The Science of Facial Attractiveness

What science actually measures

Facial attractiveness is one of the most heavily studied topics in social psychology. Across hundreds of experiments since the 1980s, three structural signals have repeatedly emerged as significant predictors of how a face is rated by independent observers: symmetry, averageness, and sexual dimorphism (femininity in female faces; a complex mix of features in male faces). Gillian Rhodes's 2006 Annual Review of Psychology synthesis [2] remains the canonical reference: each of the three contributes independently, and effects replicate across age, ethnicity, and culture.

Symmetry has the longest empirical track record. Thornhill and Gangestad's 1993 paper [10] proposed that bilateral symmetry signals developmental stability — the ability of an organism to buffer genetic and environmental noise during growth. Rhodes, Proffitt, Grady, and Sumich [7] later showed experimentally that artificially increasing symmetry via image morphing raises attractiveness ratings, establishing causation rather than mere correlation. Penton-Voak and colleagues [1] then demonstrated this effect holds for male faces independently of dimorphism.

Averageness is more counterintuitive. Composite faces — created by averaging the geometry of dozens of individual faces — are rated more attractive than most of the originals that went into them. Apicella, Little, and Marlowe [5] replicated this in the Hadza of Tanzania, an isolated hunter-gatherer population, ruling out a Western-media explanation. But Perrett, May, and Yoshikawa's 1994 Nature paper [3] revealed a wrinkle: when you exaggerate the difference between an attractive face and the average, the exaggerated face is rated even more attractive. Averageness is a strong baseline, not a ceiling.

Sexual dimorphism — the degree to which a face shows traits typical of one sex — interacts with the other two. Perrett et al. [4] demonstrated that femininity is consistently preferred in female faces; in male faces, the picture is more complex, with women's preferences shifting between more masculine and more feminine male faces depending on relationship context and hormonal state [11, 12]. This is why no single “score” tells the whole story: attractiveness is a multidimensional readout, not a scalar.

Beyond symmetry: FWHR, canthal tilt, and the modern toolkit

Two structural metrics have dominated the past 15 years of research on facial geometry. The first is the facial width-to-height ratio (FWHR) — bizygomatic width divided by upper-face height. Carré and McCormick's 2008 paper [8] showed that FWHR predicts aggressive behavior in male professional hockey players, and dozens of follow-ups have linked higher FWHR to perceived dominance, achievement drive, and short-term mating success. FWHR is one of the few facial metrics with a behavioral, not just perceptual, payoff.

The second is canthal tilt — the angle from the inner to the outer corner of the eye. Positive tilt (outer higher than inner) reads as alert and confident; neutral or negative tilt reads as tired or sad. Canthal tilt does not yet have a single landmark paper like FWHR, but it appears as a contributor in the broader literature on eye geometry, hooding, and the perception of dominance versus warmth.

Symmetry, averageness, dimorphism, FWHR, and canthal tilt — together with skin homogeneity and expression — are the core signals every modern attractiveness model is built on. Little, Jones, and DeBruine's 2011 synthesis [6] integrates them into a single evolutionary framework: each measurable feature is a candidate cue to the underlying traits (health, fertility, developmental stability) that ancestral mate preferences presumably tracked.

None of this means a single number summarizes how a face will be perceived. It means there are real, replicable, geometric inputs that any AI model can measure — and those measurements, applied correctly, do correlate with human ratings far better than gut intuition. RealSmile's 17-metric looksmaxxing test is built directly on this literature.

Bibliography

All citations are real, peer-reviewed publications. DOIs are provided where applicable; URLs link to publisher pages where the abstract or full text is available.

1. [1]

   Symmetry, sexual dimorphism in facial proportions and male facial attractiveness

   Penton-Voak, I. S., Jones, B. C., Little, A. C., Baker, S., et al. (2001). Proceedings of the Royal Society B: Biological Sciences, 268(1476), 1617–1623.

   Established that bilateral facial symmetry predicts perceived attractiveness in male faces independently of sexual dimorphism. One of the foundational symmetry-attractiveness papers.

   DOI: 10.1098/rspb.2001.1703Publisher page →
2. [2]

   Facial attractiveness

   Rhodes, G. (2006). Annual Review of Psychology, 57, 199–226.

   Landmark review of three decades of research. Concludes that averageness, symmetry, and sexual dimorphism each independently contribute to facial attractiveness across cultures.

   DOI: 10.1146/annurev.psych.57.102904.190208Publisher page →
3. [3]

   Facial shape and judgements of female attractiveness

   Perrett, D. I., May, K. A., & Yoshikawa, S. (1994). Nature, 368(6468), 239–242.

   Showed that exaggerated, non-average female faces (caricatures of attractive faces) were rated more attractive than the average — challenging the pure-averageness hypothesis.

   DOI: 10.1038/368239a0Publisher page →
4. [4]

   Effects of sexual dimorphism on facial attractiveness

   Perrett, D. I., Lee, K. J., Penton-Voak, I., et al. (1998). Nature, 394(6696), 884–887.

   Demonstrated that femininity in female faces and a complex mix of feminine + masculine traits in male faces predict attractiveness ratings.

   DOI: 10.1038/29772Publisher page →
5. [5]

   Facial averageness and attractiveness in an isolated population of hunter-gatherers

   Apicella, C. L., Little, A. C., & Marlowe, F. W. (2007). Perception, 36(12), 1813–1820.

   Replicated the averageness-equals-attractiveness finding in the Hadza of Tanzania, supporting cross-cultural and adaptive interpretations.

   DOI: 10.1068/p5601Publisher page →
6. [6]

   Facial attractiveness: Evolutionary based research

   Little, A. C., Jones, B. C., & DeBruine, L. M. (2011). Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1571), 1638–1659.

   Comprehensive synthesis of evolutionary psychology research on facial attractiveness, including symmetry, averageness, and condition-dependent cues.

   DOI: 10.1098/rstb.2010.0404Publisher page →
7. [7]

   Facial symmetry and the perception of beauty

   Rhodes, G., Proffitt, F., Grady, J. M., & Sumich, A. (1998). Psychonomic Bulletin & Review, 5(4), 659–669.

   Showed experimentally that increasing facial symmetry via image manipulation increases attractiveness ratings — direct causal evidence rather than correlation.

   DOI: 10.3758/BF03208842Publisher page →
8. [8]

   Facial structure is a reliable cue of aggressive behavior

   Carré, J. M., & McCormick, C. M. (2008). Proceedings of the Royal Society B: Biological Sciences, 275(1651), 2651–2656.

   Established that the facial width-to-height ratio (FWHR) predicts aggressive behavior in male hockey players — the foundational FWHR study cited by every subsequent paper.

   DOI: 10.1098/rspb.2008.0873Publisher page →
9. [9]

   Facial symmetry and judgements of apparent health

   Jones, B. C., Little, A. C., Penton-Voak, I. S., et al. (2001). Evolution and Human Behavior, 22(6), 417–429.

   Linked perceived facial symmetry to perceived health, supporting the "good genes" hypothesis: symmetric faces are read as more developmentally stable.

   DOI: 10.1016/S1090-5138(01)00083-6
10. [10]

    Human facial beauty: Averageness, symmetry, and parasite resistance

    Thornhill, R., & Gangestad, S. W. (1993). Human Nature, 4(3), 237–269.

    Foundational evolutionary-psychology paper proposing that attractiveness signals heritable parasite resistance via developmental stability cues like symmetry.

    DOI: 10.1007/BF02692201Publisher page →
11. [11]

    The role of masculinity and distinctiveness on the perception of attractiveness in human male faces

    DeBruine, L. M., Jones, B. C., Smith, F. G., & Little, A. C. (2010). Behavioral Ecology, 21(3), 540–545.

    Showed that perceived masculinity and distinctiveness load onto distinct dimensions, and women trade off these traits depending on context (short-term vs long-term mate preferences).

    DOI: 10.1093/beheco/arq023
12. [12]

    Human oestrus

    Gangestad, S. W., & Thornhill, R. (2008). Proceedings of the Royal Society B: Biological Sciences, 275(1638), 991–1000.

    Reviewed the evidence that women's preference for masculine, symmetric faces shifts across the ovulatory cycle — relevant to interpreting why "attractiveness" is not a single fixed score.

    DOI: 10.1098/rspb.2007.1425Publisher page →