Researchers at USC, along with psychologists from Temple University, have concluded their three-year study on the likelihood of individuals to trust others. Conducted at USC’s Institute for Creative Technologies, the study found that individuals are unreliable evaluators of other’s trustworthiness.
“Trustworthiness can be objective or subjective,” said Gale Lucas, the researcher who lead the study. “Objective is whether a person acts in a trustworthy manner in a situation, whereas subjective is whether that person is perceived as trustworthy.”
People are accustomed to making assumptions when first meeting an individual, according to the study. Assumptions range from knowledge level to morality, a phenomenon called thin slicing. One of the routes of thin slicing is that of trustworthiness.
“We often make judgements about whether we can trust someone during a social interaction,” Lucas said. “These range from everyday interactions with acquaintances to ‘bigger’ moments like buying a car.”
Using computer modeling, the researchers tracked the facial and body movements of approximately 300 participants who were placed into pairs to role-play negotiations as antique shop owners. Participants were given the value of each item and tasked with reaching an agreement about six antique items.
At the end of the allotted time, each individual rated their partner on how honest they thought their partner was. By figuring out how many lies each individual had told during the task, experimenters were able to establish a baseline for which individuals were honest and which individuals were not.
Participants believed that overall positivity, controlled smiles and head pose variability — the amount of which each person moved his or her head — represented trustworthiness. However, these expressions were found to be linked to untrustworthiness. Smiling joyously, researchers noted, indicated that a person was more likely to be untrustworthy — even though participants did not pick up on this during the study. Contempt was seen as indicative of untrustworthiness, but in reality was linked to trustworthiness. The only criteria that individuals correctly believed indicated trustworthiness was the amount of time someone spoke.
“In our context, a negotiation between two people, we found that these non-verbal signals that people use are not the ones that actually predict whether someone is lying or not,” Lucas said. “The only signal that people cued in on correctly was how much the other person talked. People who talk a lot are perceived as less trustworthy, and indeed, they are more likely to lie.”
The study concluded that people often depend too much on facial expressions when determining someone’s trustworthiness due to poorly evaluating behaviors and actions.
According to Lucas, the study has real-world implications.
“There is some potential use for this research to build computational systems that can use this information,” Lucas said. “The computational models described in the paper could be used to inform development of an agent that could assist users in making better judgements. A user could be ‘reminded’ by a computer agent to question their judgements of others’ trustworthiness when they are basing them on these ‘intuitive’ signals like positive facial expression.”