Original Article
Sexual dimorphism of facial width-to-height ratio in human skulls and faces: A meta-analytical approach

https://doi.org/10.1016/j.evolhumbehav.2016.12.002Get rights and content

Abstract

Facial width-to-height ratio (FWHR), defined as the width of the face divided by the upper facial height, is a cue to behaviour. Explanations for this link often involve the idea that FWHR is sexually dimorphic, resulting from intersexual selection pressures. However, few studies have considered sexual dimorphism in skulls since the original paper on this topic, and it is possible that different explanations may be required if faces show sex differences but skulls do not. Here, meta-analyses of skulls found that men did have larger FWHR than women, although this effect was small. However, after categorising samples by ethnicity and geographical origin, meta-analyses only found evidence of sex differences in East Asians, and again, this effect was small. A re-analysis of previous studies after excluding skull samples found little evidence of sexual dimorphism in faces. Again, considering ethnicities separately, I found no differences for White samples but a medium-sized effect with East Asians, although this was not statistically significant with only three samples. Taken together, I found no reason to consider FWHR as a sexually dimorphic measure in skulls or faces, at least not universally, and so accounts based upon this assumption need rethinking if researchers are to explain the relationship between FWHR and behaviour.

Introduction

The idea that the human face provides social information is not a new one (Darwin, 1872). We can determine the identity (Bruce & Young, 1986), sex (Burton, Bruce, & Dench, 1993), age (Rhodes, 2009), and ethnicity (Montepare & Opeyo, 2002) of a stranger with relative ease, as well as more dynamic and changing information like emotional state (Elfenbein & Ambady, 2002). There is also evidence that trait information like personality, physical and mental health, and even sexual orientation can be perceived with some accuracy from faces alone (Jones et al., 2012, Kramer and Ward, 2010, Rule et al., 2009, Scott et al., 2013).

In 2007, researchers provided evidence of one particular facial measure, the width-to-height ratio (FWHR – see Fig. 1; Weston, Friday, & Liò, 2007), which they found to be sexually dimorphic in human skulls, and has since been the subject of intense investigation as a cue to numerous behaviours. While overall size differences play a large role in general skull dimorphism (Calcagno, 1981, Lestrel, 1974, Rightmire, 1970), Weston and colleagues suggested that this ratio difference was instead due to developmental differences in shape trajectories during puberty. Specifically, the height of the upper face (defined as the nasion-prosthion distance) in adults is similar in men and women, while the (bizygomatic) width is larger in men. In other words, while sex differences in skulls are expected simply because men grow to be larger than women, the bizygomatic width in males show additional growth at puberty beyond this predicted increase. The researchers argued that this difference in skull shape might result from intersexual selection pressures, so that a region of the face has evolved which highlights the distinction between men and women.

Why evidence of sexual dimorphism predicts an association between FWHR and behaviour is less clear. If female preferences led to increased facial width in men (although evidence actually suggests that wider faces are judged to be less attractive; Geniole, Denson, Dixson, Carré, & McCormick, 2015), it may not necessarily follow that within-sex differences are correlated with behaviours. More intuitively, intrasexual selection pressures (e.g., male-male competition) could have resulted in increased success for wider-faced men, resulting in an appearance–behaviour link, especially if these two factors have the same underlying mechanism (testosterone, for instance). Might both explanations overlap, whereby a facial cue that highlights ‘maleness’ to women has become associated with masculinity (both in appearance and behaviour) in men? Of course, there is no reason to assume that the mechanisms underlying sex differences in facial development are the same as those that may drive a within-sex association between appearance and behaviour.

While the precise account and its relationship with sexual dimorphism remain unclear, FWHR does appear to function as a social cue. Levels of masculine characteristics (e.g., aggression, dominance, deception) in men correlate with FWHR, as do perceived levels of these traits (for meta-analyses, see Geniole et al., 2015, Haselhuhn et al., 2015). The explanation for this FWHR–behaviour association is thought to involve testosterone (Carré and McCormick, 2008, Sell et al., 2009), which may influence both facial development and behavioural characteristics. Indeed, initial research found significant associations between FWHR in men and baseline levels of testosterone, as well as testosterone changes in response to potential mate exposure (Lefevre, Lewis, Perrett, & Penke, 2013).

Somewhat problematically for this account, FWHR may not actually be sexually dimorphic in faces (Kramer et al., 2012, Lefevre et al., 2012, Özener, 2012) or skulls (Gómez-Valdés et al., 2013, Stirrat et al., 2012). Of course, it may be that different mechanisms drive facial development in men and women, allowing for testosterone-produced correlates of behaviour in men without differences between the sexes (Lefevre et al., 2013). In a recent meta-analysis of this field, the authors found significant (but small) sex differences when considering studies of both skulls and faces together (Geniole et al., 2015), as well as for subsets of studies (2D photographs versus other materials). However, it is not clear whether differences remain when only skulls are analysed since this distinction was not made in their analyses. It may be that skulls do not show sex differences in FWHR but faces do, perhaps through evolved cues that utilise soft tissue deposits, which differ in men and women (Enlow, 1982). This would be an important caveat when investigating the explanatory mechanisms linking behaviour and facial measures.

One potential issue with previous investigations is that they have not considered populations separately based upon ethnicity or geographical origin. Given evidence of between-population differences in skulls (Gill and Rhine, 2004, İşcan and Steyn, 1999, Ousley et al., 2009), the inclusion of all groups into a single analysis will inherently suffer from this additional source of noise. It may be that FWHR dimorphism is present in some ethnicities/populations but not others, and this could account for the mixed results that have previously been found with faces. This would also be an important caveat for theories of dimorphism and signalling.

The other problem for the ‘FWHR–testosterone–behaviour’ account is that FWHR may not actually be associated with testosterone. In recent research investigating several samples and reporting a combined meta-analysis, no relationship was found between FWHR in adult men and baseline testosterone or competition-induced testosterone reactivity (Bird et al., 2016). Even during adolescence, when testosterone is hypothesised to impact facial growth (Weston et al., 2007), no relationship was found between male FWHR and testosterone levels or other known testosterone-derived traits (Hodges-Simeon, Hanson Sobraske, Samore, Gurven, & Gaulin, 2016). Indeed, FWHR showed no change during adolescence and no growth spurt, contrary to predictions.

In the current work, I focussed specifically on whether FWHR is sexually dimorphic in adult human skulls using a meta-analytical approach. Given that the popular topic of FWHR as an important facial cue originated from this initial finding (Weston et al., 2007), it is worth further examination using multiple large samples. I also considered geographical and ethnic origins as potential factors in order to allow for the likelihood that populations may differ. For this reason, I revisit the topic of FWHR sex differences in faces, again considering ethnicity as a potential influence. Importantly, prior large-scale analyses in this area have yet to consider the distinction between faces and skulls, and the possibility (and evolutionary implications) that there may be FWHR sex differences in one but not the other.

Section snippets

Previous research

All peer-reviewed and published manuscripts that investigated human skull FWHR separately for men and women were included. This involved searching through all articles that cited Weston et al. (2007), the first paper to propose this measure as a topic of interest. Conveniently, all articles prior to the end of 2014 had already been identified in the recent meta-analysis by Geniole et al. (2015), and no newer research (as of May 2016) or omissions were found. This resulted in the inclusion of

Meta-analysis of skulls for all populations

The Supplementary materials provide a summary of the eleven databases included in the meta-analysis, comprising 4918 men and 2924 women from 87 populations.

Discussion

The current meta-analyses provide evidence that casts doubt on what seems to be the currently accepted story regarding FWHR dimorphism. Although an overall analysis of human skulls found a very small (though statistically significant) effect, where men showed larger FWHR than women, there is an argument to be made for considering ethnicities separately (Gill and Rhine, 2004, İşcan and Steyn, 1999, Ousley et al., 2009). Subgroup analyses were suggestive of a moderating effect of ethnicity

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Acknowledgements

I am very grateful to Lyman Jellema for providing access to the Hamann-Todd human osteological collection, as well as background information and other details regarding the specimens. I also thank Rolando González-José for providing the data that he and his colleagues had previously collected. Finally, I thank Geoff Cumming for statistical advice, and Alex Jones and Renée Lefebvre for comments on the manuscript.

References (55)

  • D. Arcus et al.

    Temperament and craniofacial variation in the first two years

    Child Development

    (1995)
  • B.M. Bird et al.

    Evolution and human behavior

    (2016)
  • M. Borenstein et al.

    Introduction to meta-analysis

    (2009)
  • V. Bruce et al.

    Understanding face recognition

    British Journal of Psychology

    (1986)
  • A.M. Burton et al.

    What's the difference between men and women? Evidence from facial measurement

    Perception

    (1993)
  • J.M. Carré et al.

    In your face: Facial metrics predict aggressive behaviour in the laboratory and in varsity and professional hockey players

    Proceedings of the Royal Society B: Biological Sciences

    (2008)
  • V. Coetzee et al.

    Deciphering faces: Quantifiable visual cues to weight

    Perception

    (2010)
  • G. Cumming

    Understanding the new statistics: Effect sizes, confidence intervals, and meta-analysis

    (2012)
  • G. Cumming et al.

    Introduction to the new statistics: Estimation, open science, and beyond

    (2017)
  • C.R. Darwin

    The expression of the emotions in man and animals

    (1872)
  • H. de Villiers

    The skull of the South African Negro: A biometrical and morphological study

    (1968)
  • H.A. Elfenbein et al.

    On the universality and cultural specificity of emotion recognition: A meta-analysis

    Psychological Bulletin

    (2002)
  • D.H. Enlow

    Handbook of facial growth

    (1982)
  • F. Fang et al.

    A systematic review of inter-ethnic variability in facial dimensions

    Plastic and Reconstructive Surgery

    (2011)
  • S.N. Geniole et al.

    Evidence from meta-analyses of the facial width-to-height ratio as an evolved cue of threat

    PloS One

    (2015)
  • J. Gómez-Valdés et al.

    Lack of support for the association between facial shape and aggression: A reappraisal based on a worldwide population genetics perspective

    PloS One

    (2013)
  • Cited by (0)

    View full text