Asian Scientist Magazine (Aug. 8, 2022) — Think about the last time you felt a stranger can be trusted. Did the face of that person look somewhat like you? A study by researchers from Osaka University in Japan found that people tend to trust others more when they possess facial features like them and are of the same sex. When a stranger is of the opposite sex, facial similarity does not affect the perceived trustworthiness of the stranger. This finding was published in Humanities and Social Sciences Communications.
Past studies used computer-generated faces with resemblances to the observer to establish that facial similarities can influence a degree of trust. However, no research looked into how this might hold in case of unfamiliar faces in real life. The researchers of this study hypothesised that the reverse relationship stands true as well: People who trust each other tend to have similarities in facial features. The team decided to test how perceived trustworthiness between people is correlated with their facial similarities using deep convolutional neural networks.
These networks use Artificial Intelligence to analyse pictures by detecting distinctive features and classifying them to evaluate similarities. The team conducted an experiment where volunteers had to rate trustworthiness of different standardised pictures of faces. The researchers then used neural networks to quantify the actual degree of similarity between the volunteer’s and the pictured face.
The findings were ‘surprising’ to the researchers: While the earlier understanding that facial similarity promotes feelings of trust was valid, this could only be applied on people of the same sex. The quality of social interactions between people of same-sex increases when they have similar facial features.
This presents an exciting development in current technology, says Professor Takuto Yamamoto, from the Osaka University’s Faculty of Medicine, since “it is possible to quantify the degree to which two faces are similar or different via face recognition performed by an artificial neural network”.
To account for potential confounds, the researchers limited the age range of the test pictures and evaluated a moderate number of 200 pictures. This study can be conducted on a larger scale to verify whether the association between facial similarity and trust can be generalised across all age groups.
The researchers also recognise that this study can help inform development of future online technologies in different fields like generating more personalised matching in peer to peer lending.
Source: Osaka University; Photo: Shutterstock
The article can be found at: Nakano et al. (2022) You trust a face like yours.