Facial adiposity: A cue to health?

Perception, 2009, volume 38, pages 1700 ^ 1711

doi:10.1068/p6423

Facial adiposity: A cue to health?

Vinet Coetzee, David I Perrett, Ian D Stephen

Perception Laboratory, School of Psychology, University of St Andrews, St Mary's Quad, South Street, St Andrews, Fife KY16 9JP, Scotland, UK; e-mail: [email protected], [email protected] Received 27 March 2009, in revised form 19 May 2009; published online 28 October 2009

Abstract. Facial symmetry, averageness, sexual dimorphism, and skin colour/texture all serve as cues to attractiveness, but their role in the perception of health is less clear. This ambiguity could reflect the fact that these facial traits are not the only cues to health. We propose that adiposity is an important, but thus far disregarded, facial cue to health. Our results demonstrate two important prerequisites for any health cue. First, we show that facial adiposity, or the perception of weight in the face, significantly predicts perceived health and attractiveness. Second, we show that perceived facial adiposity is significantly associated with measures of cardiovascular health and reported infections. Perceived facial adiposity, or a correlate thereof, is therefore an important and valid cue to health that should be included in future studies.

1 Introduction Healthy individuals confer direct and indirect benefits on their partners. Direct benefits include reduced risk of infection and increased resources (Kirkpatrick and Ryan 1991), while indirect benefits consist of the `good genes' passed on to the offspring. The `good genes' hypothesis posits that individuals are attracted to partners that display certain traits. These traits indicate a superior ability to survive, for instance a higher resistance to pathogens (Hamilton and Zuk 1982). Attractive individuals are therefore expected to be healthier. Various studies support the relationship between facial attractiveness and health, while some do not. Kalick et al (1998) found no significant correlation between late adolescent facial attractiveness and health scores based on detailed medical histories. A re-analysis of their data did show a significant relationship between adult facial attractiveness and adult health in both sexes, but only when using faces below the median for attractiveness (Zebrowitz and Rhodes 2004). Late adolescent facial attractiveness is also linked to longevity (Henderson and Anglin 2003) and facially attractive young adults have more heterozygous human leukocyte antigen (HLA) genes (Roberts et al 2005). These genes play a crucial role in the immune system and heterozygosity is thought to be associated with increased immune recognition of pathogens (Roberts et al 2005). Thornhill and Gangestad (2006) did not find a significant association between facial attractiveness and the number and duration of respiratory and stomach infections, or the use of antibiotics in the last three years. Studies often find a difference between the sexes. Shackelford and Larsen (1999) showed that facial attractiveness significantly correlates with cardiovascular recovery time after exercise in men, but not women, and some common physical illness symptoms (eg runny nose, nausea, backache, etc) in both sexes. Hume and Montgomerie (2001) also found a difference between the sexes, with attractive women reporting less severe diseases during their lifetime. No significant association was observed for the men. Facial attractiveness is also positively linked to reproductive health. Facially attractive men have better sperm quality (as assessed by morphology and motility) than less attractive men (Soler et al 2003, but see Peters et al 2008), while facially attractive women have higher late follicular oestrogen levels than less attractive women (Law Smith et al 2006). On the whole, most studies show a relationship between facial attractiveness and health, although the relationship is far from consistent.

Facial adiposity

1701

But which facial cues do people use to judge health and attractiveness? Past research identified four cues to attractiveness: symmetry (Grammer and Thornhill 1994; Penton-Voak et al 2001; Rhodes et al 2001), averageness (Rhodes et al 2007), sexual dimorphism (Perrett et al 1998; Penton-Voak et al 2001) and, more recently, skin colour/texture (Fink et al 2001; Jones et al 2004; Matts et al 2006; Stephen et al 2009). The link between these traits and health has been mixed. Feminine female faces are judged healthier (Rhodes et al 2003, 2007; Law Smith et al 2006), but femininity is not consistently linked to actual health in women. Thornhill and Gangestad (2006) found a significant association between femininity and number and duration of respiratory infections, but an earlier study by Rhodes et al (2003) did not show a significant association between femininity and health ratings calculated from detailed medical histories. As a rule, symmetrical faces are judged healthier (Grammer and Thornhill 1994; Jones et al 2001; Penton-Voak et al 2001; Rhodes et al 2001, 2007; Fink et al 2006). Paradoxically, actual health is only weakly associated with symmetry. Two studies reported some evidence for a relationship between measured facial symmetry and self-reported health measures (Shackelford and Larsen 1999; Thornhill and Gangestad 2006), while Rhodes et al (2001) did not find a significant association between symmetry and medically assessed lifetime health data. Averageness is fairly consistently linked to perceived health (Rhodes et al 2001, 2007). But, Grammer and Thornhill (1994) showed a significant relationship between averageness and perceived health only in men. The relationship was not significant in women's faces. Averageness is weakly associated with actual health. Rhodes et al (2001) showed that rated averageness in young adults was linked to childhood health in men, and adolescent and current health in women. Faces below median averageness mainly drive this association (Zebrowitz and Rhodes 2004). Masculinity shows a very different association with health than the other cues. Unlike the other cues, masculinity in males is linked to actual health, but not reliably to perceived health. An initial study by Rhodes et al (2003) found that male faces that are rated more masculine are also rated healthier. This association seems to apply only to natural faces and not to faces manipulated by computer graphics (Rhodes et al 2007); although Boothroyd et al (2007) did not find any clear association between masculinity and perceived health in a series of experiments. The relationship between masculinity and actual health is less ambiguous. Male face masculinity is associated with better health in puberty and adolescence (Rhodes et al 2003), use of less antibiotics, and fewer and shorter respiratory infections (Thornhill and Gangestad 2006). Colour and texture seem to be related to perceived health (Jones et al 2004; Fink et al 2006; Matts et al 2006, 2007; Stephen et al 2009) but, to our knowledge, their direct relationship to actual health has not yet been considered. Still, there is some indirect evidence for the association between colour and texture cues, actual and perceived health. Roberts et al (2005) found a significant association between HLA heterozygosity and perceived health using skin patches, where the only information available to the observer is presumably colour and texture cues. This lack of congruence in the interrelation between facial cues and health might be based on our inability to measure health, or even facial cues, accurately. Then again, we might be overlooking other important cues to health. If multiple facial cues to health and attractiveness exist, and these cues vary independently, variance in an unacknowledged cue may disrupt the relationship between other cues and health. We propose that facial adiposity, or the perception of weight in the face, could be a valid cue to health. Body mass index (BMI), or weight scaled for height, plays an important role in judgments of bodily attractiveness and health. In Western populations, obese, and to a lesser extent overweight, female bodies are judged less attractive

1702

V Coetzee, D I Perrett, I D Stephen

(Thornhill and Grammer 1999; Tove¨e et al 1998, 1999; Swami and Tove¨e 2005a) and less healthy than normal-weight bodies (BMI 18.5 ^ 25; Furnham et al 2006; Swami et al 2008). Obese (and overweight) women are also judged facially less attractive than lower-weight women (Hume and Montgomerie 2001). Preferences do change with culture and, in some non-Western (especially African) cultures, high-weight female bodies are considered attractive and fertile (Furnham and Baguma 1994; Yu and Shepard 1999; Marlowe and Wetsman 2001; Furnham et al 2002); for example, in one rural SouthAfrican Zulu population individuals with a BMI of 26.52 were considered optimally attractive (Tove¨e et al 2006). A BMI of 26.52 falls within the overweight category according to the WHO classification (WHO 2000). Both Western and non-Western observers judge underweight female bodies less attractive than normal-weight bodies (Furnham and Baguma 1994; Tove¨e et al 1998, 1999, 2006; Yu and Shepard 1999). Obese and overweight individuals are at increased risk of developing coronary artery disease, diabetes mellitus, stroke, gallbladder disease, gout, osteoarthritis, sleep apnoea, respiratory problems, and a variety of cancers (Pi-Sunyer 1993; Manson et al 1995; Must et al 1999; Brown et al 2000; Wilson et al 2002; Mokdad et al 2003). Underweight individuals, in turn, have decreased immunity (Ritz and Gardner 2006), decreased vitality, poorer mental health, increased tiredness, and show increased use of health services (Brown et al 2000) and increased mortality due to all causes compared to normal-weight individuals (Flegal et al 2005). In this study, we particularly wanted to focus on two main groups of health measures that have been related to weight: infections and cardiovascular health. Obesity is associated with impaired T and B cell function, an indication of immune dysfunction (Tanaka et al 1993), which may explain why obese surgical patients develop more post-surgical infections than non-obese patients (Choban et al 1995; Vilar-Compte et al 2000). This proneness to infection is also observed in community settings, where heavier adult women (Baik et al 2000) and children (Jedrychowski et al 1998) have a higher susceptibility to respiratory infections. Underweight individuals often have an energy deficiency due to malnutrition or under-nutrition. Subsequently, fewer resources can be allocated to the immune function, causing underweight individuals to be more prone to infection (Ritz and Gardner 2006). We therefore predict a curvilinear relationship between weight and infection, with both the heavy and underweight individuals more prone to infection than intermediate-weight individuals. Various studies show a strong relationship between excess weight and cardiovascular health. Obese and overweight adults are at increased risk of hypertension and cardiovascular disease (Hubert et al 1983; Manson et al 1995; Lusky et al 1996; Wilson et al 2002). One might argue that overweight and obese children and adolescents are not at risk of cardiovascular disease, but obese children are more likely to become obese adults (Serdula et al 1993) and childhood obesity is associated with increased adult mortality due to cardiovascular disease (Gunnell et al 1998). Underweight individuals tend to have an even lower prevalence of hypertension than normalweight individuals (Lusky et al 1996). We therefore predict a linear relationship between cardiovascular measures and weight, with heavier individuals more prone to high blood pressure than intermediate-weight and low-weight individuals. We propose that facial adiposity, or the perception of weight in the face is a valid cue to health. In order to be a valid cue, perceived facial adiposity must: (i) be used in the perception of health, and (ii) relate to actual health measures. Our aims were to test these hypotheses. First, we tested whether perceived facial adiposity is significantly associated with perceived facial health and attractiveness. Second, we tested whether perceived facial adiposity is correlated with infections and cardiovascular health. We also investigated how BMI relates to these groups of health measures in the study population.

Facial adiposity

1703

2 Methods This work was approved by the University of St Andrews Ethics Committee (Approval code: PS3137). We recruited eighty-four Caucasian participants (forty-three female; forty-one male) from the University of St Andrews (mean age ˆ 21.13 years, range ˆ 18 ^ 27 years; mean BMI ˆ 22.95, range ˆ 17.82 ^ 33.38). All participants were photographed in front of a uniform Munsell 5 background, in full colour and under standard lighting conditions. Participants were seated a set distance from the camera, asked to maintain a neutral expression, and had their hair pulled back. Each participant gave informed consent to take part in this study, was asked to complete a questionnaire, and had his/her blood pressure, body weight, and height measured. Questionnaires contained questions on gender, parental income, respiratory diseases, and use of antibiotics (appendix A). Systolic and diastolic blood pressures were measured twice with a Boots automatic blood-pressure arm monitor (Boots, England) after a minimum initial rest period of 5 min. Weight and height measures were used to calculate BMI [(weight in kilograms)/ (height in metres)2 ] and BMI categories were assigned according to WHO criteria (WHO 2000). Skewness and kurtosis values were low for all measures (ÿ1:0 4 skewness and kurtosis 5 1:02), except BMI (kurtosis 1.43), cold and flu bouts per year (skewness 1.87, kurtosis 4.25), average bout length (skewness 1.30) and use of antibiotics (skewness 1.94, kurtosis 6.23). BMI, cold and flu bouts per year, and average bout length were log-transformed, successfully normalising the data (ÿ0:94 4 skewness and kurtosis 5 0:47). Log-transformation did not sufficiently normalise the antibiotics-use data (skewness ˆ 2:13, kurtosis ˆ 3:64), because very few individuals reported using more than one course of antibiotics in the last year. We therefore grouped the antibioticsuse data into three groups: high use (twice or more in the last year; N ˆ 8), low use (once in the last year; N ˆ 34) and zero use (zero use in the last year; N ˆ 13). Antibiotics use was subsequently analysed with a Krustall ^ Wallis test followed by Mann ^ Whitney tests to compare the three groups. All other relationships between health measures, facial adiposity, and BMI were tested with Pearson's correlations. Data were missing for BMI (one participant), parental income (two participants), use of antibiotics (twentynine participants), and blood pressure (one participant). Images were resized (female images: 3876478 pixels; male images: 3896518 pixels), colour-corrected (DE ˆ 2:44) with in-house software, and standardised for inter-pupillary distance and position with PsychoMorph version 8.4.7.0. We recruited four groups of participants to rate the facial images for health, attractiveness, and weight. First, twenty-six Caucasian participants (twelve female, fourteen male; mean age ˆ 22.81 years, range ˆ 18 ^ 28 years) rated each female image for health and attractiveness on a 7-point Likert scale (0 ˆ very unhealthy/unattractive; 3 ˆ average; 6 ˆ very healthy/attractive). Second, twenty-two Caucasian participants (twelve female, ten male; mean age ˆ 21.87 years, range ˆ 18 ^ 26 years) rated each male image for health and attractiveness on the same scale. Third, twenty-six Caucasian participants (fourteen female, twelve male; mean age ˆ 20.81 years, range ˆ 20 ^ 28 years) rated each female facial image for weight on a 7-point Likert scale (0 ˆ very underweight; 3 ˆ average weight; 6 ˆ very overweight). Last, twenty-nine Caucasian participants (seventeen female, twelve male; mean age ˆ 21.1 years, range ˆ 19 ^ 26 years) rated each male facial image for weight on the same scale. In all four studies, participants were shown all the images before rating commenced to make them aware of the range and variability of the images. Images were presented in a randomised order and participants were asked to indicate whether they knew the person. We recorded the time it took the participants to rate each image and excluded all participants with an average time of less than 1.65 s per question, for two or more

1704

V Coetzee, D I Perrett, I D Stephen

images (group 1: five participants; group 2: one participant; group 3: three participants; group 4: four participants). The threshold value was defined by the maximum time it took the experimenter to select random answers as quickly as possible and included submission time. We also removed rating data if the participant knew the rated individual (group 1: 9.6% of ratings; group 2: 5.9% of ratings; group 3: 5.8% of ratings; group 4: 4.8% of ratings). Facial health ( p ˆ 0:62), attractiveness ( p ˆ 0:75) and weight ratings ( p ˆ 0:22) did not differ significantly between male and female raters (independentsamples t-test). Data from both sexes were therefore combined for analysis. Inter-rater reliability was very high for facial health (Cronbach a ˆ 0:87), attractiveness (a ˆ 0:92), and weight (a ˆ 0:84). Given the consistency of ratings, the scores were averaged across participants for each of the 84 images. Skewness and kurtosis values were low for all three measures (ÿ0:86 4 skewness and kurtosis 5 0.35). 3 Results 3.1 Perceived facial adiposity and BMI Despite the fact that a significant relationship between perceived facial adiposity and BMI might seem evident, we wanted to examine the strength of this relationship. To do so we fitted a general linear model (GLM) in SPSS version 16 to test if participant's judgment of weight using facial images related to measured body weight. BMI significantly and positively predicted perceived facial adiposity (perceived facial adiposity ˆ ÿ10:72 ‡ 10:08 BMI; F1, 81 ˆ 61:217, p 5 0:0005, R 2 ˆ 0:430, figure 1). 6

Facial adiposity

5 4 3 2 1 0

15

20

25 BMI

30

35

Figure 1. Interrelation between facial adiposity and body mass index (BMI) (perceived facial adiposity ˆ ÿ10:72 ‡ 10:08 BMI; p 5 0:005, R 2 ˆ 0:430). Heavier individuals are consistently judged to have a higher facial adiposity throughout the BMI spectrum: underweight (solid circles), normal weight (open circles), overweight (solid triangles), and obese (open triangles). The continuous line gives the best-fit general linear model.

3.2 Perceived facial adiposity, health, and attractiveness We fitted a multiple polynomial GLM to test the relationship between judgments of weight, health, and attractiveness. Second-order equations were included, since both underweight and obese individuals should be rated less healthy or attractive. The following polynomial model was fitted for both judgments: y ˆ a ‡ b1 x ‡ b2 x 2 ‡ e , where y is the health or attractiveness rating, a is the intercept, b1 and b2 are coefficients, x is perceived facial adiposity, and e is the random error. The quadratic model significantly predicted perceived health [perceived health ˆ ÿ0:33 ‡ 2:45 (perceived facial adiposity) ÿ 0:41 (perceived facial adiposity)2; F2, 81 ˆ 14:457, p 5 0:0005, R 2 ˆ 0:263; figure 2] and perceived attractiveness [perceived attractiveness ˆ ÿ0:06 ‡ 2:06 (perceived

Facial adiposity

1705

6

Perceived health

5 4 3 2 1 0

0

1

2 3 4 Perceived facial adiposity

5

6

Figure 2. Interrelation between facial adiposity and facial health judgments [perceived health ˆ ÿ0:33 ‡ 2:45 (perceived facial adiposity) ÿ 0.41 (perceived facial adiposity)2 ; p 5 0:0005, R 2 ˆ 0:263]. Individuals with intermediate perceived facial adiposity are judged healthier than individuals with a perceived facial adiposity on either side of this optimum. The solid curve is the best-fit second-order (quadratic) polynomial equation. The third-order (cubic) polynomial equation is indicated as a dashed line (virtually indistinguishable from the quadratic curve).

facial adiposity) ÿ 0.37 (perceived facial adiposity)2; F2, 81 ˆ 9:269, p 5 0:0005, R 2 ˆ 0:186; figure 3]. Exploration of the relationship between perceived facial adiposity and both perceived health and attractiveness showed that other regression models based on linear, logarithmic, inverse, compound, power, S, growth, exponential, and logistic functions did not fit the data significantly. Cubic models also fitted the perceived health (F3, 80 ˆ 9:521, p 5 0:0005, R 2 ˆ 0:263) and attractiveness data well (F3, 80 ˆ 6:371, p ˆ 0:001, R 2 ˆ 0:193), but we consider the quadratic models a more appropriate model for two reasons. First, because R 2 increases with the number of regressors in the model, higher-order models tend to have inflated R 2 values. Second, the cubic fit is very close to the quadratic fit in both cases (figures 2 and 3).

Perceived attractiveness

6 5 4 3 2 1 0

0

1

2 3 4 Perceived facial adiposity

5

6

Figure 3. Interrelation between facial adiposity and facial attractiveness judgments [perceived attractiveness ˆ ÿ0:06 ‡ 2:06 (perceived facial adiposity) ÿ 0.37 (perceived facial adiposity)2 ; p 5 0:0005, R 2 ˆ 0:186]. Individuals with intermediate perceived facial adiposity are judged more attractive than individuals with a perceived facial adiposity on either side of this optimum. The curve is the best-fit second-order polynomial equation. The third-order (cubic) polynomial equation is indicated as a dashed line.

3.3 Perceived facial adiposity and health measures To test the relationship between perceived facial adiposity and actual health, we performed separate zero-order and partial Pearson's correlations (all two-tailed). In the partial correlations we partialled out age and parental income. Age is a determining

1706

V Coetzee, D I Perrett, I D Stephen

factor of facial adiposity (Rohrich et al 2008). Socio-economic status is widely reported as a covariate of health and we therefore controlled for parental income, as a limited measure of socio-economic status. We report the partial correlations, although the zero-order correlations were similar (see table 1). The frequency (r77 ˆ 0:235, p ˆ 0:037; table 1) and duration of cold and flu bouts (r77 ˆ 0:269, p ˆ 0:016; table 1) significantly and positively correlated with how heavy the individual was perceived to be. Use of antibiotics was significantly associated with perceived facial adiposity (Krustall ^Wallis x22 ˆ 11:706, p ˆ 0:003; table 1). Facial images of individuals who reported high use of antibiotics were judged significantly heavier than those who reported low use of antibiotics (Mann ^ Whitney U ˆ 50:00, p ˆ 0:006) and those that reported zero use (U ˆ 10:50, p ˆ 0:001), while there was only a tendency for low-use individuals to have a higher perceived facial adiposity than zero-use individuals (U ˆ 150:50, p ˆ 0:094). Cold and flu bouts and bout length are similar measures; we therefore included them in a principal component analysis (PCA). Use of antibiotics was not included in the PCA because of the low sample number and skewed distribution. We identified one component with an eigenvalue 4 1, which explained 60% of the variance. This respiratory-illness component correlated significantly with perceived facial adiposity (r77 ˆ 0:327, p ˆ 0:003; table 1). Table 1. Pearson's correlations showing the relationship between perceived facial adiposity, BMI, and actual health measures. Partial correlations controlled for parental income and age. All correlations were two-tailed. Associations with antibiotics use were tested with a Krustall ^ Wallis test. Principal component analysis components are underscored. Facial adiposity zero-order, r (df)

BMI partial, r (df)

zero-order, r (df)

partial, r (df)

Infections Cold and flu number Bout length Antibiotics Respiratory component

0.235* (77) 0.209 À (81) 0.244* (81) 0.269* (77) x 2 ˆ 11:706** (2) 0.293** (81) 0.327** (77)

0.198 À (80) 0.177 (80) x 2 ˆ 9:050* (2) 0.242 * (80)

0.216 À 0.190 À

(77) (77)

0.263*

(77)

Cardiovascular health Systolic BP Diastolic BP Cardiovascular component

0.278* (81) 0.432*** (81) 0.412*** (81)

0.441*** (80) 0.399*** (80) 0.487*** (80)

0.431*** (77) 0.405*** (77) 0.487*** (77)

0.264* (77) 0.452*** (77) 0.416*** (77)

Note: *** p 4 0:001, ** p 4 0:01, * p 4 0:05, À p 4 0:10; BP ˆ blood pressure.

We also observed a significant quadratic relationship between facial adiposity and the respiratory-illness component [respiratory-illness component ˆ ÿ0:90 ‡ 0:25 (perceived facial adiposity) ‡ 0.01 (perceived facial adiposity)2 ; F2, 80 ˆ 3:752, p ˆ 0:028, R 2 ˆ 0:086], but the quadratic relationship was almost identical to the linear relationship and did not support our prediction of increased infection in the underweight group (figure 4). The linear model fitted the data better than the quadratic relationship [respiratory illness component ˆ ÿ1:02 ‡ 0:34 (perceived facial adiposity); F2, 81 ˆ 7:580, p ˆ 0:007, R 2 ˆ 0:086]. Both cardiovascular measures significantly correlated with perceived facial adiposity. Individuals who were perceived as heavier had significantly higher systolic blood pressure (r77 ˆ 0:264, p ˆ 0:019; table 1) and diastolic blood pressure (r77 ˆ 0:452, p 5 0:0005; table 1). PCA of the cardiovascular measures revealed one component with an eigenvalue 4 1, which explained 74% of the variance. The cardiovascularillness component correlated significantly with the perception of weight (r77 ˆ 0:416, p 5 0:0005; table 1).

Facial adiposity

1707

Respiratory-illness component

3 2 1 0 ÿ1 ÿ2 ÿ3 ÿ4

0

1

2 3 4 Perceived facial adiposity

5

6

Figure 4. Interrelation between perceived facial adiposity and the respiratory-illness component. The quadratic relationship between perceived facial adiposity and the respiratory-illness component [respiratory-illness component ˆ ÿ0:90 ‡ 0:25 (perceived facial adiposity) ‡ 0.01 (perceived facial adiposity)2 ; p ˆ 0:028, R2 ˆ 0:086] was almost identical to the linear relationship (respiratoryillness component ˆ ÿ1:02 ‡ 0:34 (perceived facial adiposity); p ˆ 0:007, R 2 ˆ 0:086; not indicated on the graph, because of its similarity to the quadratic relationship].

3.4 BMI and health measures If perceived facial adiposity is accurately displaying relative body weight, one might expect perceived facial adiposity and BMI to correlate with health measures in a similar way. To test this we performed zero-order and partial Pearson's correlations controlling for parental income and age. Results for BMI were fairly similar to those obtained for perceived facial adiposity (table 1). Cold and flu bouts (r77 ˆ 0:216, p ˆ 0:056; table 1) and bout length (r77 ˆ 0:180, p ˆ 0:094; table 1) showed a positive, but non-significant, correlation with BMI; the respiratory-illness component correlated significantly with BMI (r77 ˆ 0:263, p ˆ 0:019; table 1). Use of antibiotics was significantly associated with BMI (Krustall ^ Wallis x22 ˆ 9:050, p ˆ 0:011; table 1). Individuals who reported high use of antibiotics were significantly heavier than those who reported low use of antibiotics (Mann ^ Whitney U ˆ 110:00, p ˆ 0:014) and those that reported zero use (U ˆ 24:00, p ˆ 0:003), while there was no significant difference in BMI between low-use and zero-use individuals. All the cardiovascular measures significantly correlated with BMI. Individuals with higher BMIs had significantly higher systolic (r77 ˆ 0:431, p 5 0:0005; table 1) and diastolic (r77 ˆ 0:405, p 5 0:0005; table 1) blood pressure. The cardiovascular-illness component also significantly correlated with BMI (r77 ˆ 0:487, p 5 0:0005; table 1). 4 Discussion In this study we set out to test whether facial adiposity, or apparent weight judged from the face, is a cue to health. As expected, people are quite accurate at judging weight on the basis of facial cues alone, enabling facial adiposity to serve as a potential cue to health and attractiveness. In order to be a valid cue to health, any cue must fulfil two prerequisites. First, people must use this cue in their judgments of health. Second, this cue must be associated with actual health measurements. Both criteria were met in the results. We present evidence that people use perceived facial adiposity as a cue when judging health. Adiposity produces a fairly salient shape cue in the face, so it is parsimonious to assume that facial adiposity, rather than a correlate thereof, provides the basis for estimating health. In further investigations it should be possible to isolate salient shape cues. Individuals with intermediate facial adiposity are judged healthier than individuals

1708

V Coetzee, D I Perrett, I D Stephen

with a facial adiposity on either side of this optimum. For the body, weight is an extremely important cue to perceived health. Swami et al (2008) showed that BMI explains more than 70% of the variance in health judgments of female bodies. We show that this is also true for the face, although perceived facial adiposity explains only 26% of the variance in health judgments (both sexes combined). This discrepancy in the amount of variance explained is not surprising given that there is a whole range of putative facial cues such as symmetry, demeanour, colour, and texture that also signal health. Not only is perceived facial adiposity (or a correlate thereof) used as a cue to health, it is also a powerful correlate of attractiveness. Individuals with intermediate perceived facial adiposity are judged more attractive than individuals with high and, to some extent, low perceived facial adiposity. Furthermore, we show that perceived facial adiposity provides information about past health and probable future health. In general, individuals with high perceived facial adiposity report more infections than those with low perceived facial adiposity. These apparently-high-weight individuals report a significantly higher frequency of use of antibiotics and longer and more frequent respiratory infections. When we combined cold and flu frequency and duration in a single component, the respiratory-illness component significantly and positively correlated with perceived facial adiposity, indicating that individuals with higher perceived facial adiposity are more likely to have a respiratory infection. A similar relationship is also seen for the association between infections and actual BMI, further strengthening the observed effect between perceived facial adiposity and reported infections. We did not find the expected curvilinear relationship between perceived facial adiposity and the respiratory-illness component, indicating that individuals judged to be underweight do not suffer from significantly more respiratory infections than normal-weight individuals. The association with respiratory infections therefore does not explain why underweight individuals are judged less healthy and/or attractive. Underweight individuals could be judged less healthy and attractive because of other health and/or fertility factors not tested here. For instance, underweight women (BMI 5 18:5) often have amenorrhoea or anovulatory menstrual cycles (Frish 1987), both of which reduce reproductive potential. Underweight individuals also report lower iron levels (Brown et al 2000) and could be more prone than normal-weight individuals to a variety of infections that fall outside the scope of this study. Lastly, we show a strong association between perceived facial adiposity and cardiovascular health. Individuals with higher perceived facial adiposity have significantly higher blood pressure, a condition that increases their risk of coronary heart disease and stroke (MacMahon et al 1990). This association between blood pressure and weight was replicated for BMI. One might argue that cardiovascular disease did not play an important role in shaping the mate preferences of our ancestors. Yet, cardiovascular disease is currently an enormous health burden in the developed world (WHO 2003). It is therefore plausible that current mate-choice preferences are shaped by the environment the individual finds himself/herself in. People change their preferences according to their local environment (Yu and Shepard 1999; Furnham et al 2002; Sugiyama 2004; Swami and Tove¨e 2005a, 2005b). For instance, people in rural areas, where food can be scarce, find heavier women more attractive than people in urban areas (Swami and Tove¨e 2005a, 2007). Preferences can also change rapidly (Silverstein et al 1986), so it is probable that modern people are associating excess weight and obesity with negative health outcomes. In future, we will test the association between weight and health in cross-cultural settings. In summary, we set out to test the hypothesis that facial adiposity can act as a cue to health. We showed that perceived facial adiposity significantly predicts health

Facial adiposity

1709

and attractiveness in a Western population and is associated with actual health risk in the same population. Perceived facial adiposity explains a substantial amount of the variance of perceived health and attractiveness; thus studies focusing on other facial cues (eg symmetry, sexual dimorphism, averageness) could benefit from controlling for facial adiposity. For example, the association between facial symmetry and health may be strengthened by controlling for facial adiposity. Our study provides one route through which facial characteristics can provide an accurate reflection of health, and thereby influence mate choice. Acknowledgments. We thank Johannes Schindelin and Michael Stirrat for their help with software development; Lesley Ferrier, Janek Lobmaier, Lindsey Macdougall, Rachel McCarey, and Stephanie Sharples for their help with data collection, and Anne Perrett for proofreading the manuscript. References Baik I, Curhan G C, Rimm E B, Bendich A, Willett W C, Fawzi W W, 2000 ``A prospective study of age and lifestyle factors in relation to community-acquired pneumonia in US men and women'' Archives of Internal Medicine 160 3082 ^ 3088 Boothroyd L G, Jones B C, Burt D, Perrett D I, 2007 ``Partner characteristics associated with masculinity, health and maturity in male faces'' Personality and Individual Differences 43 1163 ^ 1173 Brown W J, Mishra G, Kenardy J, Dobson A, 2000 ``Relationship between body mass index and well-being in young Australian women'' International Journal of Obesity 24 1360 ^ 1368 Choban P S, Heckler R, Burge J C, Flancbaum L, 1995 ``Increased incidence of nosocomial infections in obese surgical patients'' American Surgeon 6 1001 ^ 1005 Fink B, Grammer K, Thornhill R, 2001 ``Human (Homo sapiens) facial attractiveness in relation to skin texture and color'' Journal of Comparative Psychology 115 92 ^ 99 Fink B, Neave N, Manning J T, Grammer K, 2006 ``Facial symmetry and judgements of attractiveness, health and personality'' Personality and Individual Differences 41 491 ^ 499 Flegal K M, Graubard B I, Williamson D F, Gail M H, 2005 ``Excess deaths associated with underweight, overweight, and obesity'' Journal of the American Medical Association 293 1861 ^ 1867 Frish R E, 1987 ``Body fat, menarche, fitness and fertility''Human Reproduction 2 521 ^ 533 Furnham A, Baguma P, 1994 ``Cross-cultural differences in the evaluation of male and female body shapes'' International Journal of Eating Disorders 15 81 ^ 89 Furnham A, Moutafi J, Baguma P, 2002 ``A cross-cultural study on the role of weight and waistto-hip ratio on female attractiveness'' Personality and Individual Differences 32 729 ^ 745 Furnham A, Swami V, Shah K, 2006 ``Body weight, waist-to-hip ratio and breast size correlates of ratings of attractiveness and health'' Personality and Individual Differences 41 443 ^ 454 Grammer K, Thornhill R, 1994 ``Human (Homo sapiens) facial attractiveness and sexual selection: the role of symmetry and averageness'' Journal of Comparative Psychology 108 233 ^ 242 Gunnell D J, Frankel S J, Nanchanal K, Peters T J, Smith G D, 1998 ``Childhood obesity and adult cardiovascular mortality: a 57-y follow-up study based on the Boyd Orr cohort'' American Journal of Clinical Nutrition 67 1111 ^ 1118 Hamilton W D, Zuk M, 1982 ``Heritable true fitness and bright birds: a role for parasites?'' Science 218 384 ^ 387 Henderson J J A, Anglin J M, 2003 ``Facial attractiveness predicts longevity'' Evolution and Human Behavior 24 351 ^ 356 Hubert H B, Feinleib M, McNamara P M, Castelli W P, 1983 ``Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study'' Circulation 67 968 ^ 977 Hume D K, Montgomerie R, 2001 ``Facial attractiveness signals different aspects of `quality' in women and men'' Evolution and Human Behavior 22 93 ^ 112 Jedrychowski W, Maugeri U, Flak E, Mroz E, Bianchi I, 1998 ``Predisposition to acute respiratory infections among overweight preadolescent children: an epidemiological study in Poland'' Public Health 112 189 ^ 195 Jones B C, Little A C, Penton-Voak I S, Tiddeman B P, Burt D M, Perrett D I, 2001 ``Facial symmetry and judgements of apparent health: Support for a `good genes' explanation of the attractiveness ^ symmetry relationship'' Evolution and Human Behavior 22 417 ^ 429 Jones B C, Little A C, Feinberg D R, Penton-Voak I S, Tiddeman B P, Perrett D I, 2004 ``The relationship between shape symmetry and perceived skin condition in male facial attractiveness'' Evolution and Human Behavior 25 24 ^ 30

1710

V Coetzee, D I Perrett, I D Stephen

Kalick S M, Zebrowitz L A, Langlois J H, Johnson R M, 1998 ``Does human facial attractiveness honestly advertise health? Longitudinal data on an evolutionary question'' Psychological Science 9 8 ^ 13 Kirkpatrick M, Ryan M J, 1991 ``The evolution of mating preferences and the paradox of the lek'' Nature 350 33 ^ 38 Law Smith M J, Perrett D I, Jones B C, Cornwell R E, Moore F R, Feinberg D R, Boothroyd L G, Durrani S J, Stirrat M R, Whiten S, Pitman R M, Hillier S G, 2006 ``Facial appearance is a cue to oestrogen levels in women'' Proceedings of the Royal Society of London, Series B 273 135 ^ 140 Lusky A, Barell V, Lubin F, Kaplan G, Layani V, Shohat Z, Lev B, Wiener M, 1996 ``Relationship between morbidity and extreme values of body mass index in adolescent'' International Journal of Epidemiology 25 829 ^ 834 MacMahon S, Peto R, Collins R, Godwin J, MacMahon S, Cutler J, Sorlie P, Abbott R, Collins R, Neaton J, Abbott R, Dyer A, Stamler J, 1990 ``Blood pressure, stroke and coronary heart disease: Part 1. Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias'' The Lancet 335 765 ^ 774 Manson J E, Willett W C, Stampfer M J, Golditz G A, Hunter D J, Hankinson S E, Hennekens C H, Speizer F E, 1995 ``Body weight and mortality among women'' The New England Journal of Medicine 333 677 ^ 685 Marlowe F, Wetsman A, 2001 ``Preferred waist-to-hip ratio and ecology'' Personality and Individual Differences 30 481 ^ 489 Matts P, Fink B, Grammer K G, Burquest M, 2006 ``Skin color distribution plays a role in the perception of age, attractiveness and health in female faces'' Journal of the American Academy of Dermatology 56 AB26 Matts P J, Fink B, Grammer K, Burquest M, 2007 ``Color homogeneity and visual perception of age, health, and attractiveness of female facial skin'' Journal of the American Academy of Dermatology 57 977 ^ 984 Mokdad A H, Ford E S, Bowman B A, Dietz W H, Vinicor F, Bales V S, Marks J S, 2003 ``Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001'' The Journal of the American Medical Association 289 76 ^ 79 Must A, Spadano J, Coakley E H, Field A E, Colditz G, Dietz W H, 1999 ``The disease burden associated with overweight and obesity'' The Journal of the American Medical Association 282 1523 ^ 1529 Penton-Voak I S, Jones B C, Little A C, Baker S, Tiddeman B, Burt D M, Perrett D I, 2001 ``Symmetry, sexual dimorphism in facial proportions and male facial attractiveness'' Proceedings of the Royal Society of London, Series B 268 1617 ^ 1623 Perrett D I, Lee K J, Penton-Voak I, Rowland D, Yoshikawa S, Burt D M, Henzi S P, Castles D L, Akamatsu S, 1998 ``Effects of sexual dimorphism on facial attractiveness'' Nature 394 884 ^ 887 Peters M, Rhodes G, Simmons L W, 2008 ``Does attractiveness in men provide clues to semen quality?'' Journal of Evolutionary Biology 21 572 ^ 579 Pi-Sunyer F X, 1993 ``Medical hazards of obesity'' Annals of Internal Medicine 119 655 ^ 660 Rhodes G, Chan J, Zebrowitz L A, Simmons L W, 2003 ``Does sexual dimorphism in human faces signal health?'' Proceedings of the Royal Society of London, Series B 270 S93 ^ S95 Rhodes G, Yoshikawa S, Palmermo R, Simmons L W, Peters M, Lee K, Halberstadt J, Crawford J R 2007 ``Perceived health contributes to the attractiveness of facial symmetry, averageness, and sexual dimorphism'' Perception 36 1244 ^ 1252 Rhodes G, Zebrowitz L A, Clark A, Kalick S M, Hightower A, McKay R, 2001 ``Do facial averageness and symmetry signal health?'' Evolution and Human Behavior 22 31 ^ 46 Ritz B W, Gardner E M, 2006 ``Malnutrition and energy restriction differentially affect viral immunity'' Journal of Nutrition 136 1141 ^ 1144 Roberts S C, Little A C, Gosling L M, Perrett D I, Carter V, Jones B C, Penton-Voak I, Petrie M, 2005 ``MHC-heterozygosity and human facial attractiveness'' Evolution and Human Behavior 26 213 ^ 226 Rohrich R J, Pessa J E, Ristow B, 2008 ``The youthful cheek and the deep medial fat compartment'' Plastic and Reconstructive Surgery 121 2107 ^ 2112 Serdula M K, Iver D, Coates R J, Freedman D S, Williamson D F, Byers T, 1993 ``Do obese children become obese adults? A review of the literature'' Preventative Medicine 22 167 ^ 177 Shackelford T K, Larsen R J, 1999 ``Facial attractiveness and physical health'' Evolution and Human Behavior 20 71 ^ 76 Silverstein B, Perdue L, Peterson B, Kelly E, 1986 ``The role of the mass media in promoting a thin standard of bodily attractiveness for women'' Sex Roles 14 519 ^ 532

Facial adiposity

1711

Soler C, Nunez M, Gutierrez R, Nunez J, Medina P, Sancho M, Alvarez J, Nunez A, 2003 ``Facial attractiveness in men provides clues to semen quality'' Evolution and Human Behavior 24 199 ^ 207 Stephen I D, Perrett D I, Coetzee V, Law Smith M, 2009 ``Skin blood perfusion and oxygenation color affect perceived human health'' PLoS ONE http://dx.plos.org/10.1371/journal.pone.0005083 Sugiyama L S, 2004 ``Is beauty in the context-sensitive adaptations of the beholder?: Shiwiar use of waist-to-hip ratio in assessments of female mate value'' Evolution and Human Behavior 25 51 ^ 62 Swami V, Miller R, Furnham A, Penke L, Tove¨e M J, 2008 ``The influence of men's sexual strategies on perceptions of women's bodily attractiveness, health and fertility'' Personality and Individual Differences 44 98 ^ 107 Swami V, Tove¨e M J, 2005a ``Female physical attractiveness in Britain and Malaysia: A crosscultural study'' Body Image 2 115 ^ 128 Swami V, Tove¨e M J, 2005b ``Male physical attractiveness in Britain and Malaysia: A cross-cultural study'' Body Image 2 383 ^ 393 Swami V, Tove¨e M J, 2007 ``Differences in attractiveness preferences between observers in low and high resource environments in Thailand'' Journal of Evolutionary Psychology 5 149 ^ 160 Tanaka S, Inoue S, Isoda F, Waseda M, Ishihara M, Yamakawa T, Sugiyama A, Takamura Y, Okuda K, 1993 ``Impaired immunity in obesity: suppressed but reversible lymphocyte responsiveness'' International Journal of Obesity and Related Metabolic Disorders 17 631 ^ 636 Thornhill R, Gangestad S W, 2006 ``Facial sexual dimorphism, developmental stability, and susceptibility to disease in men and women'' Evolution and Human Behavior 27 131 ^ 144 Thornhill R, Grammer K, 1999 ``The body and face of woman: one ornament that signals quality?'' Evolution and Human Behavior 20 105 ^ 120 Tove¨e M J, Maisey D S, Emery J L, Cornelissen P L, 1999 ``Visual cues to female physical attractiveness'' Proceedings of the Royal Society of London, Series B 266 211 ^ 218 Tove¨e M J, Reinhardt S, Emery J L, Cornelissen P L, 1998 ``Optimum body-mass index and maximum sexual attractiveness'' The Lancet 352 548 Tove¨e M J, Swami V, Furnham A, Mangalparsad R, 2006 ``Changing perceptions of attractiveness as observers are exposed to different culture'' Evolution and Human Behavior 27 443 ^ 456 Vilar-Compte D, Mohar A, Sandoval S, de la Rosa M, Patricia G, Volkow P, 2000 ``Surgical site infections at the National Cancer Institute in Mexico: a case-control study'' American Journal of Infection Control 28 14 ^ 20 WHO, 2000 ``Obesity: Preventing and managing the global epidemic'', in WHO Technical Report Series 894 (Geneva: WHO), page 9 WHO, 2003 ``Diet, nutrition and the prevention of chronic diseases'', in WHO Technical Report Series 916 (Geneva: WHO) page 4 Wilson P W F, D'Agostino R B, Sullivan L, Parise H, Kannel W B, 2002 ``Overweight and obesity as determinants of cardiovascular risk: The Framingham experience'' Archives of Internal Medicine 162 1867 ^ 1872 Yu D W, Shepard G H, 1999 ``Reply: the mystery of female beauty'' Nature 399 216 Zebrowitz L A, Rhodes G, 2004 ``Sensitivity to `bad genes' and the anomalous face overgeneralization effect: cue validity, cue utilization, and accuracy in judging intelligence and health'' Journal of Nonverbal Behavior 28 167 ^ 185

ß 2009 a Pion publication

ISSN 0301-0066 (print)

ISSN 1468-4233 (electronic)

www.perceptionweb.com

Conditions of use. This article may be downloaded from the Perception website for personal research by members of subscribing organisations. Authors are entitled to distribute their own article (in printed form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online distribution system) without permission of the publisher.