Friday, 11 April 2014

Facial expressions as social camouflage

Can making faces mask your personality?

According to a group of University of Glasgow psychologists, Daniel Gill and colleagues, it can. Writing in the journal Psychological Science, these researchers say that human facial expressions can signal how dominant, trustworthy, or attractive we are – and that these ‘dynamic’ signals can mask or override the impression given off by the ‘static’ structure of the face.

In other words, someone might have a face that ‘seems untrustworthy’, but if they make the right face, they’ll still look like someone you’d trust with your housekeys.

To reach this conclusion, the researchers made use of software that allows them to generate realistic animated face images. These ‘faces’ are programmed with 42 different sets of muscles – called ‘action units’. Each one of these units could be switched on or off independently from the others, creating billions of possible animated facial expressions – only a tiny proportion of which are likely to be seen in real life. This software has been used before in studies of emotional expressions.

Gill and colleagues generated thousands of random expressions and got volunteers to rate each one for how dominant, trustworthy, and attractive it appeared. From all of these ratings they were able to determine the essence – or prototype – of, for example, a highly trustworthy look. Which, it turns out, involves the activation of the ‘Dimpler’, ‘Lip corner and cheek raiser’, and ‘Sharp lip puller’.

Can a facial expression tell you whether somebody is a good egg or not?
Armed with these dynamic prototypes of dominance, trustworthiness and attractiveness, Gill et al then tested whether they could counteract the effects of static impressions of the same traits. They used the same software to generate thousands of static faces, got volunteers to rate them, and worked out what made someone just look trustworthy, for example.

Then, they overlaid the dynamic expressions on top of the static ones. This revealed that, in general, the dynamic expressions were more powerful than the static traits. Mathematically speaking, the effect of static structure was linear while the dynamic effect was nonlinear and larger in magnitude.

They dub this social camouflaging: ‘Even the most submissive face [was] transformed into a dominant face by social camouflaging and reaches the same level of dominance as the most dominant static facial morphology.’

As well as with trustworthiness, the same effect worked for dominance and attractiveness as well, although it wasn’t quite as effective in the latter case, suggesting that ‘facial attractiveness is more difficult to mask than are facial dominance and trustworthiness’.

This, they say, is no big surprise: ‘Casting directors are probably aware that not all social traits are equal. An attractive character will require an actor with attractive morphology; however, social camouflage can help an actor fake a dominant or trustworthy character.’

However, all of this research was based on computer-generated faces. This provided Gill and colleagues with the ability to examine a wider range of expressions than would have been possible using actual models, but it does mean that these results might need to be confirmed with real faces to verify the relationships between dynamic and static faces.


Gill, D., Garrod, O., Jack, R., & Schyns, P. (2014). Facial Movements Strategically Camouflage Involuntary Social Signals of Face Morphology Psychological Science DOI: 10.1177/0956797614522274

Post written for the BPS Research Digest by guest host Neuroskeptic, a British neuroscientist who blogs for Discover Magazine.

1 comment:

Anonymous said...

I wonder how this would work of people with prosapagnosia? As a face-blinder myself I am not sure the process I use when I judge someone's face on first sight. When I think now of someone with an untrustworthy face what is that based on when I can't even picture my mothers face in my mental eye?

Post a Comment

Note: only a member of this blog may post a comment.