Beauty, novelty and threat

By Gordon Rugg

In an earlier article in this short series, we looked at what happens when you treat two concepts such as liking and disliking as two separate scales, rather than as opposite ends on a single scale. The answer was that often this makes sense of results which would otherwise look contradictory.

In another article, we looked at what happens when you apply this approach to the well established literature on perceptions of attractiveness. The result was that this provided a clear, simple way of explaining an apparent paradox within the literature on perceptions of attractiveness in human faces.

Covering these topics raised a lot of other questions, which we’ll tackle in this article. The questions relate to three main themes:

  • Social context
  • Cognitive load
  • Perceptions of novelty and threat

Social context

There’s been a lot of research into attractiveness, and much of that research has involved perceptions of female attractiveness. There has been less research into male attractiveness, and even less research into perceptions of human ugliness. Also, much of the early work on averaged faces reflects common social biases of the time, such as averaging Danish faces separately from Swedish faces, but readily lumping together diverse non-European ethnic groups into e.g. an “average Mexican face” or “average Chinese face”. None of this is likely to surprise anyone with even a superficial knowledge of sociology or social anthropology or women’s studies or media studies. I won’t go further into this topic here. Instead, I’ll work through some of the issues raised in the attractiveness literature, to show how they tie in with more recent work on the uncanny valley, as well as how they can be represented using this two-concept plot approach.

Some of the underlying mechanisms proposed for perceptions of attractiveness and ugliness relate to evolutionary biology, and in particular to honest costly signals of health (e.g. Rhodes et al, 2001; Foo et al, 2017). This makes sense in terms of, for instance, skin blemishes being associated with ugliness, since they might relate to an underlying health problem. However, making sense is not necessarily the same as being true, which raises the question of whether there might be other, better, explanations for some or all aspects of attractiveness.

An example is that humans prefer symmetrical faces to asymmetric faces, which could arguably be due to bodily symmetry reflecting health. This preference also appears in insect preferences for faces of other insects from the same species (e.g. Rodriguez et al, 2004) which could be argued as further proof of an underlying cause in evolutionary biology.

However, the same findings can be explained in terms of a very different mechanism, namely cognitive load, i.e. the amount of mental processing required to handle a given set of sensory input, such as a picture of a face.

Cognitive load

Numerous researchers have pointed out that making judgments about a symmetrical, unblemished face requires less mental processing than making judgments about asymmetric and/or blemished faces (e.g. Reber et al, 2004; Winkieman et al, 2006). This is also consistent with findings about preferences for averaged images of objects other than human faces, such as automobiles, where evolutionary biology is not a very plausible mechanism (e.g. Halberstadt & Rhodes, 2003). An excellent example of this approach being applied to aesthetics in general is a classic article by Ramachandran & Hirstein (1999). This model is also consistent with findings from using machine learning to identify predictors of human facial attractiveness (e.g. Kagian et al, 2008).

An obvious apparent objection both to the evolutionary explanation and to the cognitive explanation involves changes in aesthetic preferences across cultures and across time. At first glance, both the evolutionary and the cognitive mechanisms would seem to select for a single optimal form of attractiveness, but this is clearly not what happens in human societies, where there are usually several different forms of preferred beauty.

However, a simple and powerful explanation for this apparent paradox is described in Ramachandran & Hirstein’s 1999 article. In brief, the brain needs input to stay at an optimal level of functioning; novel stimuli (e.g. an unusual face) provide this input. Similarly, at a social level, ownership of unusual items is usually a marker of higher social status (e.g. Rogers, 2003) and attractive people are more likely to be imitated than less attractive people (van Leeuwen et al, 2009). So, novelty and interestingness can add attractiveness, even if the novelty and interestingness are achieved via a departure from the classic features associated with beauty. This makes further sense of an apparent paradox covered in the previous article in this set, namely that averaged faces are attractive, but very attractive faces are not average (Alley & Cunningham, 1991).

Novelty, interestingness and threat

Novelty may bring interestingness and more attractiveness. However, novelty may also be threatening.

This is where we loop back round to sociology and social anthropology and media studies, via themes such as infantilisation of women in popular culture, plus Necker shifts and uncanny valleys. There’s been a lot of solid research into the interactions between traditional models of femininity and sending out non-threatening signals. There’s also been a lot of good popular work within the TV tropes tradition on conventions for portraying women as strong and/or threatening in popular culture.

Most of that work has used text or tables as the medium for analysing these issues. What I’ll do here is to look at what happens when you plot reassuring safe signals against threatening signals, using the format below, and applying it to three photos of women with blonde hair.

Some forms of novelty send out strong signals of safety, without any threatening signals; for instance, this image of a smiling young woman (model Melvnin) with an unusual combination of hair colour, eye colour and skin colour. The combination is unusual, but the individual features are common. Used under fair use academic terms

Other forms of novelty/rarity send out strong signals of threat, such as this image of a scowling old woman with an eye patch and disheveled hair (played by Helena Bonham Carter) which taps into assorted tropes about witches and threatening mad women. Used under fair use academic terms

Then, in the top right quadrant, we get strong signals both of safety and of threat, as in the picture below of Yolandi Visser. The facts that she is a young woman, with neat makeup, and a symmetrical face, should send out signals of safety and attractiveness. However, the black lipstick combined with the black contact lenses tap into well established tropes about aliens, demons and other threats. Her facial expression is neutral, leaving the viewer with an uncomfortable set of conflicting signals which can’t be resolved. I’ve written about non-resolvable facial signals here. Used under fair use academic terms

With safety/risk, uncertainty is an important variable, particularly when there is strong evidence in both directions, as in the uncanny valley. We’ve blogged about this before, in relation to Necker shifts, when the viewer switches from seeing the situation from one perspective to seeing it in an utterly different perspective. It’s no accident that this is a key dynamic in horror stories and horror movies.

Returning to representing signals of safety and threat, the three images can be located in terms of safe and threatening signals as shown below.


For dealing with tropes about attractiveness and depictions of women, this is a simple but powerful representation of findings from the relevant literatures. It also raises interesting questions about what attractiveness actually is.

In terms of scales, the general assumption in the past was that attractiveness and ugliness are opposite ends of a single scale, with attractiveness at the positive end, and ugliness as a negative value opposite to attractiveness.

However, it’s also possible to argue that attractiveness and ugliness are two separate privative scales, with the features that constitute ugliness not always being the opposite of the features that constitute attractiveness.

More intriguingly, perceived attractiveness may be nothing more than a sweet spot between the values for ease of cognitive processing, novelty, and perceived threat. This raises a lot of interesting questions about the focus of research for work relating to aesthetics. Cognitive load and novelty can be quite easily modeled using well-established mathematical and computational approaches (e.g. Kagian et al, 2009, and Shannon, 1948, respectively), but perceptions of threat are not so tractable. Arguably, to understand aesthetics better, we need to understand threat better, which raises some very interesting research questions.

Some of those questions can be tackled simply and effectively by using the two-scale approach that we’ve examined in this series of articles. Risk assessment, for instance, often involves weighing apparently contradictory evidence, and the safe/threatening plot provides a quick, simple way of showing where a risk or risks should be located.


There can be a lot to gain from asking whether two apparent opposites would be better treated as two separate concepts, and then plotted against each other.

In some fields, this approach (or something similar) is well established. In others, though, this approach hasn’t yet been widely used, and there are rich opportunities for early adopters.

Notes, references and links

You’re welcome to use Hyde & Rugg copyleft images for any non-commercial purpose, including lectures, provided that you state that they’re copyleft Hyde & Rugg.

There’s more about the theory behind this article in my latest book: Blind Spot, by Gordon Rugg with Joseph D’Agnese 

You might also find our website useful:

Related articles about uncertainty, risk and horror:

Overviews of the articles on this blog:


Alley, T.R. & Cunningham, M.R. (1991), Averaged Faces Are Attractive, but Very Attractive Faces Are Not Average. Psychological Science, 2(2), pp 123-125.

Foo, Y.Z., Simmons, L.W. & Rhodes, G. (2017). Predictors of facial attractiveness and health in humans. Scientific Reports, DOI: 10.1038/srep39731.

Halberstadt, J.B., & Rhodes, G. (2003). It’s not just average faces that are attractive: Computer-manipulated averageness makes birds, fish, and automobiles attractive. Psychonomic Bulletin & Review, 10, 149–156.

Kagian, A., Dror, G., Leyvand, T., Meilijson, I., Cohen-Or, D. & Ruppin, E. (2009). A machine learning predictor of facial attractiveness revealing human-like psychophysical biases. Vision Research, 48, pp 235-243.

Ramachandran, V.S. & Hirstein, W. (1999). The science of art: A neurological theory of aesthetic experience. Journal of Consciousness Studies, 6, pp 15-51.

Reber, R., Schwarz, N. & Winkielman, P. (2004). Processing fluency and aesthetic pleasure: Is beauty in the perceiver’s processing experience? Personality and Social Psychology Review, 8(4), pp. 364-382.

Rhodes, G., Zebrowitz, L.A., Clark, A., Kalick, M., Hightower, A. & McKay, R. (2001).

Do facial averageness and symmetry signal health? Evolution and Human Behavior, 22, pp. 31-46.

Rodriguez, I., Gumbert, A., de Ibarra, N.H., Kunze, J. & Giurfa, M. (2004). Symmetry is in the eye of the ‘beeholder’: innate preference for bilateral symmetry in flower-naïve bumblebees. Naturwissenschaften, 91, pp 374-377.

Rogers, E.M. (2003.) Diffusion of Innovations (5th edition). Free Press, New York.

Shannon, C.E. (1948). A Mathematical Theory of Communication. The Bell Systems Technical Journal, 27, pp 379-423, 623-656.

van Leeuwen, M.L., Veling, H., van Baaren, R.B. & Dijksterhuis, A. (2009). The influence of facial attractiveness on imitation. Journal of Experimental Social Psychology, 45(6), pp 1295-1298.

Winkielman, P., Halberstadt, J., Fazendeiro, T. & Catty, S. (2006). Prototypes are attractive because they are easy on the mind. Psychological Science 17(9) pp 799-806.



5 thoughts on “Beauty, novelty and threat

  1. Pingback: Modeling misunderstandings | hyde and rugg

  2. Pingback: Content analysis, part 2 | hyde and rugg

  3. Pingback: Tacit and semi tacit knowledge: Overview | hyde and rugg

  4. Pingback: Explicit and semi-tacit knowledge | hyde and rugg

  5. Pingback: Tacit knowledge: Can’t and won’t | hyde and rugg

Leave a Reply

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

You are commenting using your 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

This site uses Akismet to reduce spam. Learn how your comment data is processed.