AI approach suggests facial expressiveness is a stable trait

Most of us have an intuitive idea of what facial behaviour entails, and the benefits being a master of it might convey. A stoic expression certainly has its advantages in poker, for example, but the majority of us can deploy a well-timed smile when the moment calls for it.

Studying facial behaviours, however, has historically been tricky. As Eithne Kavanagh and colleagues note in their latest paper, humans have the most complex facial musculature of any animal, and capturing all that movement and meaning in a decently-sized study sample has previously been too time-consuming to be workable.

In their new study, published last month in Scientific Reports, the team employs modern solutions this age-old problem. Using AI-powered software, they take some of the first steps towards quantifying facial behaviour. By combining this method with human judgements, they find evidence facial expressiveness a stable individual difference, and that those with more expressive faces enjoy various social advantages.

For their first experiment, the Nottingham-based team recruited 52 participants to participate in semi-structured video calls, during which they chatted with a confederate. Between telling jokes, navigating embarrassment as the confederate 'accidentally' muted their mic, negotiating their share of a set reward and more, the team collected a wealth of facial behaviour data across different conversational contexts.

Six months after this call, 34 of these participants also uploaded videocall-style clips of themselves trying to achieve a number of social goals that would benefit from facial expressions, such as giving a friendly greeting or reassuring a friend. These clips were viewed and evaluated by 172 further participants, who rated them on various social dimensions.

Analyses showed that, generally, those who were more facially expressive were rated to be more likeable and agreeable. Not only that, but those with more facial expressivity were thought to have greater success in situations that required negotiation (though only if they were also agreeable).

Beyond just general expressiveness, those with more readable facial expressions were highly rated, too, and were considered to be more well-liked than their less-readable counterparts.

Though there's some debate about how to define a trait, in this case the team adopted the definition that a trait can be defined when behaviour is determined more by individual differences than environmental context. Participants were pretty consistent in their level of facial expressiveness; scores on measures of facial expressivity, as created by the AI-powered facial analysis software, didn't significantly change across the various social contexts and conditions in these clips. This finding in particular seems to indicate that facial expressiveness could be considered a trait.

In a second study involving 1,315 new participants, the team was able to replicate the above findings using informal, totally unstructured video chats between pairs of participants. They also found that facial expressivity was associated the Big Five dimensions of agreeableness, extraversion, and neuroticism — though the ties between these aspects of personality and facial behaviour remain to be explored in detail.

The authors believe that their findings overall suggest facial expressiveness can be considered a "stable individual difference that proffers social advantages." Researching facial behaviour (and, of course, how it's received) in ecologically valid contexts could therefore offer a wealth of new and exciting insights on the dynamics of social interactions. The team hopes that similar efforts which make use of modern tech could lay the path towards the tantalising idea of a "unified theory of facial behaviour."

Read the paper in full:

Kavanagh, E., Whitehouse, J., & Waller, B. M. (2024). Being facially expressive is socially advantageous. Scientific Reports, 14(1), 12798. https://doi.org/10.1038/s41598-024-62902-6