that such characteristics may convey (e.g., Johnston and Franklin, 1993; Perrett et al., 1994, 1998; Jones, 1995). Yet studies of male attractiveness, often using techniques identical to those used in studies of female faces, have failed to isolate features that are consistently found attractive. Cross-cultural studies of mate preferences indicate that men are more influenced by physical appearance than women in mate choice (Buss, 1989). Observation suggests that women, rather than men, are the sex that advertise through extensive use of makeup and clothing: "Faced with these facts, a biologist would be forced to suspect that he was looking at a society in which females compete for males, rather than vice versa"(Dawkins, 1976). To those familiar with sexual selection in nonhuman species this seems to present a paradox, as H o m o sapiens is moderately sexually dimorphic in body size, males mature later than females, and testis size tentatively suggests a unimale mating system (Harcourt, 1996). These sex differences indicate male competition for females, not vice versa. Indeed, the majority of documented human societies are best described as facultatively polygynous (Murdock, 1967; Daly and Wilson, 1983). In an attempt to resolve this paradox, many evolutionary theorists have been quick to point out that in a social species with biparental care such 219

Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0065-3454/01$35.00

220

IAN S. PENTON-VOAK AND DAVID I. PERRETT

as ours, both males and females should be choosy. Female reproductive potential is more easily assessed from physical appearance than male reproductive potential due to declining female fertility with age. Male ability to invest in children may even increase with age as status gains allow greater access to resources (e.g., Symons, 1979). Women, however, as well as men, are concerned with good looks in a mate. Although 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 young adults' 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 by both sexes, and the centrality of mate choice theories to evolutionary approaches to behavior, men's facial characteristics may reflect the action of sexual selection. The purpose of this review is to examine both theories and studies of male facial attractiveness (Section II) and studies of personality attributions made to face shapes that may potentially influence female judgments of male attractiveness (Section III). Finally, we outline some of our recent work investigating how shifting female preferences for male faces across the menstrual cycle may balance the costs and benefits associated with different male traits (Section IV). We hope that this summary may at least partially unite some of the conflicting findings in the current literature.

II.

THEORIES OF H U M A N FACIAL ATrRACTIVENESS

Despite increasing professional and public interest in the field, systematic studies of facial attractiveness have not yet reached consensus on the characteristics that make human male faces attractive to females. The three main hypotheses are that (1) symmetrical faces, (2) average faces, or (3) faces with exaggerated secondary sexual characteristics are found attractive. Each of these hypotheses (which are not mutually exclusive) can be justified by "good genes" theories, and each has received some empirical support, briefly reviewed here. All three of these theories rely on facial characteristics honestly advertising biological quality: symmetry is related to developmental stability, which may be linked to genotypic quality; averageness may indicate heterozygosity; and exaggerated sex-typical traits may represent an "honest handicap." All, then, are thought to be phenotypic signals linked to heritable genotypic quality and reproductive potential, and, hence, preferences for such characteristics should be favored by selection. Empirical proof that facial attractiveness is associated with characteristics that advertise health is, however, conflicting in contemporary populations (Kalick et al., 1998; Shackleford and Larsen, 1999; see also Daly and Wilson, 1999).

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

A.

221

FACIAL SYMMETRY

Symmetry is a marker of developmental stability that may reflect genotypic as well as phenotypic quality (M¢ller, 1997; M¢ller and Thornhill, 1997, for reviews). Bodily symmetry appears to be related to reproductive success in many species, including humans (Gangestad and Thornhill, 1997; Manning et al., 1997; M¢ller and Thornhill, 1998). Meta-analysis suggests that symmetry is more important to male attractiveness than to female attractiveness across species (M¢ller and Thornhill, 1998). A simple prediction is that facial symmetry will be important in judgments of attractiveness. Studies addressing this question are reviewed in the following two sections.

1. Studies o f Naturally Occurring Facial Asymmetries There is no standardized technique for measuring the absolute symmetry of a human face, which may be a factor contributing to the relatively small amount of research that has been conducted into the symmetry of natural faces. Grammer and Thornhill (1994) estimated asymmetry in male and female faces by bisecting lines that delineated left and right positions of bilateral features, marked on each face in a small sample of 16 men and 16 women. Horizontal symmetry measured in this way correlated with attractiveness judgments of both male and female faces. Using a similar measurement technique, but with the inclusion of a measure of vertical as well as horizontal symmetry, Scheib et al. (1999) also found that symmetry and rated attractiveness correlated for a sample of 40 male faces. The relationship between symmetry and facial attractiveness, however, was still observed when only the left or right half of each face was presented (i.e., without cues to symmetry), suggesting that some covariate of symmetry that can be ascertained from half-faces may influence attractiveness judgments. Scheib et al. found that size of the lower face and cheekbone prominence were related to symmetry in both half- and full-face attractiveness judgments. It should be noted, however, that some cues to symmetry may still be present in half-faces. The midline used to generate the half-face stimuli bisected the middle of the nose tip and the "v" of the upper lip. In faces with low symmetry, other points on this midline may reveal more (or less) than half of centrally placed features (e.g., bottom of chin, bridge of nose). Hence, some cues to symmetry may still remain in half-faces. Mealey etal. (1999) studied the covariation of symmetry and attractiveness in monozygotic twin pairs. Such twins are genetically but not developmentally identical, and, hence, manifest differing levels of facial symmetry when adult. The symmetry of each twin pair was assessed by presenting left-left (L-L) and right-right (R-R) chimeric image pairs of each of the two twins to adult participants, and asking them to rate the similarity of each pair of images (Twin i L1-L1 and R1-R1; Twin 2 Lz-L2 and Rz-R2). The most similar

222

IANS.PENTON-VOAKANDDAVIDI. PERRETT

pair of images (i.e., L1-L1 and R1-R2 or Lz-L2 and R2-R2) was assumed to belong to the more symmetric of the two twins. Secondly, the natural images of the twins were presented to separate raters, who were asked to rate the more attractive of each twin pair using a rating task. The results indicated a significant correlation between symmetry and attractiveness for both male and female twins--the more symmetrical of each twin pair was generally preferred, and bigger differences in estimated symmetry led to larger attractiveness differentials within twin pairs. 2.

Studies Using Computer Graphic Symmetry Manipulations

Given the theoretical benefits ascribed to symmetry (e.g., M¢ller and Thornhill, 1998), and the results of studies of naturally occurring asymmetry in faces, it is surprising that several studies directly manipulating human facial images have found that asymmetry is generally preferred to symmetry (Kowner, 1996; Langlois et al., 1994; Samuels et al., 1994; Swaddle and Cuthill, 1995). Most of these studies have created symmetric face images by aligning one half-face with its mirror reflection (Kowner 1996; Langlois et al., 1994; Samuels et a1.,1994). These techniques may induce additional stimulus differences unrelated to symmetry. The mirror-reflecting technique can introduce abnormal feature shapes. For example, a mouth of normal width displaced to the right of the midline will assume atypical widths in left-mirrored and right-mirrored chimeric face images (Fig. 1, C,D). Asymmetry may have been preferred in a further study (Swaddle and Cuthill, 1995) because asymmetric original faces were compared to symmetric images with different skin textures (induced by combining original and mirror images); see Perrett et al. (1999) for a discussion of these problems, and Fig. 1. Despite the fact that experiments using chimeric stimuli have failed to detect a preference for symmetry, several studies have demonstrated that symmetry predisposes favorable facial attractiveness judgments, reconciling the results of computer graphic studies with studies of naturally occurring asymmetry. Perrett et al. (1999) present three experiments using novel techniques for manipulating symmetry: these experiments are described in some detail next.

Fr~. 1. Symmetrymanipulationfor facialimageswithnatural skintextures.Real faceimages with normal and symmetricshapes. (A, B) Note that asymmetriesin pigmentationand shadows present in the originalfaces (A) remain in the more symmetricallyshaped versions (B). (C, D) Imagesmade with techniquesemployedin previousstudiesof facialsymmetry.Chimeric faces made by combining the left sides of the original faces with their mirror reflections (C) and similarlyfor the right sides of the faces (D) illustrate the shape abnormalities that this techniqueinduces.From Perrett et al. (1999).

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

A

B

C

D

223

224

I A N S. P E N T O N - V O A K A N D D A V I D I. P E R R E T T

In Experiment 1, symmetry in face shape was improved without changing the symmetry of face textures; natural asymmetries in skin pigmentation were present in both the original and more symmetric versions of the same face (Fig. 1). Adults' responses to pair-wise presentation of these two versions of each face indicated a clear preference for the symmetrically remapped stimuli. Experiment 2 used stimuli with average texture information generated from a set of faces. This average texture was rendered into both the original shapes and the symmetrically remapped shapes of a set of individual faces, giving perfect symmetry in the remapped version. Pair-wise presentation showed a preference for perfectly symmetrical face stimuli. Experiment 3 used a rating task rather than a forced choice paradigm (stimuli were presented one at a time rather than in pairs), replicating the preference for symmetry shown in Experiments i and 2. In Experiment 1, symmetry manipulations were rather subtle; perhaps because of this subtlety of the manipulation, 75 % of subjects were unaware that symmetry had been manipulated. Despite this unawareness, subjects preferred symmetrical faces, demonstrating that symmetry influences attractiveness judgments even without conscious awareness. Experiment 1 goes some way toward establishing the ecological validity of computer graphic research into facial symmetry because it demonstrates a preference for more symmetrical male and female facial images with entirely natural skin pigmentation and texture. Rhodes et al. (1998) used a technique similar to that of Swaddle and Cuthill (1995), but avoided stimulus artifacts such as double blemishes by retouching the manipulated images. Using this improved technique, they too found a preference for symmetry over asymmetry in faces of both sexes. Further support for a preference for symmetry in human faces comes from Hume and Montgomerie (1999) who replicated Rhodes et al.'s technique with similar results. Thus, the three most recent, and perhaps methodologically superior, computer graphic studies (Rhodes et al., 1998; Perrett et al., 1998; Hume and Montgomerie, 1999) parallel the findings of investigations into naturally occurring facial asymmetries (Grammer and Thornhill, 1994; Mealey et al., 1999; Scheib et al., 1999). There are progressive changes in the shape of faces, particularly in soft tissues, throughout life (e.g., Burt and Perrett, 1995). Asymmetries in face shape due to differential growth and aging will, therefore, be more prominent in older faces. Kowner (1996) demonstrated that asymmetry itself may be a perceptual cue to age because asymmetric faces were perceived as older than symmetric faces. To summarize, it is apparent that symmetry is one factor contributing to judgments of facial attractiveness. It is also apparent, however, that symmetrical faces may possess other properties that are associated with attractiveness.

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

225

For example, sex hormones may influence the symmetry of growth (Thornhill and Gangestad, 1993) and chin shape and size, which independently affect attractiveness (Perrett et al., 1994; Scheib et al., 1999). A covariate of symmetry, rather than symmetry itself, may be a primary cue to attractiveness judgments (Scheib et al., 1999). B.

AVERAGENESS

The most influential and most investigated theory of facial attractiveness is that average features, not exaggerated characteristics, are optimal, as such faces are thought to indicate high levels of heterozygosity in their owners (Symons, 1979; Thornhill and Gangestad, 1993). As extreme genotypes that are more likely to be homozygous for deleterious alleles (Symons, 1979) and less likely to have alleles to which pathogens are poorly adapted (Thornhill and Gangestad, 1993) are selected against, a preference for average faces may be adaptive. A preference for averageness is compatible with both cognitive theories of prototyping (see Langlois and Roggman, 1990) and work in theoretical biology suggesting that a preference for "average" phenotypes would rapidly replace random mating (Koeslag, 1994; Koeslag, 1994). However, the averageness hypothesis has received only mixed empirical support. Modern computer graphic techniques develop a method first suggested by Francis Galton (1878) to allow composite "average" faces to be constructed from much larger sets of individual photographs. Langlois and Roggman (1990) found that composite (average) faces were rated as more attractive than the individual faces that formed them. A preference for average faces over extremely distinctive faces has been demonstrated using line drawings, although the comparison is confounded by symmetry (Rhodes and Tremewan, 1996). A preference for averageness may explain the popular belief that people are attracted to opposite sex individuals that resemble themselves. Penton-Voak et al. (1999a) presented female subjects with a series of synthetic male faces, one of which had been generated from their own face (a gender transform--see Rowland and Perrett, 1995). It was found that subjects rated faces more similar to their own as more attractive than other faces. A preference for averageness, however, rather than similarity, could generate this finding, because faces very far from average receive low attractiveness ratings and such atypical faces differ from the faces of most individuals more than average faces. Alley and Cunningham (1991) proposed that average-sized features would be attractive but not optimal, and briefly reviewed studies using methodologies not reliant on digital averaging that failed to support the averageness hypothesis. Perrett etal. (1994) demonstrated that, although average composite male and female faces are indeed attractive, more attractive composites can

226

IAN S. PENTON-VOAK AND DAVID I. PERRETT

FIG, 2. A n original male face (top left) can be warped into an: average face shape derived from a sample of 26 young adult males (bottom left). Color information from an average face can be warped into an individual face shape (top right). The average face is shown b o t t o m right,

SHIFTINGFEMALEPREFERENCESACROSSTHEMENSTRUALCYCLE

227

be generated by using a subset of the 15 most attractive faces from a set of 60 individuals. A counterargument to this finding suggests that the 15 most attractive individuals in the group may be, in fact, the 15 most "average" faces in the group, and, hence, a composite of them represents the population average better than the full 60-face group. This argument does not explain why a caricature of the 15-face female average (exaggerating the differences between the 60- and 15-face average by 50%) creates a yet more attractive face (Perrett et al., 1994). Pollard et al. (1999) used facial-metric measurements of feature size rather than composites to calculate averageness, and found that faces with features close to the average size of the population studied were rated as exactly t h a t - - a v e r a g e - - i n attractiveness. 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 and Cunningham, 1991; Benson and Perrett, 1992). Manipulations of averageness using combinations of average shape and average color information from composites reveal that both skin texture and face shape influence attractiveness. Individual faces warped into average face shapes (Fig. 2) are rated as more attractive than the original (Figs. 2 and 3). Similarly, rendering the color information from a blended average into an original shape improves attractiveness ratings. Average shape

5 m

~0 °~,~ t~

l" 4,

ia 0.1

a.

t~

3, 11"

Original

Shape and Color

Original Shape,

Average Color

Average Shape, Original

Average Shape and Color

Color

FI6. 3. Mean and SE of attractivenessratings givenby 28 subjectsto 16 originalmale images, 16 original face shapes with average color, 16 faces with original color and average shape, and the average image (N = 28).

228

IAN S. PENTON-VOAKAND DAVID I. PERRETT

and color together typically result in the highest attractiveness ratings (Figs. 2 and 3; Benson, 1993; Little, 1999). Dissociating symmetry and averageness is problematic. As no reliable directional asymmetries have been found in human faces at rest (e.g., Perrett et aL, 2001), average faces are by definition symmetric. Average size and symmetry of real faces, however, may not be simply related. Studies of bird species indicate that larger than average secondary sexual characteristics (such as swallows' tails) are more symmetric than smaller traits (MOller and Hoglund, 1991). In women, breast asymmetry is negatively related to breast size (large breasts are relatively more symmetric than small breasts; Manning et aL, 1997) and, in men, facial symmetry appears to be related to jaw and cheekbone size (Scheib et aL, 1999). So although in composite images symmetry increases as facial averageness increases, this may not reflect patterns of symmetry in real faces. Symmetry, rather than averageness, may drive the attractiveness judgments of composite faces. However, a study by Rhodes et al. (1999) independently manipulated averageness and symmetry and concluded that both positively influence attractiveness judgments.

C.

SECONDARY SEXUAL CHARACTERISTICS AND SEXUAL DIMORPHISM

Symons (1994) proposes that averageness may be a "default" position that defines facial attractiveness unless another psychological mechanism is operating. A preference for secondary sexual characteristics may be one such alternative mechanism operating in women's judgments of male facial attractiveness, as it appears to be in other species (Andersson, 1994). The growth of male secondary sexual characteristics in mammals reflects androgenic activity; such hormones are thought to be immunosuppressive (e.g., Hillgarth and Wingfield, 1997). Consequently, male traits are often hypothesized to represent an "honest handicap" (Zahavi, 1975; Folstad and Karter, 1992). 1.

Sex Differences in Facial Growth

Sexual selection is often implicated in sexual dimorphism; male-male competition leads to sexual selection for successful (i.e., large, strong) males. Male-female dimorphism in human faces radically increases at puberty under the influence of sex steroids, particularly testosterone and estrogen. Levels of both hormones increase in both pubescent boys and girls, although the ratio of androgens to estrogens is dependent on sex. 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 (Thornhill and Gangestad, 1996; Gage et al., 1999). As masculinity in mammals is

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

229

testosterone dependent, whereas femininity represents the more neutral developmental state (Haaq and Donahoe, 1998; Owens and Short, 1995), male faces grow more than female faces, which remain relatively childlike. High estrogen apparently prevents facial bone growth in pubertal females, and leads to fat deposition in the lips and cheek area (Thornhill and Gangestad, 1999). Considerable evidence suggests that extremely feminine female faces are considered attractive. A wide variety of techniques ranging from (1) measurement of facial photographs of women (Cunningham, 1986; Grammer and Thornhill, 1994; Jones and Hill, 1993) through (2) studies of facial composites (Perrett et aL, 1994, 1998; Rhodes et aL, 2001) to (3) the generation of attractive female face shapes using genetic algorithms (Johnston and Franklin, 1993), show that feminine features indicating estrogenized female faces increase 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 (estrogen levels decrease in adult females with age) and reproductive health (Symons, 1979,1994; Thornill and Gangestad, 1993,1996,1999; Singh, 1993). A wealth of studies indicate that femininity (an exaggeration of female sextypical features) rather than averageness is most 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 (cheek bone prominence and jaw size). Cunningham et aL (1990) and Grammer and Thornhill (1994) used facial measurements, and found a female preference for large jaws in males. "Masculine" features, such as a large jaw and a prominent brow ridge, are also reliably associated with ratings of dominance in photographic, indenti-kit and composite stimuli by male and female raters (McArthur and Apatow, 1983-1984; Berry and Brownlow, 1989; Berry and Wero, 1993; McArthur and Berry, 1987; Perrett et al., 1998). Facial dominance appears to correlate with status in some human hierarchies (Mueller and 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 the opposite (McArthur and Apatow, 1983-1984; Berry and McArthur, 1985; Perrett et aL, 1998). Berry (1991) had a large set of male photographs rated for attractiveness and, independently, babyishness. She found, essentially, evidence for two types of attractiveness--sincerity (associated with attractive baby-faces) and

230

I A N S. P E N T O N - V O A K A N D D A V I D I. P E R R E T T

power (associated with attractive "mature" faces). As well as preferences for some typically "masculine" features, Cunningham etal. (1990) also found preferences for more feminine features (such as large eyes and an expressive mouth). Cunningham et al. suggest that women have "multiple motives" when choosing mates: optimally attractive faces may simultaneously appear dominant and yet elicit nurturing responses.

2. Manipulating Sexual Dimorphism in Faces Using Computer Caricaturing Techniques Although comparative approaches to human facial preferences suggest that exaggerated sex-typical traits will be attractive in both male and female faces, earlier studies (reviewed above) suggest that although this may hold true for female faces, preferences for male faces are more complex. Our recent work has used computer graphic techniques to manipulate facial sexual dimorphisms in stimuli used in preference tests; some of this work is reviewed below. Computer graphic techniques can be used to construct average male and female faces by digitally blending photographs of individuals of one sex. Sexual dimorphism in face shape can then be enhanced or diminished by looking at the geometrical differences between male and female face shapes and either exaggerating or minimizing them. For example, "masculinizing" a male face shape by increasing the differences between a male and female average increases the size of the jaw and reduces lip thickness (among all other dimorphic characteristics), as male jaws are larger than female jaws and the lips of men are thinner than those of women. These "masculinized" or "feminized" versions of male and female composite faces are useful stimuli for testing the influence of sexually dimorphic characteristics on judgments of facial attractiveness, although they are not without critics. Meyer and Quong (1999) suggest that as our stimuli are based on male-female differences rather than within-sex differences they fail to accurately simulate differing androgen levels in male faces. Although we concede that stimuli based on within-sex differences in sex steroids would be superior to our stimuli, we assume that male-female differences parallel the differences between individuals with high and low androgens (Perrett and Penton-Voak, 1999). Femininity in mammals represents the more neutral developmental state: masculinization is dependent on androgens (Owens and Short, 1995). Typically male XY individuals with complete androgen insensitivity syndrome have a female appearance (although they do have "male" dentition; Haaq and Donahoe, 1998). It seems unlikely that feminizing a face by applying male-female differences produces facial shapes associated with higher than average levels of androgens.

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

231

Fie. 4. Maleimages:(left)50% feminizedand (right) 50% masculinized.

Using this technique, the amount of "masculinity" or "femininity" transform applied to a face is specified as a percentage. The shape differences between male and female faces are described by a set of vectors between marked delineation points on the features of the male and female averages (172 delineation points define the outline of the face, the eyes, the mouth, the nose, etc.). Transforms are expressed as a percentage of the distance traveled along these vectors: in a 25% feminized male face shape, each delineation point is moved 25% of the way along the vector to the female average face. The color information from the original male average is then warped into this new shape. To masculinize male face shapes, the direction of the malefemale vector is reversed before the points are moved along it (see Fig. 4 for examples of feminized and masculinized male face stimuli). Such manipulations of both Japanese and Caucasian face stimuli were presented to Japanese and Caucasian adult males and females in their country of origin (Perrett et al., 1998). Participants could alter the appearance of a face on a computer monitor by using a computer mouse. Moving the mouse left or right would change the amount of masculinity or femininity in the face in real time, under the subject's control. For the male face stimuli, the shape selected by Caucasians as most attractive (from the shape range available) was significantlyfeminized for both the

232

IAN S. PENTON-VOAKAND DAVID I. PERRETT

Caucasian male face and the Japanese male face continua. Japanese participants also selected significantly feminized versions of the male stimuli for both the Japanese and Caucasian male face continua. There were no effects of subject sex, subject population, type of stimulus face or any interactions between main effects (Perrett et aL, 1998). Previous studies show crosspopulation consistency in attractiveness judgments. Our study showed crosscultural (between population) agreement in the preference for feminized over average face shapes, which further refutes the averageness hypothesis. This preference for feminized male faces seems contrary to predictions from indicator mechanisms of sexual selection and to some other published studies of male facial attractiveness reviewed briefly above. Yet, other studies report a somewhat similar trend [e.g., Berry and Zebrowitz McArthur's (1986) reported preference for "babyfaced" men; the "multiple motives" theory of Cunningham et al. (1990)]. Rather than preferring typically masculine faces (with prominent brow ridges and large jaws), both male and female adults appear to favor a small amount of femininity in men's faces, using stimuli prepared in this way. One possible explanation for this somewhat unusual finding may rest on stereotypical personality attributions that individuals make to faces. Evidence in favor of this hypothesis is reviewed next.

III.

A.

PERSONALITY ATTRIBUTIONS AND FACIAL CHARACTERISTICS

Is PHYSIOGNOMY ACCURATE?

The explanation for the unexpected preferences for feminized male faces reported above may lie in the personality attributions these composite faces elicit (Perrett et al., 1998). The attribution of personality to faces, physiognomy, has had a long and checkered history across many cultures. Yet, on the whole, scientists of the twentieth century have dismissed physiognomists as being, at best, misguided: Respectable science h o l d s . . , that there is not any connection between the features of the face and the character of the p e r s o n . . . Any connection between these two wholly different kinds of personal qualities would bespeak of some mystical system of correspondences between the mind and body; scientists could only regard such a system as absurd (Brandt, 1980; in Berry and Brownlow, 1989).

Despite such condemnation, Liggett (1974) demonstrated that the advice of science has clearly been ignored: 90% of university students see the face as a valid guide to a person's character. Contrary to the orthodox scientific view that physiognomy is a worthless area for study that should be classified (and condemned) along with phrenology, recent literature contains a number

SHIFTING FEMALE PREFERENCES ACROSSTHE MENSTRUAL CYCLE

233

of studies concerning personality judgments from physical characteristics. These will be briefly reviewed next. B.

STUDIESOF ZERO ACQUAINTANCEAND THE "BIG FreE" PERSONALITY FACTORS

Although physiognomists have often been dismissed as charlatans or worse (e.g., Cleeton and Knight, 1924), their work may have been, in at least part, based on observable relationships between physical appearance and personality. Studies using "zero acquaintance" paradigms (in which participants rate the personality of strangers) have found a surprising degree of correlation between self-ratings and stranger ratings on personality factors, often using five-factor models of personality (e.g., Norman, 1963). Many trait theorists agree that five trait dimensions seem to represent a reasonable compromise that is at least a partially accurate measure of people's personalities (e.g., McCrae and Costa, 1987; Watson, 1989; Barrett and Pietromonaco, 1997). These five dimensions have been derived from factor analyses of ratings of individuals using trait adjectives (e.g., talkative-quiet, cold-warm, etc., See Norman, 1963; McCrae and Costa, 1987; Botwin et al., 1997). Although different researchers disagree on the best names for these five factors, they are commonly referred to as Extraversion, Agreeableness, Conscientiousness, Emotional Stability (Neuroticism) and Openness-Intellect (or Culture). Without verbal interaction, individuals in small groups are to some extent able to make accurate perceptions of others' self-reported personalities. Using small group paradigms, in which four or five unacquainted individuals assess each others' personalities, factors of extraversion, agreeableness, and conscientiousness are moderately accurately inferred; culture and emotional stability less so (e.g., Passini and Norman, 1966; Albright et al., 1988; Watson, 1989; see Kenny et aL, 1994 for a review). These findings seem to generalize cross-culturally: Albright et al. (1997) used a similar paradigm with Chinese students in Beijing, and reported a similar pattern of results to those found in Western studies. Small group paradigms, of course, allow nonverbal communication and interaction to convey personality. Nonetheless, Kenny et al. (1992) replicated consensus and accuracy in personality perception using videotapes of individuals, and Borkenau and Liebler (1992) showed that somewhat legitimate personality judgments can be made from still photographs alone. There are both individual and sex differences in accuracy of personality attributions. Ambady et al. (1995) report that women are more accurate judges of strangers' personalities than men, consistent with other literature on nonverbal behavior.

234 C.

IAN S. PENTON-VOAKAND DAVID I. PERRETT THE EFFECTS OF "MATURE" AND "BABYFACE" CHARACTERISTICS ON ATTRIBUTIONS OF FACIAL DOMINANCE

Several researchers have devoted much effort to examining the characteristics and consequences of faces that are described as "dominant" or "babyfaced." Dominance can be considered as a trait encompassed by the extraversion factor in the five-factor model (although it appears to load more heavily on the agreeableness factor in some studies), but it has received considerable attention in studies outside of this theoretical orientation. Dominance is a characteristic that is closely linked to male reproductive opportunities and success in many species including humans. Dominant looking male teenagers, for example, copulate earlier than less dominant looking peers (Mazur et al., 1994). Thus, the characteristics that lead to attributions of dominance may influence attractiveness judgments. Most studies agree on the characteristics that make male faces appear dominant: mature features, such as a large jaw and a prominent brow ridge, are reliably associated with ratings of dominance in photographic, indentikit, and schematic stimuli (Berry and Wero, 1993; Berry and Brownlow, 1989; McArthur and Berry, 1987; McArthur and Apatow, 1983-1994). Similarly, the literature agrees that babyfaces are characterized by smaller chins, high eyebrows, and larger eyes. Such faces are generally rated as being warmer, more honest, and more sincere, but also more naive and less physically strong (McArthur and Apatow, 1983-1984; Berry and McArthur, 1985; Berry, 1991). Babylike and dominant faces reliably elicit personality attributions cross-culturally (e.g., Keating et aL, 1981; McArthur and Berry, 1987) but their effect on attractiveness judgments is equivocal as discussed above. Such dominance attributions, however, influence potentially important social interactions. Facial maturity has been shown to have more of an effect on male daily social experience than on female routine interactions (Berry and Landry, 1997). By simulating a trial situation, Berry and Zebrowitz McArthur (1986) demonstrated that manipulating the maturity of the "defendant" influences both the chance of "conviction" and the "sentence" given. Babyfaced individuals were less likely to be found guilty of charges involving intentional criminal behavior, although they were more likely to be found guilty of a negligent crime. Babyfaced defendants were also given lighter sentences for negligen t crimes, reflecting the effects of attributed naivety and honesty. In a study with perhaps more ecological validity, Mueller and Mazur (1997) showed that facial dominance (as rated by undergraduates) of the graduates from the West Point Military Academy in 1950 predicted their final rank at the end of their careers. Such an environment may reward success in physical male-male competition more than many other walks of life.

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

235

Consensus in judgments of the personality of mature or babylike faces does not necessarily indicate that such attributions are valid, even though they influence potentially important social situations. Many of the personality attributions made to babylike faces are also, unsurprisingly, made to infants themselves. Lorenz (1943) suggested that positive behavioral responses to infants are adaptive in that they increase offspring viability. Similar responses to adult faces with some childlike qualities may be inappropriate generalizations of an otherwise adaptive behavior. This argument alone, however, fails to explain the evidence that babyfaced or mature individuals do to some extent self-report that they have the personality characteristics attributed to them. Berry and Brownlow (1989) found that ratings of male babyishness were positively correlated with the face owner's self-reported approachability and warmth, but negatively related to self-reported aggression. For female faces, babyishness was associated with low self-reported levels of physical power and assertiveness. Bond et al. (1994) demonstrated that individuals whose faces were rated as being "less honest" were more likely to volunteer for experiments that involved them in deceiving others than people who were judged to look more honest. Such correspondence between self-reported and observed personality variables is often explained in terms of a behavioral confirmation or "selffulfilling prophecy" effect (Watson, 1989; Berry and Wero, 1993). It seems that certain facial configurations reliably elicit certain personality judgments. If someone is regularly perceived and treated as if they are submissive (or dominant), they may modify their behavior so that they act in a more submissive (or dominant) manner. This, in turn, may lead to an internalization of such behavior, and an individual may come to see themself as possessing the characteristics with which they are stereotypically attributed. The behavioral confirmation theory may operate to strengthen the relationship between expected and observed behaviors of male strangers. It does not, however, provide a convincing explanation of the origin of stereotypes necessary for consistent ratings of dominance and other characteristics (Watson, 1989). Cultural factors are often invoked, and yet, cross-cultural studies show considerable agreement in attributions of dominance. This has led some researchers to posit a biological basis for the personality ratings given to faces. As the differences between "dominant" and "baby" faces clearly parallel the differences between "masculine" and "feminine" composite stimuli, they probably reflect underlying androgen levels. Testosterone is also linked to male dominance behaviors (Mazur and Booth, 1998), tentatively providing a biological link between facial appearance, behavior, and valid (though stereotypical) personality judgments. Given the reported importance of personality to human mate choice (e.g., Buss, 1989), it seems likely that stereotypical personality judgments may

236

I A N S. P E N T O N - V O A K A N D D A V I D I. P E R R E T T

influence attractiveness judgments of unfamiliar faces. First impressions last, after all.

1. Personality Judgments Made to Composite Faces To see whether personality attributions influenced the attractiveness judgments of our composite stimuli, we presented sets of three Caucasian and Japanese male composite stimuli (in masculinized, average, or feminized versions) to university undergraduates and research staff in Scotland (Perrett et aL, 1998). Increasing the masculinity of face shape across the three set members increased ranking of perceived dominance, masculinity, and age, but decreased ranking of perceived warmth, emotionality, honesty, cooperativeness, and quality as a parent (Friedman's X2; N = 20,p < 0.0005 for each dimension). Even with the small number of raters in our sample, and the relatively small differences between the stimuli, consensus in attributions was nearly total. These results demonstrate that exaggerating or reducing sexual dimorphism in composite faces alters their perceived masculinity or femininity in the predicted direction. Thus, manipulations of this sort clearly influence perceptions that convey the "psychological meaning" of masculinity and femininity (Meyer and Quong, 1999; Perrett and Penton-Voak, 1999). In addition, the considerable consensus in the ranking of stimuli indicates that, although masculinized and feminized versions of stimuli are fairly similar, they must contain stereotypical cues to personality. Socially valued traits such as honesty, warmth, cooperation, and skill as a parent are associated with feminized versions of male faces, whereas traits such as dominance are, as predicted, associated with masculinized face shapes. A female preference for slightly feminized male faces can perhaps be at least partially explained by these personality attributions. Although biological predictions such as handicap theory indicate that females should prefer masculinized faces, such faces elicit negative personality attributions. Crossculturally, personality factors are reported to be the most important factor in mate choice by both sexes (Buss and Barnes, 1986; Buss, 1989). So, it seems inconceivable that personality attributions could have no effect on attractiveness judgments. Feminization of male face shape may increase attractiveness because it softens particular features that are perceived to be associated with negative personality traits.

2. Testosterone, Behavior, and Face Shapes The behavioral confirmation theory of accurate personality ratings and a putative biological theory of facial dominance may in fact interact and reinforce one another. Male testosterone levels are known to increase in the winners and decrease in the losers of competitions (e.g., Mazur et aL, 1997). Success in dominance interactions may increase testosterone levels in

S H I F T I N G F E M A L E PREFERENCES ACROSS T H E M E N S T R U A L CYCLE

237

a way that influences the growth of male facial characteristics, especially in adolescents. A positive reciprocal relationship between dominance behavior influenced by stereotypical personality attributions and levels of testosterone is therefore plausible (Gage et aL, in preparation; Manning, pets. comm.). Increasing the testosterone level in adult males is also associated with more troubled sexual relationships. A sample of 4462 former members of the U.S. armed forces showed that men with testosterone levels one standard deviation above the mean (estimated from one measurement) were 50% more likely never to marry than men with testosterone one standard deviation below the mean (Booth and Dabbs, 1993). The same study demonstrated that married men with high testosterone were also more likely to suffer troubled relationships, increased incidence of domestic violence, and extra-pair sex. The implications of these findings for theories of sexual selection in humans are clear: whatever indirect "good genes" benefits high testosterone may offer in terms of offspring viability, there is an apparent cost in terms of likely paternal investment. It is worth reiterating that women make more accurate personality judgments in zero-acquaintance studies than men (Ambady et aL, 1995). As lower initial investment in offspring makes males the sex most likely to desert (Symons, 1979), selection pressures to make accurate personality judgments might be expected to be stronger in females than males. In the holistic transforms employed to masculinize and feminize the stimuli used in our experiments, all male-female facial differences visible in frontal facial photographs were manipulated. Our methodology has generated a robust and reliable preference for slightly feminized male stimuli using participants (university students) matched in age with those stimuli, both in the United Kingdom and Japan and also in South Africa (Perrett et al., 1998). This preference is unexpected given that dominant-looking males copulate earlier than their less-dominant-looking peers (Mazur et al., 1994) and succeed in competitive hierarchies (Mueller and Mazur, 1997), and that "cultural success" as reflected by status in social hierarchies is positively correlated with copulation frequency in contemporary Canada (Perusse, 1993). Cues to cultural success (such as high-status clothing) undoubtedly positively influence attractiveness judgments of males (e.g., Townsend and Roberts, 1993; Townsend and Wasserman, 1998). Testosterone appears to be related to dominance behavior and so, by extension, position in hierarchies, but simultaneously testosterone is associated with less likelihood of extended paternal investment (Mazur and Booth, 1998). One aspect of female mate choice may be a cost-benefit analysis of such conflicting factors: a high status/high testosterone partner may have access to more resources, and offer heritable benefits to offspring; but he may also be more likely to desert or even injure the female who chooses him (Mazur and Booth, 1998). A female preference for a lower status partner

238

IAN S. PENTON-VOAK AND DAVID I. PERREq'T

who is more likely to provide long-term paternal investment may, overall, pay a greater reproductive dividend. Given that stereotypical personality judgments have some accuracy, such attributions may form an important component in women's mate-choice decisions. With our stimuli, personality attributions apparently outweigh biological fitness cues in attractiveness judgments of male faces. In real faces, perceived personality may not be as closely linked to masculine characteristics: combinations of masculine and feminine features, indicating both status and kindness may be optimally attractive. There is some support for such multiple motives influencing judgments of facial attractiveness (Cunningham et al., 1990; see also Scheib et al., 2001). The preferences we have found indicate a selection pressure that has acted against exaggerated dimorphism in male and female faces. Because more feminine face shapes are perceived as younger, the preferences would encourage a "youthful" facial structure in the human species generally. There is some paleo-anthropological evidence consistent with such a hypothesis, as both cranial and postcranial robusticity have been declining throughout the recent evolution of H o m o (e.g., Brace et al., 1987; Ruff et al., 1993). A move toward more gracile morphology is often attributed to advances in food preparation and other technologies, and it is, of course, unclear whether sexual selection has played a part.

IV. MENSTRUAL CYCLE SHIFFS IN FACE PREFERENCES A.

THE HUMAN OVARIAN CYCLE--CONCEALED OVULATION AND CONSTANT RECEPTIVITY.9

In contrast to most other mammals and many other primates, human female reproductive cycles do not have a clearly defined estrus period of sexual activity around ovulation. Instead ovulation is hidden both from males and to (most) females themselves. Women have intercourse, and similarly can refrain from intercourse, at any point in their menstrual cycle, leading some theorists to view human sexual behavior as "emancipated" from hormonal influences (Ford and Beach, 1951). To consider women continuously receptive implies a continuous state of estrus in which females are active solicitors of male attention (Hrdy and Whitten, 1987). This is an overstatement, as Frank Beach memorably noted in 1974: "Any male who entertains this illusion must be a very old man with a short memory or a very young man due for a bitter disappointment," a sentiment echoed by Symons (1979): "The sexually insatiable woman is to be found primarily, if not exclusively, in the ideology of feminism, the hopes of boys, and the fears of men."

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

239

The evolution of concealed ovulation and continuous receptivity has nonetheless attracted much attention from evolutionarily minded scholars of human behavior. Various hypotheses have suggested that, for example: (1) Concealed ovulation and continual sexual receptivity would strenthen the human pair-bond, and hence, cooperation within the group necessary for successful hunting/gathering lifestyles (e.g., Morris, 1968; Lancaster, 1975); (2) Ovulation was concealed as an adaptation to force desirable males into consort relationships, and thereby increase paternal investment (e.g., Alexander and Noonan, 1979; Symons, 1979); (3) Hominid females, with increasing intelligence, wished to avoid the danger and pain of pregnancy by consciously avoiding copulation around ovulation, but those females that could detect ovulation and avoided conception were selected against (Burley, 1979); (4) Cryptic ovulation confuses paternity, preventing infanticide (Hrdy, 1981); or (5) It allows hominid females to gain good genes benefits in extra-pair partners (Benshoof and Thornhill, 1979; Schr6der, 1993). There are several in-depth reviews of these theories in the literature (Strassman, 1981; Gray & Wolfe, 1983; Steklis and Whiteman, 1989). Choosing between such hypotheses is difficult, as many are based on the assumptions that human concealed ovulation is a derived trait (from an ancestor with cyclic estrus swellings) and that something is known about the social structure and mating patterns of ancestral hominids. In fact, it is not at all clear that concealed ovulation is a derived trait from either comparative research with extant primates or studies of primate phylogeny (e.g., Dixson, 1983; Hrdy and Whitten, 1987; Burt, 1992; Sill6n-Tulberg and M¢ller, 1993). Furthermore, little can be concluded about mating systems from sexual dimorphism in fossil hominids. First, small sample sizes for separate hominid species make sexing fossil specimens problematic (e.g., Hausler and Schmid, 1995; Tague and Lovejoy, 1998) and estimating dimorphism unreliable (McHenry, 1994). Second, those estimates of sexual dimorphism that do exist reveal a curious pattern in the australopithecines. Whereas body size dimorphism is large, indicating strong male-male competition in early hominids, canine dimorphism is unexpectedly low (Plavcan and van Schaik, 1997). Although it is probably safe to say that neither polyandry nor monogamy was the mating system of ancestral hominid species, it is hard to say whether a unimale or multimale system was prevalent in our ancestors (Sill6n-Tulberg and M¢ller, 1993; Plavcan and van Schaik, 1997). B.

CHANGES IN SEXUAL B E H A V I O R ACROSS THE M E N S T R U A L CYCLE

Despite the obvious lack of a true estrus phase in women, many researchers have proposed that female sexual desire, behavior, and preferences may vary across the menstrual cycle. The sources of this variation may

240

IAN S. PENTON-VOAK AND DAVID I. PERRETT

be social (e.g., sex during menstruation and postpartum is taboo in many cultures; Hrdy, 1981) or biological (resulting from the influence of fluctuating levels of sex hormones). Women's menstrual cycles last between 21 and 35 days, with a mean duration of around 28 days. Many commentators have pointed out the great irregularity of cycle lengths both within and between women, which is especially interesting in comparison to the regularity of most mammals (e.g., Burley, 1979). Irregularity, however, also seems common in some nonhuman primates (e.g., Papio ursinus and Pan troglodytes, Butt, 1992). Nevertheless, most standard models of the menstrual cycle are based on a 28-day duration. In such models, ovulation occurs on approximately day 14 at the end of the follicular phase (Regan, 1996). The first 5 days (early follicular phase) is the period of menstruation, in which levels of estrogen, progesterone, and androgen are all low. In the midfollicular phase (days 6-10), estrogen levels rise steadily, androgens increase somewhat and progesterone levels remain low. Days 11-14 (the late-follicular phase) are characterized by a rise in androgen and progesterone and a rapid increase of estrogen levels, which peak approximately 2 days before ovulation. This estrogen peak causes a surge of luteinizing hormone, which, in turn, leads the follicle to rupture and the ovum to be released on day 14 in standard models. The 14 days following ovulation, the luteal phase, are characterized by secretion of progesterone from the corpus luteum. In this phase, progesterone and estrogen levels rise until a drop in the level of both hormones occurs before the onset of menstruation (the beginning of the next cycle). As human sperm may survive in the female reproductive tract for around 5 days (Baker and Bellis, 1995), copulations in the 5 days preceding ovulation (mid- to late-follicular phase) are most likely to result in conception. Despite the model given above, ovulation is actually fairly unpredictable, especially in the first years following menarche. Social factors, such as a lot of time spent in the presence of males, may shorten the cycle; external factors such as stress may result in anovulatory cycles (Metcalf et al., 1983). There is even the possibility that women, like some other mammals, may ovulate in response to copulation during the follicular phase (J6chle, 1973; Baker and Bellis, 1995). Peaks in sexual desire and activity have been reported around ovulation (e.g., Urdy and Morris, 1968; Adams et al., 1978; Stanislaw and Rice, 1988); in the midfollicular phase (Urdy and Morris, 1977; Bancroft et al., 1983), and also in the late-luteal/premenstrual period (e.g., Stewart, 1989). Further studies report multiple peaks of sexual desire and/or activity, often at ovulation and in the late-luteal phase (e.g., Silber, 1994). Dissociating various factors (e.g., male or female initiation of sexual activity) in this body of

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

241

research is problematic. Cultural proscriptions against sex during menstruation may influence reported patterns of desire. Some evidence that culturally enforced abstinence may account for perimenstrual peaks in sexual desire comes from Matteo and Rissman (1984), who investigated peaks of sexual activity in lesbian couples who did not reduce their sexual activity during menstruation, but did report an increased activity midcycle. Regan (1996) presents a comprehensive review of the literature and concludes that there is substantial variation between women who may have no, one, or two peaks of sexual desire. Women who do have peaks in sexual desire tend to have them either in the midfollicular, late-follicular (ovulation), or late-luteal phases of the menstrual cycle. Follicular phase or midcycle peaks in sexual interest would parallel many other primate species, as such peaks would be associated with likely conception following sex (Hrdy, 1981). Sexual desire, then, varies across individual women's cycles and between different women (Regan, 1996). There appears to be no simple species-typical pattern of human female proceptivity or receptivity that links sexual desire to hormonal activity. Humans are not unique in this regard--sexual receptivity in other primates varies across species, and between and within groups (Rowell, 1972), and is clearly influenced by social, as well as hormonal factors (Keverne, 1976). Despite confusion as to exactly when peaks in sexual interest occur, two studies report that women with partners may be more inclined to seek sex outside the pair bond when conception following sex is most likely. Baker and Bellis (1990, 1995) report data indicating that the rate of female extrapair copulations (EPCs) is around 2.5 times higher during the follicular phase than in the luteal phase, which possibly promotes sperm competition when conception is likely. A decrease in EPC frequency during nonfertile phases of the menstrual cycle may reduce the risk of being discovered, and hence, the risk of desertion by the cuckolded male. Baker and Bellis' study has generated much controversy as the data were collected from a questionnaire in Company magazine, introducing the possibility of some self-selection bias among participants. A further study (Worthman, 1978; reported in Hrdy, 1981) supports the findings in a hunter-gatherer society, using hormonal measurements to determine menstrual status. Worthman reports that there is a midcycle increase in sexual activity with both husbands and lovers among women of the !Kung San. A series of studies of women's clothing in Viennese discotheques (e.g., Grammer et al., 1997) indicate that women dressed in tighter, more revealing clothes in the ovulatory phase of the menstrual cycle (as revealed by estradiol assays). As such clothing increases women's self-rated "sexiness" in the context of a nightclub, this may be an example of proceptive female sexual behavior related to cyclic hormonal fluctuations.

242 C.

IAN S. PENTON-VOAKAND DAVID I. PERRETT CHANGES IN FEMALE PREFERENCES FOR MALE SCENT ACROSS THE MENSTRUAL CYCLE

Some evidence also suggests that female preferences for male characteristics may change across the menstrual cycle. Fitness benefits to offspring can only be realized if conception follows copulation, so females may be more attentive to phenotypic markers indicating fitness during the follicular phase of the menstrual cycle when conception is most likely (days 6-14; Baker and Bellis, 1995; Regan, 1996; Gangestad and Thornhill, 1998). Female preferences for odor seem to have a cyclic component, with females being more receptive to odors at ovulation, coincident with increases in estrogen and luteinizing hormone (Doty et aL, 1981). Odor cues appear to play an important role in human mate choice, and women report odor to be a more important cue in mate choice than men (Herz and Cahill, 1997). In addition, there is some evidence that women are able to make biologically significant judgments of male odor that may be fitness relevant. Wedekind e¢ al. (1995) performed a "T-shirt experiment," in which males wore clean T-shirts for 2 nights. The worn shirts were then presented to female subjects, who rated their odor. Both female (the smellers) and male (the smelled) participants were typed for various genes in the major histocompatibility complex (MHC). Women rated the odors of men who had dissimilar MHC genes to themselves as preferable to MHC-similar men, which might maximize immune system function in offspring through heterozygosity (Wedekind and Furl, 1997). Although comparisons across various phases of the menstrual cycle were not made in this experiment, all women were tested around the 12th day of their menstrual cycle, coinciding with the occurrence of peak olfactory sensitivity and peak fertility. These preferences for men differing in immune system genotypes were reversed in women using oral contraception. Grammer (1993) studied female perception of androstenone, a chemical structurally related to testosterone that is found mainly in male rather than female sweat and urine. In contrast to the closely related androstenol, which has a pleasant, sandalwoodlike smell that is attractive to females, androstenone has an unplesant, urinelike odor. The exact mechanism of androstenone and androstenol production is unclear: they may both be byproducts of enzyme action on bacteria, or androstenol may oxidize to androstenone (see Grammer, 1993, for a brief review). Grammer 1993 showed that the generally unpleasant odor of androstenone becomes more acceptable in midcycle to women who do not use hormonal contraception. Furthermore, androstenol influences female mood at midcycle: females who placed a drop of androstenol on their top lip each morning for a month reported themselves to be more submissive than a placebo group at midcycle (Benton, 1982; see also Kirk-Smith et al., 1978).

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

243

1. The Scent of Symmetry? Three other studies of women's olfactory preferences across the cycle show that women may also make discriminations of male quality in the follicular phase of the menstrual cycle that they do not make at other times (Gangestad and Thornhill, 1998; Rikowski and Grammer, 1999; Thornhill and Gangestad, 1999). These studies employed similar methodologies to Wedekind et al. (1995) in which females rated the odor of T-shirts that had been worn by males for 2 nights. Although these studies are concerned with signaling qualities of male odor, unlike the studies mentioned above no specific chemical is proposed as a signal. Instead these studies concentrate on bodily and facial symmetry of the male T-shirt weavers. All three studies demonstrate that women are sensitive to and prefer the scent of males who are symmetrical, through some as yet undetermined chemical signal. This discrimination, however, appears to be influenced by hormonal status across the menstrual cycle. In the first of these scent of symmetry experiments (Gangestad and Thornhill, 1998), the level of fluctuating asymmetry of 41 men had been estimated from measurements of digit length, elbow width, wrist width, ankle width, foot breadth, and ear length and width. These men wore T-shirts for 2 nights before they were returned for rating by 45 female participants, 17 of whom used oral hormonal contraception. The nonpill using women were split into two groups according to menstrual cycle phase (with the follicular phase labeled "high conception risk," and menses and the luteal phase characterized as "low conception risk"). Men's fluctuating asymmetry significantly predicted the attractiveness ratings given to their scent by "high" but not by "low" conception risk women, and had no relationship with the ratings of women on the pill. Thornhill and Gangestad (1999) replicate their original finding with a larger sample (74 males, 78 females) and better controls of personal hygiene. Additionally, this more recent study showed that women prefer the scent of men whose facial photographs are rated as attractive at times of high fertility. Men, on the other hand, showed no preference for the odor of symmetrical females. Rikowski and Grammer (1999) replicated the findings of Thornhill and Gangestad in a similar study in Austria. Sixteen males, whose body asymmetry had been estimated from 7 bilateral body traits and whose facial asymmetry had been estimated from the position of bilateral facial structures (see Grammer and Thornhill, 1994), wore T-shirts for 3 nights. Forty female raters (14 of whom were estimated to be in the follicular phase when tested) assessed the odor of the worn shirts. Overall, women in the follicular phase rated male odor as sexier overall than women in other menstrual cycle phases. In addition, it was found that the ratings of male odor "sexiness"

244

I A N S. P E N T O N - V O A K A N D D A V I D I. P E R R E T T

given by women in the follicular phase of their menstrual cycle correlated significantly with bodily symmetry and nonsignificantly with facial symmetry and facial attractiveness (r = 0.48 and 0.47, respectively, p < 0.1). These relationships were not present in women in other, low conception risk phases of the menstrual cycle.

2. Interpretation of Cyclic Preferences These cyclic changes have been interpreted from a "good genes" perspective. Symmetry is a marker of (possibly) heritable developmental stability arid is a correlate of secondary sexual trait size in some species (M¢ller and Hoglund, 1991; Andersson, 1994), and hence of high level of androgens in mammals (Thornhill and Gangestad, 1996; Scheib et al., 1999). A preference for symmetric men when conception is most likely may be adaptive, increasing offspring viability. In combination with other evidence, such a mechanism may be especially important in extra-pair sexual activity. Females prefer symmetrical men as extra-pair copulation partners, and evidence reviewed above indicates that EPC activity peaks in the follicular phase (Thornhill and Gangestad, 1994; Baker and Bellis, 1995; Gangestad and Thornhill, 1997). As symmetry appears to be inversely related to the investment that men make in relationships, such choices probably do not reflect selection for paternal effort (Thornhill and Gangestad, 1997, 1998, 1999). D.

CYCLICSHIFTSIN PREFERENCESFOR FACES

Cyclic changes analogous to those found for odors have been shown to operate in the visual modality, specifically in preferences for faces. Frost (1994) developed photographs of 3 pairs of male faces to manipulate the skin darkness. Skin color is sexually dimorphic within all races, with males having darker skin than females (Van Den Berghe and Frost, 1986). This may be due to a link between the production of melanin and that of gonadotropins by the anterior pituitary. Frost's study showed that French Canadian women (N = 98) preferred the lighter face, but for women not using oral contraception (N = 56) this preference was attenuated when the estrogen/progesterone ratio was high--the follicular phase of the menstrual cycle. Breakdown of women by percentage of cycle completed shows a peak around ovulation in the number of dark faces preferred. Given the sexual dichromatism of skin color, a preference for darker skin at ovulation could be considered a preference for exaggeration of a male trait at ovulation.

1. Cyclic Preferences for Masculinity in Male Face Shapes The findings that traits relating to possible male "quality" are apparently more attractive to women at high-fertility phases of the menstrual cycle,

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

245

coupled with Frost's evidence for female cyclic visual preferences for male skin color, suggested that a study of cyclic female preferences for male facial shape might be fruitful. We performed a series of experiments in which composite male face stimuli with manipulated levels of masculinity and femininity were presented to women whose menstrual phase was estimated. We hypothesized that females may be attracted to more masculine faces with exaggerated secondary sexual facial characteristics when conception is most likely, that is, in the follicular phase of the menstrual cycle. This prediction was supported by the three experiments briefly described below.

2. A United Kingdom-Based Questionnaire Study We presented 5 male stimuli with differing levels of masculinity/femininity (50 and 30% feminized, the average, and 30 and 50% masculinized) in the BBC Tomorrow's Worm magazine, along with a short questionnaire with details of the female respondents' age, use of oral contraception, and number of days since the onset of their last period. Of 178 women who completed all of the relevant sections of the questionnaires, 39 used oral contraception. Following Gangestad and Thornhill (1998), a standard 28-day model of the female menstrual cycle was used to assign the 139 remaining female respondents (mean age of 30.7 years with a range of 14-50) to one of two groups based on their chance of conception. This was estimated from the number of days since the onset of the participants last menses: high conception risk, (N = 55; the follicular phase, days 6-14) or low conception risk (N = 84; menses, days 0-5, and the luteal phase, days 15-28). Figure 5 shows the percentage of women in each of the two groups who selected each of the 5 available faces as most attractive. Those in the lowconception risk group effectively picked faces at chance levels (though showing a tendency to prefer feminized faces), whereas the high risk group showed a significant preference for masculinized faces (Penton-Voak and Perrett, 2000). These preliminary data indicated that women are attracted to exaggerated male traits when conception following coitus is most likely (the follicular phase) and not at other times of the menstrual cycle.

3. A Longitudinal Study of Japanese Females A second experiment aimed to address some of the methodological weaknesses of the questionnaire study. Instead of a single-shot methodology, a repeated-measures design was employed, allowing the same women to be tested in different phases of their menstrual cycle. This study was conducted in Japan with Japanese students, and used male facial stimuli sets of both Japanese and white British origin. Oral contraception alters the hormonal

246

I A N S. P E N T O N - V O A K A N D D A V I D I. P E R R E T T

• high conception risk '-" . . . . . . . ".ption risk

40% 30% ° ~

"~ 20% to% 0% -50%

-30%

0%

30%

50%

Masculinity of face shape FIG.5. Percentageof subjects in high- and low-conceptionrisk groups selectingeach of the 5 stimuli (with varying levels of masculinityand femininity)as most attractive in the U.K. questionnaire study.From Penton-Voakand Perrett (2001).

processes of the menstrual cycle and appears to disrupt cyclic and other preferences (e.g., Wedekind et al., 1995; Thornhill and Gangestad, 1999). At the time of the study, oral contraception was unavailable in Japan. The young Japanese women tested here are very unlikely to have used hormonal contraception and hence provide a good population with which to test hypotheses based on normal cyclic changes in hormonal activity. If cyclic preferences are an adaptation to maximize fitness in offspring through extra-pair copulations and such shifts represent the interaction of hormonal and social factors, it is possible that women within relationships may manifest different patterns of response than those without a partner. Previous research suggests that women in all cultures value positive personality characteristics in a partner above all else (Buss, 1989). Research has indicated that stereotypical personality characteristics do indeed influence attractiveness judgments (Perrett et al., 1998; see review earlier in chapter). Once women have secured a partner, however, those seeking extra-pair partners may be influenced by considerations other than personality and future paternal investment. For example, cues to possibly heritable quality (as signaled by androgenic effects on face shape) may become more important. This experiment investigated such a possibility by presenting two sets (a Caucasian male average, and a Japanese male average) of 5 stimuli (40 and 20% masculinized, the average, and 20 and 40% feminized) to 39 Japanese students (Penton-Voak et al., 1999b). The participants took part in two experimental sessions, one in the follicular phase of their menstrual cycle, and one

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

,~ 30%

247

[] low conception risk • high conception risk

25% 20% o

'=~ 15%

;i 1°% 5% 0%

i

Japanese f a c e s .~

Caucasian faces

30% 25%

2O°/o ,.=

15%

lO% 0% -5% no partner (n =19)

partner (n =20)

FIG. 6. Effects of conception risk on femininity preferred by Japanese subjects in Japanese and Caucasian faces (top, from Penton-Voak et al., 1999b) and for subjects with (n = 20) and without (n = 19) a partner (below). Mean and SE illustrated.

in the luteal phase, as estimated from self-reports of average cycle length. In addition, participants reported whether they were currently in a "steady" heterosexual relationship. Overall, subjects preferred feminized faces (a replication of Perrett et al., 1998), and the origin of stimuli (Japanese or Caucasian) had no effect. As in the questionnaire experiment, women in the follicular phase showed a preference for significantly less-feminized faces (Fig. 6). Although cross-cultural consistency does not p r o v e that a behavior is an adaptation (ecological pressures may cause local adaptations), crosscultural replication adds some weight to an evolutionary interpretation of the results. Although no significant effects or interactions were found when the relationships of participants were analyzed, there was a tendency for women in relationships to prefer more masculine faces in general (p = 0.066; Fig. 6) and to show a greater follicular phase shift toward masculinity than women without partners (p = 0.08; Fig. 6). Both of these trends are consistent with an extra-pair copulation interpretation of cyclic preferences: Once a main

248

IAN S. PENTON-VOAK AND DAVID I. PERRETT

partner has been acquired, women may be less concerned with perceived personality indicating beneficial paternal behavior, and more concerned with cues to genetic quality in possible extra-pair partners (i.e., those features associated with testosterone). Needless to say, conscious awareness of such concerns is not necessarily implicated here. These cues may be especially important when extra-pair copulations are most likely to occur and when conception is most likely during the follicular phase of the menstrual cycle.

4. A United Kingdom-based Longitudinal Study Another laboratory-based within-subjects experiment tested the preferences of U.K. university students across their menstrual cycles (Penton-Voak et al., 1999b). An interactive methodology was used allowing female undergraduates to alter the apparent masculinity and femininity of the five male facial continua (four Caucasian and one Japanese). To further investigate the possible role of cyclic preference changes in short-term or possibly extrapair sexual relationships, participants who were not using oral contraception were asked to make one of two attractiveness judgments: to pick a face that would be attractive in a "long-term partner" (N = 26), or a face that would be attractive in a "short-term sexual partner" (N = 23). In this sample, cyclic shifts favoring relative masculinity in the follicular phase only occurred in females judging attractiveness for a short-term sexual partner; preferences for a long-term partner did not differ significantly across the cycle (Fig. 7, Penton-Voak et aL, 1999b).

20%

,w conception risk igh conception risk

15% O

"~

10%

0% "short term" context (n=23)

"long term"context (n=26)

Fie. 7. U.K. longitudinal study data. Effects of relationship context and conception risk on mean femininity preferred in five male facial continua. From Penton-voak et aL, 1999b. Mean and SE illustrated.

SHIFTING F E M A L E P R E F E R E N C E S A C R O S S T H E M E N S T R U A L C Y C L E

249

Females asked to pick a long-term partner in this experiment always preferred the same level of feminization in male face shapes across the menstrual cycle; such shapes elicit favorable personality attributions that may be related to behavior as discussed above. When likely paternal investment is low, however, as in short-term sexual encounters or extra-pair copulations, more weight is given to characteristics that may indicate "good genes" in partners, through facial traits thought to be linked to androgens. To some extent, this finding reconciles the overall preference for femininity in male faces with other literature on male sexual activity. As reported earlier, dominant (presumably high testosterone) men copulate earlier (Mazur et al., 1994) and achieve more sexual access as adults (Perusse, 1993). Townsend et al. (1995) report that high-status male college students, such as athletes, often reported more than 100 sex partners. Such a number of partners obviously reflects low male investment in each female. It has been suggested that this low investment in a given woman reflects the quality of the male involved: they are more attractive to women and can therefore obtain more partners, despite the women's alleged desire for a monogamous relationship (e.g., Symons, 1979, and pers. comm.; Townsend et al., 1995; Gangestad and Thornhill, 1997; Wiederman and Dubois, 1998). Nonetheless, it is widely accepted that biparental care is an adaptation: men and women desire a long-term pair bond as it has led in the past to greater reproductive success. Men do not have to be tricked into socially monogamous relationships: strategies of zero male parental investment will be selected against if they result in lower reproductive success (Burley, 1979). High-testosterone men apparently fail either to attempt or to succeed in marital relationships (Mazur and Booth, 1998). The data of our U.K.-based longitudinal cycle study imply that higher testosterone men may be favored for short-term sexual relationships when conception is likely, whereas men with faces that indicate relatively lower testosterone, with their concomitant likelihood of extended paternal investment, are preferred for long-term relationships. At present, however, this hypothesis clearly relies on several assumptions that remain untested. In two of three of these cycle/face preference experiments, and all of our previous work using composite stimuli, we have found an overall preference for more feminine than average faces. Facial masculinity, however, is associated with dominance and some aspects of reproductive success such as sexual access (as reviewed above). Access to reproductive opportunities will reflect diverse influences. Male-male competition may favor different characteristics to those favored by female choice. Nonetheless, success in male-male competition and the accompanying status and dominance advantages that it bestows will be another factor influencing judgments of attractiveness in real-life situations.

250 E.

IAN S. PENTON-VOAK AND DAVID I. PERRETT CYCLIC PREFERENCES, MIXED FEMALE MATING STRATEGIES, AND ~GOOD GENES"

The experiments summarized above found evidence of a cyclic preference for male face shape: women exhibit preferences for biologically relevant aspects of facial structure that may signal heritable genetic characteristics when conception is most likely. The results are also consistent with other cyclic preferences relevant to mate choice. Menstrual phase may be one variable that contributes to the difficulty of defining what females find attractive in male faces. Female sexual interest has a peak at ovulation, and preferences for men honestly advertising immunocompetence at this time may be adaptive. In spite of this, women are potentially sexually receptive across the menstrual cycle. Sex when conception is unlikely probably serves purposes not directly linked to fertilization (Hrdy, 1981). Our work, and the work of other researchers into the influence of face shape on personality, indicates that dominance and quality as a parent are attributions made to opposite ends of the continuum that relates to facial masculinity and femininity (or mature and babyfaces). Both masculinity and femininity may be associated with behaviors that have costs and benefits to reproductive success (Perrett et aL, 1998; Berry and Wero, 1993). A preference for males with a more masculine (i.e., high testosterone) appearance when conception is most likely may confer benefits for offspring in terms of heritable immunocompetence but also costs due to potentially decreased paternal investment. Femininity in male faces may be associated with the opposite collection of characteristics. In humans, paternity uncertainty results from internal fertilization, concealed ovulation, limited visual similarity between offspring and their fathers, and the apparently cross-cultural finding that couples prefer to copulate clandestinely (Schr6der, 1993; Pagel, 1997; Christenfeld and Hill, 1995). Rates of extra-pair paternity are certainly nonzero, although well-controlled studies are the exception rather than the rule (Macintyre and Sooman, 1991). As with some other species (Graves et aL, 1993; Andersson, 1994), human females may have been selected to pursue a mixed mating strategy in some circumstances. Adaptations allow female sexual behavior to serve multiple functions in addition to fertilization, and different male characteristics may indicate different potential benefits. Some females may choose a primary partner whose relatively low masculine appearance suggests cooperation in parental care, while occasionally pursuing extra-pair copulations with males with a relatively masculine appearance when conception is most likely. Potentially, sexual behavior arising from cyclic preferences may provide the benefits of polyandry (genetic diversity in offspring, good genes benefits in offspring, encouraging competition at the level of sperm) without losing

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

251

the advantages of ostensive monandry (such as extended paternal investment), thus realizing a reproductive advantage that may have been favored by selection. The behavioral adaptation that allows such mixed mating strategies is the female ability to copulate across the menstrual cycle, allowing most sexual encounters to serve purposes not directly involved in fertilization. For the majority of time, women appear to prefer smaller than average male facial secondary sexual characteristics. This is in sharp contrast to the many species in which female choice of male characteristics reveals preferences for exaggerated male traits (Andersson, 1994). Most species, however, copulate only when conception is likely during brief periods of female receptivity. In social species (such as many primates), with long-term, reciprocal relationships between individuals, females may have adapted to benefit from nonreproductive sex. Perhaps, then, we should not be surprised if the criteria of male attractiveness differ in these species. The results of the studies reported here have several caveats that can be addressed by future research. First, confirmation of an overall preference for femininity and cyclic shifts in such preferences using real rather than composite faces would add strength to both hypotheses. Facial-metric measurements of such stimuli may also isolate the features driving attractiveness judgments and allow multiple-motives models (e.g., Cunningham et aL, 1990) to be tested thoroughly. Cyclic shifts, coupled with context-dependent mate choice criteria, may explain the difficulty of defining male facial attractiveness and the inconsistencies in the current literature. Although the preferences we have described in this chapter and in earlier papers generalize to participants in the United Kingdom and Japan, further cross-cultural work is needed to investigate the possible consequences of local environmental conditions. Yu and Shepard (1998) indicate that male preferences for low female waist-to-hip ratios (e.g., Singh, 1993) are not universal, as previously thought. Similarly, overall preferences for feminized male faces may not be universal: in societies living under the pressure of a higher parasite load than modern United Kingdom and Japan, possible indicators of good genes (i.e., testosterone-dependent features) may be given more weight in attractiveness judgments (Gangestad and Buss, 1993). Pathogen prevalence appears to be one factor influencing human mating systems (Low, 1990). Fashion and media coverage also seem likely to influence judgments of male attractiveness in Japanese and British society: contemporary film stars, musicians and other male icons may be relatively feminine in comparison with past stars of stage and screen. However, in studies examining the preferences of a more diverse sample of women than undergraduate students, subject age has not influenced attractiveness judgments of male faces (e.g.,

252

IAN S. PENTON-VOAK AND DAVID I. PERRETT

Penton-Voak and Perrett, 2001). This indicates that even if older women once preferred the more rugged stars of their youth, the recent crop of boyish idols has influenced their current preferences. Additionally, it seems unlikely that cyclic preferences result from cultural factors. The possibility, however, that cultural fads have a radical impact on mate choice cannot be ignored. Further cross-cultural research may shed some light on such issues. Finally, and perhaps most importantly, further studies of actual partner choice must be conducted. Although the timing of extra-pair sex, and the characteristics of men chosen for such extra-pair sex, appear consistent with the hypotheses proposed here (Baker and Bellis, 1995; Gangestad and Thornhill, 1997), it is unclear whether the small shifts in preferences that we report here influence actual behavior, and hence have biological significance. Ethological studies are necessary to expand our knowledge of human sexual behavior:

V.

SUMMARY

On the evidence of experimental studies, female preferences for even static images of male faces represent a complex set of decision-making processes, and the differing techniques employed by different researchers often generate conflicting results. The findings reported by Perrett et al. (1998) indicate that women do not have clear preferences for masculinized (high testosterone) face shapes as predicted by indicator models of sexual selection, and some other studies of male faces (e.g., Grammer and Thornhill, 1994). Stereotypical personality judgments attributed to static faces appear to influence attractiveness judgments. Masculinized faces (indicating high levels of androgens) are considered to possess fewer desirable personality traits than feminized faces. These attributions may have some validity, and a reasonable (although, as yet unsupported) biological model linking androgen levels, behavior, and facial shape fits in with the observed preference pattern and, apparently, the fossil record: an overall preference for relatively low-testosterone men may be a somewhat unexpected adaptation. Biological facial characteristics that are considered putative indicators of good genes are not, however, ignored. They appear to be appraised in the light of stereotypical personality judgments and in the context of personal life-history factors (such as the type of relationship sought and possibly the relationship status of the woman). Furthermore, preferences are mediated and interact with cyclic hormonal changes linked to the likelihood of conception following sex. Relative masculinity in faces seems to be preferred at

SH/FTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

253

times when conception is most likely, paralleling other researchers' work in the olfactory modality. These preferences have been found in two cultures (United Kingdom and Japan), with student-aged and older women, and with multiple stimulus sets. Cyclic preferences for male face shapes are consistent with other reported cyclic preferences for male odor. If preference data are reliably linked to actual sexual behavior (a question that should be addressed by future research), a model linking likelihood of parental investment and facial masculinity preferred by females is suggested: when parental investment is sought (i.e., for a long-term relationship) facial shapes associated with relatively lower testosterone levels are preferred. Such faces may reliably indicate prosocial personality characteristics. When, however, likelihood of parental investment is low (in short-term relationships or possibly extra-pair copulations when the likelihood of conception is high) relatively more masculine faces are preferred, in a fashion more consistent with "good genes" hypotheses.

Acknowledgments IPV was supported by an ESRC Ph.D. studentship. This work was supported by project grants to DP from Unilever Research and the ESRC-ROPA. We are indebted to the help from D. Carrington at the BBC Tomorrow's World magazine, D. Castles at the Hasegawa Laboratory, University of Tokyo, and S. Yoshikawa at the Department of Cognitive Psychology in Education, University of Kyoto. We thank A. Whiten, R. Byrne, M. Goodale, J. Graves, R. Thornhill, and T. Roper for comments and suggestions on much of the work presented here.

References Adams, D., Burt, A., and Gold, A. R. (1978). Rise in female-initiated sexual activity at ovulation and its suppression by oral contraception. N. Engl. J. Med. 299, 1145-1150. Albright, L., Kenny, D. A., and Malloy, T. E. (1988). Consensus in personality judgments at zero acquaintance. J. Pers. Soc. Psychol. 55(3), 387-395. Albright, L., Malloy, T. E., Dong, Q., Kenny, D. A., Fang, X., Winquist, L., and Yu, D. (1997). Cross-cultural consensus in personality judgments. J. Pers. Soc. PsychoL 72(3), 558-569. Alexander, R. D., and Noonan, K. M. (1979). Concealment of ovulation, parental care and human social evolution. In "Evolutionary Biology and Human Social Behavior" (W. N. Chagnon & Irons, ed.), Duxbury Press, Massachusetts. Alley, T. R., and Cunningham, M. R. (1991). Average faces are attractive, but very attractive faces are not average. Psychol. Sci. 2(2), 123-125. Ambady, N., Hallahan, M., and Rosenthal, R. (1995). On judging and being judged accurately in zero-acquaintance situations. J. Pers. Soc. Psychol. 69(3), 518-529. Anderssou, M. (1994). "Sexual Selection." Princeton Univ. Press, New Jersey. Baker, R. R., and Bellis, M. A. (1995). "Human Sperm Competition: Copulation, Masturbation and Infidelity," 1st ed. Chapman & Hall, London.

254

IAN S. PENTON-VOAKAND DAVID I. PERRETF

Bancroft, J., Sanders, D., Davidson, D., and Warner, E (1983). Mood, sexuality, hormones, and the menstrual cycle. 3. Sexuality and the role of androgens. Psychosorn. Med. 45, 509516. .. Barrett, L. E, and Pietromonaco, E R. (1997). Accuracy of the five-factor model in predicting perceptions of daily social interactions. Pers. Soc. Psychol. Bull. 23, 1173-1187. Beach, E (1974). Human sexuality and evolution. In "Reproductive Behavior" (W. Montagna and W. Sadler, eds.) Plenum, New York. Bellis, M. A., and Baker, R, R. (1990). Do females promote sperm competition? Data for humans. Anim. Behav. 40, 997-999. Benshoof, L., and Thornhill, R. (1979). The evolution of monogamy and concealed ovulation in humans. J. Soe. Biol. Struct. 2, 95-106. Benson, E (1993). Perception and recognition of computer-enhanced facial attributes and abstracted prototypes. Ph.D. Thesis, University of St. Andrews. Benson, P., and Perrett, D. (1992). Face to face with the perfect image. N. Sei. 133, 32-35. Benton, D. (1982). The influence of androstenol--a putative human pheromone--on mood throughtout the menstrual cycle. Biol. Psyehol. 15, 249-256. Berry, D. (1991). Attractive faces are not all created equal: Joint effects of facial babyishness and attractiveness on social perception. Pers. Soc. Psychol. Bull. 17(5), 523-533. Berry, D. S., and Brownlow, S. (1989). Were the physiognomists right? Personality correlates of facial babyishness. Pets. Soe. Psychol. Bull. 15, 266-279. Berry, D. S., and Landry, J. C. (1997). Facial maturity and daily social interaction. J. Pers. Soe. Psychol. 72, 570-580. "Berry, D. S., and McArthur, L. Z. (1985). Some components and consequences of a babyface. J. Pets. Soe. PsyehoI. 48(2), 312=323. Berry, D. S., and McArthur, L. Z' (1988). What's in a face? Facial maturity and the attribution of legal resporlsibility. Pers. Soc. Psychol. Bull. 14(1), 23-33. Berry, D. S., and Wero, J. L. E (1993). Accuracy in face Perception: A view from ecological psychology. Ecol. Psychol. 497-519. Berry, D. S., and Zebrowitz MeArthur, L. (1986). Perceiving character in faces: The impact of age-related craniof~icial changes on social perception. Psyehol. Bull. 100, 3-18. Blaffer Hrdy, S., and Whitten, R (1987). Patterning of sexual activity. In "Primate Societies" (B. B. Smutts, D. L. Cheney, R. M. Seyfarth, R. Wrangham, and T. T. Strusaker, eds.), pp. 370-384. Univ. Chicago Press, Illinois. Bond, C. E, Berry, D. S., and Omar, A. (1994). The kernal of truth in judgements of deceptiveness. Basic AppL Soe. Psyehol. 15, 523-534. Booth, A., and Dabbs, J. (1993). Testosterone and men's marriages. Soc. Forces 72, 463-477. Borkenan, E, and Liebler, A. (1992). Trait inferences: Sources of validity at zero acquaintance. J. Pers. Soc. Psyehol. 62(4), 645~557. Botwin, M. D., Buss, D. M., and Shaekleford, T. K. (1997). Personality and mate preferences: Five factors in Mate selection and marital satisfaction. J. Pers. 65(1), 107-136. Brace, C. L., Rosenberg, K. R., and Hunt, K. D. (1987). Gradual change in human tooth size in the late pleistocene and post-pleistocene. Evolution 41, 705-720. Brandt, A. (1980). Face reading: The persistence of physiognomy. PsychoL Today (Dec.) 90-96. Burley, N. (1979). The evolution of concealed ovulation. Am. Nat. 114, 835-858. Burt, A. (1992). "Concealed ovulation" and sexual signals in primates. Folia Primatol. 58, 1-6. Butt, D. M., and Perrett, D. I. (1995). Perception of age in adult Caucasian male faces: Computer graphic manipulation of shape and colour,information. Proc. R. Soc. London B 259, 137-143. Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behav. Brain Sci. 12, 1-49.

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

255

Buss, D. M., and Barnes, M. (1986). Preferences in human mate selection. J. Pers. Soc. Psychol. 50, 559-570. Christenfeld, N., and Hill, E. A, (1995). Whose baby are you? Nature 378, 669. Cleeton, G. U., and Knight, E B. (1924). Validity of character judgement based on external criteria. J. AppL Psychol. 8, 215-231. Cunningham, M. R. (1986). Measuring the physical in physical attractiveness: Quasiexperiments on the sociobiology of female facial beauty. J. Pers. Soc. PsychoL 50(5), 925935. Cunningham, M. R., Barbee, A. R, and Pike, C. L. (1990). What do women want? Facialmetric assessment of multiple motives in the perception of male facial physical attractiveness. J. Pers. Soc. Psychol. 59(1), 61-72. Daly, M., and Wilson, M. (1983). "Sex, Evolution and Behavior" 2nd ed. PWS, Boston. Daly, M., and Wilson, M. I. (1999). Human evolutionary psychology and animal behaviour. Anita. Behav. 57, 509-519. Dawkins, R. (1976). "The Selfish Gene." Oxford Univ. Press, UK. Dixson, A. E (1983). Observations on the evolution and behavioral significance of"sexual skin" in female primates. Adv. Study Behav. 13, 63-106. Dory, R. L., Snyder, P. J., Huggins, G. R., and Lowry, L. D. (1981). Endocrine, cardiovascular, and psychological correlates of olfactory sensitivity changes during the human menstrual cycle. J. Comp. Physiol. Psychol. 95, 45-60. Folstad, I., and Karter, A. J. (1992). Parasites, bright males and the immunocompetence handicap. Am. Nat. 139, 603-622. Ford, C. S., and Beach, E A. (1951). "Patterns of Sexual Behavior." Eyre and Spottiswood, London. Frost, R (1994). Preference for darker faces in photographs at different phases of the menstrual cycle: Prefiminary assessment of evidence for a hormonal relationship. Percept. Motor Skills 79, 507-514. Gage, A. R., Manning, J. T., Scutt, D., Diver, M. J., and Fraser, W. D. (1999). Testosterone, cortisol and jaw size in men. Galton, E J. (1878). Composite Portraits. Anthropological Institute of Great Britain and Ireland. 8, 132-142. Gangestad, S. W., and Buss, D. M. (1993). Pathogen prevalence and human mate preferences. Ethol. Sociobiol. 14, 89-96. Gangestad, S. W., and Thornhill, R. (1997). The evolutionary psychology of extrapair sex: The role of fluctuating asymmetry. Evol. Hum. Behav. 18(2), 69-88. Gangestad, S. W., and Thornhill, R. (1998). Menstrual Cycle Variation in Women's Preferences for the scent of symmetrical men. Proc. R. Soc. London B 265, 927-933. Grammer, K. (1993). 5-alpha-androst-16en-3alpha-on: A male pheromone? A brief report. Ethol. Sociobiol. 14, 201-208. Grammer, K., and Thornhill, R. (1994). Human (Homo sapiens) facial attractiveness and sexual selection: The role of symmetry and averageness. J. Comp. Psychol. 108, 233-242. Grammer, K., Jutte, A., and Fischmann, B. (1997). Der Kampf der Geschlechter und der Krieg der Signale. In "Liebe, Lust und Leidenschaft. Sexualitat im Spiegal der Wissenschaft" (B. Kanitscheider, ed.), Hirzel. Stuttgart. Graves, J., Ortega-Ruano, J., and Slater, P. J. B. (1993). Extra-pair copulations and.paternity in shags: Do females choose better males? Proc. R. Soc. London B 253, 3-7. Gray, J. P., and Wolfe, L. D. (1983). Human female sexual cycles and the concealment of ovulation problem. J. Soc. BioL Struct. 6, 345-352. Haaq, C. M., and Donahoe, P. K. (1998). Regulation of Sexual Dimorphism in Mammals. Physiol. Rev. 78(1), 1-33.

256

IAN S. PENTON-VOAK AND DAVID I. PERRETr

Harcourt, A. H. (1996). Sexual selection and sperm competition in primates: What are the male genitalia good for? EvoL Anthropol. 4, 121-1.29. Hansler, M., and Schmid, E (1995). Comparison of the pelves of Sts 14 and AL 288-1: Implications for birth and sexual dimorphism in australopithecines. J. Hum. EvoL 29, 363-383. Herz, R. S., and Cahill, E. D. (1997). Differential use of sensory information in sexual behavior as a function of gender. Hum. Nat. 8, 275-286. Hillgarth, N., and Wingfield, J. C. (1997). Testosterone and immunosuppression in vertebrates: Implications for parasite mediated sexual selection. In "Parasites and Pathogens" (N. E. Beckage, ed.). Chapman and Hall, New York. Hrdy, S. B. (1981). "The Woman That Never Evolved." Harvard Univ. Press, Cambridge, MA. Hume, D. K., and Montgomerie, R. (1999). Facialattractiveness and symmetry in humans. Paper presented at the Human Behavior and Evolution Society, Utah. J6chle, W. (1973). Coitus-induced ovulation. Contraception 7(6), 523-564. Johnston, V. S., and Franklin, M. (1993). Is beauty in the eye of the beholder? Ethol. Sociobiol, 14, 183-199. Jones, D. (1995). Sexual selection, Physical attractiveness, and facial neoteny. Curr. Anthropol. 36(5), 723-748. Jones, D., and Hill, K. (1993). Criteria of facial attractiveness in five populations. Hum. Nat. 4(3), 271-296. Kalick, S. M., Zebrowitz, L. A., Langlois, J. H., and Johnson, R. M. (1998). Does human facial attractiveness honestly advertise health? Longitudinal data on an evolutionary question. PsychoI. Sci. 9(1), 8-13. Keating, C. E (1985). Gender and the physiognomy of dominance and attractiveness. Soc. Psychol. Q. 48, 61-70. Kenny, D. A., Homer, C., Kashy, D. A., and Chu, L. (1992). Consensus at zero acquaintance: Replication, behavioral cues and stability. J. Pers. Soc. Psychol. 62, 88-97. Kenny, D. A., Albright, L., MaUoy, T. E., and Kashy, D. A. (1994). Consensus in interpersonal perception: Acquaintance and the big five. Psychol. Bull, 116, 245-258. Keverne, E. B. (1976). Sexual receptivity and attractiveness in the female rhesus monkey. Adv. Study Behav. 7, 155-200. Kirk-Smith, M., Booth, D., Carroll, D., and Davies, P. (1978). Human social attitudes affected by androstenol. Res. Commun. Psychol. Psychiatr. Behav. 3, 379-384. Koeslag, J., and Koeslag, P. (1994). Koinophilia. J. Theor. Biol. 167, 55-65. Koeslag, J. H. (1994). Koinophilia replaces random mating in populations subject to mutations with randomly varying fitnesses. J. Theor. BioL 171, 341-345. Kowner, R. (1996). Facial asymmetry and attractiveness judgment in developmental perspective. J. Exp. Psychol.: Hum. Percept. Perform. 22(3), 662-675. Lancaster, J. (1975). "Primate Behavior and the Emergence of Human Culture." Holt, Reinhart, and Winston, New York. Langlois, J. H., and Roggman, L. A. (1990). Attractive faces are only average. PsychoL Sci. 1(1), 115-121. Langlois, J. H., Roggman, L. A., and Musselman, L. (1994). What is average and what is not average about attractive faces? PsychoL Sci. 5, 214-219. Liggett, J. (1974). "The Human Face." Constable, London. Little, A. C. (1999). The influence of facial masculinity and distinctiveness on judgements of attractiveness in human male faces. M. S. Thesis, Department of Psychology. Stifling University. Lorenz, K. (1943). The innate forms of potential experience. Z. Tierpsychol. 5, 234-409. Low, B. S. (1990). Marriage systems and pathogen stress in human societies. Am. ZooL 30, 325-339.

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

257

Macintyre, S., and Sooman, A. (1991). Non-paternity and prenatal genetic screening. Lancet 338, 869. Manning, J. T., Scutt, D., Whitehouse, G. H., and Leinster, S. J. (1997). Breast asymmetry and phenotypic quality in women. Evol. Hum. Behav. 18, 223-236. Matteo, S., and Rissman, E. F. (1984). Increased sexual activity during the midcycle portion of the human menstrual cycle. Horm. Behav. 18, 249-255. Mazur, A., and Booth, A. (1998). Testosterone and dominance in men. Behav. Brain Sci. 21, 353-371. Mazur, A., Halpern, C., and Udry, J. R. (1994). Dominant looking males copulate earlier. Ethol. Sociobiol. 15, 87-94. Mazur, A., Susman, E. J., and Edelbrock, S. (1997). Sex difference in testosterone response to a video game contest. Evol. Hum. Behav. 18, 317-326. McArthur, L., and Apatow, K. (1983-1984). Impressions of baby-faced adults. Soc. Cog. 2(4), 315-342. McArthur, L. Z., and Berry, D. S. (1987). Cross-cultural agreement in perceptions of babyfaced adults. J. Cross-cultural Psychol. 18(2), 165-192. McCrae, R. R., and Costa, P. T. (1987). Validation of the Five Factor model of personality across instruments and observers. J. Pers. Soc. Psychol. 52(1), 81-90. McHenry, H. M. (1994). Behavioral ecological implications of early hominid body size. J. Hum. Evol. 27, 77-87. Mealey, L., Bridgestock, R., and Townsend, G. (1999). Symmetry and perceived facial attractiveness. J. Pers. Soc. Psychol. 76, 151-158. Metcalf, M. G., Skidmore, G. F., Lowry, G. F., and Mackenzie, J. A. (1983). Incidence of ovulation in the years after the menarche. J. Endocrinol. 97, 213-219. Meyer, D. A., and Quong, M. W. (1999). The bio-logic of facial geometry. Nature 397, 661~562. MOiler, A. P. (1997). Developmental stability and fitness: A review. Am. Nat. 149, 916942. MOiler, A. E, and Hoglund, J. (1991). Patterns of fluctuating asymmetry in avian feather ornaments: Implications for models of sexual selection. Proc. R. Soc. London B 245, 1-5. M011er, A. P., and Thornhill, R. (1997). A meta-analysis of the heritability of developmental stability. J. Evol. Biol. 10, 1-16. MOiler, A. P., and Thornhill, R. (1998). Bilateral Symmetry and Sexual Selection: A metaanalysis. Am. Nat. 151, 174-192. Morris, D. (1994). "The Naked Ape." Vintage. Mueller, U., and Mazur, A. (1997). Facial dominance in Homo sapiens as honest signaling of male quality. Behav. Ecol. 8(5), 569-579. Murdoek, G. P. (1967). "Ethnographic Atlas." Univ. of Pittsburgh Press, Pennsylvania. Norman, W. T. (1963). Toward an adequate taxonomy of personality attributes: Replicated factor structure in peer nominated personality ratings. J. Abnorm. Soc. Psychol. 66, 574583. Owens, I. P. F., and Short, R. V. (1995). Hormonal basis of sexual dimorphism in birds: Implications for new theories of sexual selection. Trends Ecol. Evol. 10(1), 44-47. Pagel, M. (1997). Desperately concealing father: A theory of parent-infant resemblance. Anim. Behav. 53, 973-981. Passini, E T., and Norman, W. T. (1966). A universal conception of personality structure? J. Pers. Soc. Psychol. 4, 44-49. Penton-Voak, I. S., and Perrett, D. I. (2000). Female preference for male faces changes cyclically-- Further evidence. Evol. Hum. Behav. 21, 39--48. Penton-Voak, I., Perrett, D., and Pierce, J. (1999a). Computer graphic studies of the role of facial similarity in attractiveness judgements. Curr. Psychol. 18, 104-117.

258

IAN S. PENTON-VOAK AND DAVID I. PERRETT

Penton-Voak, I. S., Perrett, D., Castles, D., Burt, M., Koyabashi, T., and Murray, L. K. (1999b). Female preferences for male faces change cyclically. Nature 399, 741-742. Perrett, D., and Penton-Voak, I. (t999). The Bio-logic of facial geometry. Nature 397, 662. Perrett, D. I., May, K. A., and Yoshlkawa, S. (1994). Facial shape and judgements of female attractiveness. Nature 368, 239-242. Perrett, D. I., Lee, K. J., Penton-Voak, I. S., Rowland, D. R., Yoshikawa, S., Burt, D. M., Henzi, S. E, Castles, D. L., and Akamatsu, S. (1998). Effects of sexual dimorphism on facial attractiveness. Nature 394, 884-887. Perrett, D. I., Butt, D. M., Penton-Voak, I. S., Lee, K. J., Rowland, D. A., and Edwards, R. (1999). Symmetry and Human Facial Attractiveness. EvoL Hum. Behav. 20, 295-308. Perusse, D. (1993). Cultural and reproductive success in industrial societies: Testing the relationship at proximate and ultimate levels. Behav. Brain Sci. 16, 267-322. Plavcan, J. M., and van Schaik, C. E (1997). Interpreting hominid behavior on the basis of sexual dimorphism. J. Hum. Evol. 32, 345-374. Regan, E C. (1996). Rhythms of desire: The association between menstrual cycle phases and female sexual desire. Can. J. Hum. Sex. 5(3), 145-156. Rhodes, G., and Tremewan, T. (1996). Averageness, exaggeration and facial attractiveness. Psychol. Sci. 7(2), 105-110. Rhodes, G., Proffitt, E, Grady, J., and Sumich, A. (1998). Facial Symmetry and the perception of beauty. Psychonom. Bull. Rev. 5, 659-669. Rhodes, G., Sumich, A., and Byatt, G. (1999). Are average facial configurations attractive only because of their symmetry? Psychol. Sci. 10, 52-58. Rhodes, G., Byatt, G., Hiekford, C., and Jeffery, L. (2001). Sex-typicality and facial attractiveness: Are supermale and superfemale faces super-attractive? Br. J. PsychoL Rikowski, A., and Grammer, K. (1999). Human body odour, symmetry and attractiveness. proe. R. Soc. London B 266, 869-874. Rowell, T. E. (1972). Female reproduction cycles and social behavior in primates. Adv. Study Behav. 4, 69-105. Ruff, C. B., Trinkans, E., Walker, A., and Larsen, C. S. (1993). Postcranial robusticity in Homo. 1: Temporal trends and mechanical interpretation. Am. J. Phys. Anthropol. 91, 21-53. Samuels, C. A., Butterworth, G., Roberts, T., Graupner, L., and Hole, G. (1994). Facial aesthetics: Babies prefer attractiveness to symmetry. Perception 23, 823-831. Scheib, J. E., Gangestad, S. W., and Thornhill, R. (1999). Facial attractiveness, symmetry, and cues to good genes. Proc. R. Soc. London B. 266, 1913-1917. Schr6der, I. (1993). Concealed ovulation and clandestine copulation: A female contribution to human evoluton. Ethol. SoeiobioL 14, 381-389. Silber, M. (1994). Menstrual cycle and work schedule: Effects of women's sexuality. Arch. Sex. Behav. 23, 397-404. Silldn-Tulberg, B., and M¢ller, A. R (1993). The relationship between concealed ovulation and mating systems in anthropoid primates: A phylogenetic analysis. Am. Nat. 141(1), 1-25. Singh, D. (1993). Adaptive significance of,female physical attractiveness: Role of waist-to-hip ratio. J. pers. Soc. PsychoL 65(2), 293-307. Stanislaw, H., and Rice, E J. (1988). Correlation between sexual desire and menstrual cycle characteristics. Arch. Sex. Behav. 17, 499-508. Steklis, H. D., and Whiteman, C. H. (1989). Loss of estrns in human evolution: Too many answers, too few questions. EthoL SociobioL 10, 417-434. Stewart, D. E. (1989). Positive changes in the pre-menstrual period. Acta Psychiatr. Scand. 79, 400-405. Strassman, B. I. (1981). Sexual selection, paternal care, and concealed ovulation in humans. Ethol. SoeiobioL 2, 31-40.

SHIFTING FEMALE PREFERENCES ACROSS THE MENSTRUAL CYCLE

259

Swaddle, J. E, and Cuthill, I. C. (1995). Asymmetry and human facial attractiveness: Symmetry may not always be beautiful. Proc. R. Soc. London 261, 111-116. Symons, D. (1979). "The evolution of human sexuality." Oxford Univ. Press, UK. Symons, D. (1994). Beauty is in the adaptions of the beholder: The evolutionary psychology of human female sexual attractiveness. In "Sexual Nature/Sexual Culture" (P. R. R Abramson, S. D., ed.). Univ. of Chicago Press, Illinois. Tague, R. G., and Lovejoy, C. O. (1998). AL 288-1--Lucy or Lucifer: Gender confusion in the Pliocene. J. Hum. Evol. 35, 75-94. Thornhill, R., and Gangestad, S. W. (1993). Human facial beauty: Averageness, symmetry, and parasite resistance. Hum. Nat. 4(3), 237-269. Thornhill, R., and Gangestad, S. W. (1994). Human fluctuating asymmetry and sexual behavior. Psychol. Sci. 5, 297-302. Thornhill, R., and Gangestad, S. W. (1996). The evolution of human sexuality. Trends Ecol. Evol. 11(2), 98-102. Thornhill, R., and Gangestad, S. W. (1999). The scent of symmetry: A human sex pheromone that signals fitness? Evol. Hum. Behav. 20, 175-202. Thornhill, R., and Gangestad, S. W. (1999). Facial Attractiveness. Trends Cog. Sci. 3, 452-460. Townsend, J. M., and Roberts, L. W. (1993). Gender differences in mate preference among taw students: Divergence and convergence of criteria. J. Psychol. 127(5), 507-528. Townsend, J. M., and Wasserman, T. (1998). Sexual attractiveness: Sex differences in assessment and criteria. Evol. Human. Behav. 19, 171-191. Townsend, J. M., Kline, J., and Wasserman, T. H. (1995). Low investment copulation: Sex differences in motivations and emotional reactions. Ethol. Soeiobiol. 16, 25-51. Udry, J. R., and Morris, N. (1968). Distribution of coitus in the menstrual cycle. Nature 200, 593-596. Urdy, R. J., and Morris, N. M. (1977). The distribution of events in the human menstrual cycle. J. Reprod. Fert. 51, 419-425. Van Den Berghe, E L., and Frost, E (1986). Skin color preference, sexual dimorphism and sexual selection: A case of gene culture co-evolution? Ethn. Rac. Studies 9(1), 87-113. Walster, E., Aronson, V., Abrahams, D., and Rottman, L. (1966). Importance of physical attractiveness in dating behavior. J. Pers. Soc. Psychol. 4(5), 508-516. Watson, D. (1989). Strangers' rating of the five robust personality factors: Evidence of a surprising convergence with seif-report. J. Pets. Soc. Psychol. 57, 120-128. Wedekind, C., and Furi, S. (1997). Body odour preferences in men and women: Do they aim for specific MHC combinations or simply heterozygosity? Proc. R. Soc. London B 264, 1471-1479. Wedekind, C., Seebeck, T., Bettens, E, and Paepke, A. J. (1995). MHC-dependent mate preferences in humans. Proc. R. Soc. London B 260, 245-249. Wiederman, M. W., and Dubois, S. L. (1998). Evolution and sex differences in preferences for short-term mates: Results from a plociy capturing study. Evol. Hum. Behav. 19, 153-170. Worthman, C. (1978). Psychoendocrine study of human behavior: some interactions of steroid hormones with affect and behavior in !Kung San. Ph.D. Thesis, Harvard University. Yu, D. W., and Shepard, G. H. (1998). Is beauty in the eye of the beholder? Nature 396, 321-322. Zahavi, A. (1975). Mate selection: A selection for a handicap. Z Theor. Biol. 53, 205-214.