Can We Read Someone’s Trustworthiness in Their Face?

by Bastian Jaeger

In his autobiography, Charles Darwin recounts the time he approached the captain of the HMS Beagle. Darwin was keen on joining the crew on an upcoming trip around the world as the expedition’s naturalist, but he needed the captain’s approval. Despite recommendations by friends, the captain was reluctant at first. He thought that Darwin wasn’t cut out for the long journey – because of the shape of his nose. The captain was a strong believer in physiognomy, the idea that a person’s character is revealed in their facial appearance. Luckily, he ultimately decided to give Darwin a chance after all.

Physiognomy has a long history in scholarly thought, dating back to at least ancient Greece. Today, the idea that a person’s nose is indicative of their character may seem laughable. In scientific circles, physiognomy and phrenology (the practice of inferring character traits from skull measurements) are widely regarded as pseudoscience. But decades of research on first impressions also suggest that we are all physiognomists to some degree. We might not measure a person’s nose, but we still tend to form judgments about others based on their facial appearance. Research shows that when we meet a stranger, we immediately form an impression based on their facial appearance. This process seems to occur relatively automatically –  it only takes a few hundred milliseconds to judge whether a person seems trustworthy or not. These split-second judgments can be very consequential because people rely on them when making many important decisions. A person’s facial appearance can influence if they are hired for a job, found guilty in court, or trusted in a social interaction.

Can we actually tell how trustworthy someone is based on their facial appearance? This is what we set out to test. We were not the first to tackle this question, but previous research on the accuracy of trustworthiness impressions has yielded mixed results. Some researchers found that people can discriminate between trustworthy and untrustworthy individuals based on a facial photograph with slightly-higher-than-chance accuracy. Other studies yielded no evidence for accuracy. The plausibility of accurate trustworthiness detection has also been questioned (why wouldn’t people evolve to look trustworthy but then exploit the trust of others by acting selfishly?) and it is unclear which facial features “give away” a person’s trustworthiness.

It seems like more evidence is needed to address this question, so in a recent article, we set out to test the accuracy of trustworthiness impressions while also addressing some limitations of previous work. Misplaced trust can be consequential in everyday life. In our studies, we therefore let participants make decisions with tangible outcomes. Similar to previous studies, we let participants play a trust game. We gave some money to a sample of trustors, 131 students from the University of Zurich, who decided whether to keep the money or send it to a trustee. If the money was sent, it was tripled and the trustee had to decide if they want to split the tripled amount with the trustor (reciprocating their partner’s trust) or keep all of the money for themselves (betraying their partner’s trust). Trustors interacted with 31 different trustees, also students at the University of Zurich, and they saw a photo of the trustee’s face on each trial.

Replicating previous work, we found that trustors relied on the facial appearance of their interaction partners when deciding whom to trust. Trustees who were rated as more trustworthy-looking based on their photo were more likely to receive money from their partners. However, trustors were not able to invest their money in a smart way – they were not more likely to send money to trustees who were actually more trustworthy (those who indicated that they would return more of the money if trusted). We found similar results when analyzing how much money trustors earned across all interactions. If seeing the face of an interaction partner gives trustors a strategic advantage (because they can infer their partner’s trustworthiness to some degree) then they should be able to earn more money than a person who trusts at random. This was not the case. In fact, trustors would have earned more money by following another very simple strategy – always keeping their endowment and never trusting anyone (participants in our study turned out to be not very trustworthy).

Maybe participants didn’t send money to the trustees because they believed that they would get more in return, but because it felt wrong to signal to a stranger that they can’t be trusted? We tested this alternative explanation in another study where we simply asked participants to predict how trustworthy the photographed individuals are. We paid them when they made more accurate predictions, but again, trustworthiness judgments were not accurate. We also tested the robustness of our findings in a few other ways. Instead of showing participants cropped photos that only revealed internal facial features, we showed them the original photos that also revealed trustees’ facial structure and hairstyle. We tested detection accuracy for male and female trustees separately. We explored if trustors who are more confident in their decisions are more accurate. In all cases, we found that our participants were unable to detect the trustworthiness of their interaction partners.

Our studies suggest that modern science was right to brand physiognomy as a pseudoscience. It doesn’t seem like we can judge a person’s character just by looking at their face. First impressions may come to mind quickly, but our findings suggest that we better second-guess them. So before you trust or distrust a stranger because it feels like it’s the right decision, ask yourself what your evaluation is based on. If they simply “look trustworthy” to you, maybe find some better information first before making a decision.

Read the paper: Can people detect the trustworthiness of strangers based on their facial appearance?