Can You Judge Trustworthiness from a Face?

When going with your gut is probably good.

Posted Oct 07, 2017

Source: baranq/Shutterstock

Trust is a fundamental building block of human society. We trust not only those we know well, but also strangers we've never met before. The server brings us a meal in a restaurant, trusting we’ll pay after we’ve eaten it. We put in long hours at work, trusting that our employer will give us our wages at the end of the month. And we zip through green lights, trusting that cross traffic will stop at their red lights.

Among family members, close friends, and colleagues at work, we have a good sense of who we can trust — and how far. That’s because we’ve had plenty of interactions with them in which we’ve tested their trustworthiness. But what do we do when we have to decide whether to trust a stranger? New research by French psychologist Jean-François Bonnefon and his colleagues suggests that, when it comes to judgments of trustworthiness, our intuitions are fairly accurate guides, but they're far from foolproof.

Judgments of trustworthiness can be observed and measured in the laboratory by having two research participants — a “Trustor” and a “Trustee” — play the trust game. We give the Trustor $10 and tell them they can either keep the money or give it to the Trustee. We also explain that if they give it to the Trustee, the amount will be tripled to $30, and the Trustee can then decide whether to split the money with the Trustor or keep the entire amount.

The best-case scenario is when there is complete cooperation. The Trustor gives the Trustee all $10, which is tripled to $30, and the Trustee splits this evenly with the Trustor so that each gets $15. In other scenarios, either the Trustor keeps the full $10 and the Trustee gets nothing, or the Trustee keeps the full $30 and the Trustor gets nothing. The game is set up so that the Trustor has something to gain from trusting the Trustee, but of course the Trustee could still take the money and run.

The beauty of the trust game is that we not only get a judgment of trustworthiness, we also get an assessment of the accuracy of that judgment. Further, real money is at stake, so the Trustor has an incentive to judge the Trustee as accurately as possible. The trust game is widely used in research on economic decision-making and is believed to accurately reflect judgments of trustworthiness in real life.

Bonnefon and colleagues wanted to see how much and what kind of information was needed to form good intuitions about another person’s trustworthiness. First, they found that if you let the two participants interact on other tasks for half an hour before playing the trust game, the Trustor is fairly accurate at predicting the Trustee’s trustworthiness. Of course, in half an hour you can learn a lot about a person — you’ve got their words, their body language, their facial expressions, and maybe even some behaviors that tell you whether you can trust them.

Next, the researchers asked what would happen if the Trustor could only observe the Trustee interacting with another person. In short, could they still read the cues of trustworthiness if they weren’t directly interacting with them? The experimenters tried out two conditions: Half the Trustors observed a video of the Trustee with the sound on, and the other half with the sound off. In both cases, the Trustors were fairly good at predicting the trustworthiness of the Trustee. Apparently, it’s body language, not spoken words, that count.

The next question was: Can Trustors predict the trustworthiness of Trustees just from pictures of their faces? Here’s where the story gets interesting. When the experimenters showed the Trustors pictures of the Trustees from the chest up in full color, so that clothing and hair style were clearly visible, performance dropped to chance. However, when those pictures were cropped to show only the eyes, nose, and mouth, and then converted to black-and-white and degraded to give a grainy appearance, performance returned to above-chance levels.

So here’s one of those conundrums that delights psychologists: How is it that less information (a cropped, degraded, gray-scale photo) can yield better judgments than more information (a complete, detailed, full-color photo)? Of course, the researchers can only speculate at this point, but here’s what they think is going on.

In the full-color photos, there’s simply too much information, so the viewers are overloaded. Further, they tend to overthink the problem, weighing in irrelevant information, such as clothing and hair styles. Instead of going with quick intuitions, they engage in a slow, rational process.

However, the cropped photos hone in on the relevant facial information for conveying emotional states, namely the eyes and mouth. Further, the graininess of the cropped images encourages viewers to rely on their intuition, while the high-detail photos induce a more rational approach.

Bonnefon’s research falls in line with dual-process theory, which suggests that we have two modes of thinking — a quick, intuitive mode and a slow, rational mode. Each has strengths and weaknesses. When making snap judgments in the social realm, intuition will usually yield better decisions than reason. In other cases, reason tends to trump intuition.

An important point  the authors emphasize is that even though intuition works better than reason in making social judgments, it’s still not all that good. Performance greater than chance — which is the typical result for intuitive social judgments — means that there are still many times that intuition will lead us astray.

The bottom line for making social judgments is this: If you have absolutely no other information to go on, trust your gut; it will be right more often than not. But keep in mind that your gut is far from foolproof, so don’t ignore contextual cues of untrustworthiness — such as offers that sound too good to be true — just because you’ve got a good feeling about the person making the offer. “There’s a sucker born every minute,” the old saying goes — but you don’t have to be one of them. 


Bonnefon, J.-F., Hopfensitz, A., & De Neys, W. (2017). Can we detect cooperators by looking at their face? Current Directions in Psychological Science, 26, 276-281.