Home > Sex/Gender Differences, The Body > Another Eye Tracking Study of Female Body Attractiveness

Another Eye Tracking Study of Female Body Attractiveness

Seda Dural et.al. (2008) used eye tracking combined with attractiveness ratings to explore the effect of altering waist-hip ratio and body fat levels in figures generated from 3D modeling software.

The stimuli were presented individually to 50 males and 50 females aged 18-22 (probably Turkish college students) on an LCD screen for viewing and rated for attractiveness while eye movements were tracked using, what seems to be, in-lab created eye tracking hardware. (The article is published in Turkish and I have been forced to rely on the English language summary combined with obscure, computer-generated translations of sections of the full text.) That being said, the stimuli are very similar to those used in Singh’s original studies, though more detail and dimensionality are visible. Attractiveness ratings data are contained in this chart:

For eye tracking analysis, the authors divided the figures into 3 areas:

and used eye tracking data from the most and least attractive stimuli and found that raters looked more often at area B than at areas A and C in both the most and least attractive stimulus figures. The authors conclude that their study provides significant support for the role of waist-hip ratio in the evaluation of female body attractiveness.


  • Utilized 3D software to generate stimulus figures
  • Utilized eye tracking data
  • Non-U.S. college student sample


  • Considering that there is: minimal variation in section A across the 3 weight categories; minimal variation in section C within weight categories; and the most variation between all stimuli in section B; it is possible that the subjects in this study learned that section B contained the most variable information in the stimuli and that section B viewing frequencies are an artifact of the experimental controls themselves
  • As the table above shows (labeled Sekel 5), normal body weight images were significantly more attractive than either under or over weight images
  • Eye fixations in section B do not necessarily indicate that attention is being directed to waist-hip ratio (Cornelissen, 2009)

All-in-all, this study further confirms that both total body fat and body fat distribution as revealed in the WHR are components of female body attractiveness. The eye tracking data, however, do not support the assertion that WHR is the primary determinant of body attractiveness in women.

Wayne Hooke

ResearchBlogging.orgDural, Seda, Cetinkaya, Hakan, Gulbetekin, Evrim (2008). The role of waist-to-hip ratio in evaluation of female physical attractiveness: Eye-tracker data.
Kadinin fiziksel cekiciliinin deerlendirilmesinde bel-kalca-oraninin rolu: Goz-izleme sistemi verileri. Turk Psikoloji Dergisi, 23(61), 75-88

  1. Nadia
    August 2, 2009 at 6:21 am

    I just wanted to point out something that you often point out, but that is actually refuted by two recent studies. You did not discuss multivariate analysis in the Rilling paper. In neither Rilling et. al 2009 or Smith et. al 2007, the two studies that use digital video of real women, found any effects of WHR in multivariate analysis. These data don’t manipulate leg length, bust size, or waist and hip size independently. This is important given the finding that waist circumference independently influences attractiveness judgments.

  2. Nadia
    August 2, 2009 at 7:26 am

    *often point out that studies indicate WHR does contribute to attractiveness judgments. WHR did not survive the multiple regression in Smith et. al 2007 or Rilling et. al 2009. Interestingly enough you write that “All-in-all, this study further confirms that both total body fat and body fat distribution as revealed in the WHR are components of female body attractiveness.” Neither WHR or BMI were actually predictive in the multivariate analysis of Rilling et. al.

    I really don’t want to be redundant, because I’ve mentioned this point before, but Thornhill and Grammer 1999. Mean BMI ~20, mean WHR ~.72, and the correlations are similar to other studies using less attractive stimulus sets. Wayne, you often bring up the point of few figures who are attractive or have optimal measurements. I would not argue that it should never be addressed, but how much of an issue is it? If say Thornhill and Grammer found substantially different correlations using their image set, then other studies (Smith et. al 2007, Rilling et. al 2009, Tovee et. al 2002) that would support your contention that it’s a caution or weakness. But the fact that correlations between attractiveness and WHR and BMI were similar in Thornhill and Grammer suggest that this is not such a big deal.

    You write that the study further shows that fat distribution as measured by WHR influences attractiveness, but it may well have actually been another measure of fat distribution, like waist size that influenced attractiveness. This paper is not sufficient to show that fat distribution as gauged by WHR per se influences judgments of attractiveness, because it did not use a variety of waist and hip sizes. There are other measures of fat distribution: waist circumference, skinfold thickness in the central abdomen, etc.

    I would actually argue that both eye tracking studies seem to support the role of abdominal fat as an important factor in judgments of women’s attractiveness. If abdominal fat in general influences attractiveness judgments, but evidence for WHR is mixed, that could potentially reconcile Cornelissen et. al and this paper.

    • whooke
      August 3, 2009 at 12:49 pm

      Hi Nadia,

      Thanks for the thoughtful comments! I agree with your caution that WHR might not actually be the relevant visual cue for raters in this study (though it was intentionally manipulated). Thanks for the reminder! About your other main points: I do continue to assert that WHR matters. I’ll quote the Smith 2007 paper you just referenced: “although WHR can be significantly correlated with attractiveness judgements, it is a much weaker cue than BMI.” (K.L. Smith et al. / Evolution and Human Behavior 28 (2007) 48–54. Quotation from page 52). It’s a weaker cue but is relevant, at least in some instances. In the end, neither (body fat or body fat distribution based on WHR) is close to a comprehensive explanation. The real evaluations are often based on more subtle aspects of bodies.

      I do have serious concerns about drawing conclusions about what is and is not beautiful when a full range of body types/face types/etc. are not in the stimulus sets for a number of reasons. Probably the most compelling for me is that restricted ranges conceal relationships between variables that exist when the range is not restricted. For example, there is a correlation between IQ and income when IQ varies e.g., from 60-140 in representative samples of Americans. If we limit the IQ range to any 10 point range we won’t find it.


  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: