1. Introduction

One of the most robust findings in studies of human attractiveness is the high level of agreement among individuals on what is an attractive face (see Langlois et al., 2000 for a meta-analytic review). In many species, however, individual differences in mate preferences are evident. For example, female sticklebacks (Gasterosteus aculeatus) and guppies (Poecilia reticulata) in good physical condition are more likely to prefer cues to immunocompetence in males than are relatively poor condition females (Bakker et al., 1999, Lopez, 1999). Such condition-dependent preferences are potentially adaptive if only females in good physical condition are able to obtain the healthiest mates. As the face plays an important role in human mate choice (Miller & Todd, 1998, Thornhill & Gangestad, 1999), women's own condition may influence their preferences for male faces with characteristics that are thought to be indices of immune system strength and health (e.g., symmetry and masculinity; Gangestad & Simpson, 2000, Thornhill & Gangestad, 1999).

Consistent with this proposal, Little, Burt, Penton-Voak, and Perrett (2001) reported positive relationships between women's ratings of their own attractiveness and the extent to which they preferred male faces with symmetric and masculine shapes. Penton-Voak et al. (2003) replicated the finding of Little et al. regarding female preferences for sexual dimorphism in male face shape using independent measures of female physical condition (waist-to-hip ratio and other-rated facial attractiveness, see Miller & Todd, 1998). Although psychological factors (e.g., anxiety, depression, sand tress) might also reflect female condition, and psychological factors are known to influence physical health (for a review, see Brunner & Marmot, 1999), possible relationships between women's psychological condition and their face preferences have not yet been investigated.

Visible skin condition is positively associated with indices of men's genotypic health (e.g., MHC heterozygosity, Roberts et al., 2005), phenotypic health (e.g., pallor during illness, Roujeau, 2001), and facial attractiveness (Jones et al., 2004, Roberts et al., 2005). In women, a low waist-to-hip ratio is associated with an attractive body shape (Singh, 1993), medical health (Singh, 1993, for a review), and fertility (Jasienska, Ziomkiewicz, Ellison, Lipson, & Thune, 2004). Given the above, we investigated if women with low waist-to-hip ratios expressed stronger attraction to apparent health in faces than did women with relatively high waist-to-hip ratios. We also tested if women who scored low on assessments of anxiety, depression, and stress expressed stronger attraction to apparent health in faces than did women who scored relatively high on these measures. As the effects of psychological and physical conditions on face preferences could be mediated by women's beliefs about their own attractiveness, the possible effects of self-rated attractiveness were also investigated. Stimuli (composite male and female faces) were manipulated in color and texture cues associated with apparent health and did not differ in symmetry or masculinity of 2D shape.

2. Methods

3. Results

One-sample t tests (comparing preferences with chance) indicated that women generally preferred faces with raised apparent health (male faces: t=5.509, df=4.64, p<.001; female faces: t=5.741, df=68, p<.001).

We analysed preference for apparent health in male faces with linear regression (dependent variable: proportion of trials on which health was preferred for male faces; independent variables entered using the enter method: waist-to-hip ratio, age, psychological condition, and self-rated attractiveness). The overall model was significant (adjusted r2=.100, F=2.890, df=68, p=.029). Waist-to-hip ratio (β=−.238, t=−2.059, p=.044) and psychological condition (β=.256, t=2.173, p=.044) were independently related to preference for apparent health in male faces. Women with low waist-to-hip ratios or who were in good psychological condition expressed stronger attraction to apparent health in male faces than did women with high waist-to-hip ratios or who were in relatively poor psychological condition. There were no effects of age (β=.090, t=0.776, p=.441) or self-rated attractiveness (β=.128, t=1.090, p=.280).

Preferences for apparent health in female faces were analysed in the same way. By contrast with our findings for male faces, the overall model was not significant (adjusted r2=.002, F=1.036, df=68, p=.396) and there were no independent effects of any of the independent variables (all absolute β values <.184, all absolute t values <1.516, all p>.135).

There were no significant relationships between the independent variables (all p>.328).

4. Discussion

Although women generally preferred faces with raised apparent health to faces with lowered apparent health, systematic variation in women's preference for apparent health in male faces was also observed. Women with low waist-to-hip ratios expressed stronger attraction to apparent health in male faces, but not female faces, than did women with relatively high waist-to-hip ratios. This finding complements those from earlier studies in which positive relationships between estimates of women's physical condition and their preferences for male faces with masculine and symmetric shapes were observed (e.g., Little et al., 2001, Penton-Voak et al., 2003). Independent of the effect of waist-to-hip ratio, women who scored low on the anxiety, depression, and stress measures demonstrated stronger preferences for apparent health in male faces than did women who scored relatively high on the anxiety, depression, and stress measures. Again, this relationship between women's condition and their preferences for apparent health in faces was observed for judgments of male faces but not for judgments of female faces. In addition to the effect of physical condition, psychological condition also appears to contribute to individual differences in women's preferences for male faces. Both the effect of waist-to-hip ratio and that of psychological condition were independent of women's beliefs about their own attractiveness.

Color and texture cues to apparent health in male faces (e.g., skin tone and evenness of complexion; Jones et al., 2004) signal both genotypic (Roberts et al., 2005) and phenotypic (Roujeau, 2001) health. Our findings therefore suggest that women's physical and psychological conditions are positively related to their preference for healthy potential mates. This systematic variation in health preference is potentially adaptive if only women who are in good condition (physically and psychologically) are able to obtain the healthiest mates. Condition-dependent preferences for healthy males could yield reproductive benefits by optimising mating effort (e.g., reducing mate search time) while simultaneously allowing for pursuit of the healthiest mate that is attainable. Given the high attractiveness of healthy-looking individuals, the condition-dependent preferences we report here may contribute to the similarity in the attractiveness of partners (see Feingold, 1988). By contrast, with previous studies where matching for attractiveness in partner choice was observed when there was an actual possibility of rejection by the partner, but not when there was no opportunity for rejection to occur (Huston, 1973, Walster et al., 1966), the condition-dependent preferences we report here suggest that matching for attractiveness is also evident on preference tests where there is no possibility of actual rejection.