Instagram photos featuring faces are more likely to attract attention on the popular photo- and video-sharing website than images with only objects or landscapes, a new U.S. study has revealed.

After analyzing 1.1 million photos on Instagram using face detection software, researchers at Georgia Institute of Technology found that on average, a photo that contained a human face receives 38 per cent more “likes” and 32 per cent more comments compared to a photo that does not contain any faces.

The study, conducted in partnership with Yahoo labs researchers, further found that the age, gender and number of faces in the photo did not impact social engagement, despite the fact that Instagram is more popular among younger people.

In total, Instagram has more than 150 million active monthly users who, according to the study, “collectively generate 1.2 billion likes per day.”

Study lead and Georgia Tech Ph.D student Saeideh Bakhshi said even as infants, people enjoy looking at faces, which could explain why such photos generate more social engagement.

“Faces are powerful channels of non-verbal communication,” Bakhshi said in an article posted to the Georgia Tech website. “We constantly monitor them for a variety of contexts, including attractiveness, emotions and identity.”

Perhaps unsurprisingly, Instagram users with more followers attracted more social activity, but the study also found that on Instagram, less is more.

Bakhshi said the more photos uploaded by one user, the lower the probability that any single photo had of getting “likes” or comments.

“The more you post, the less feedback you’re going to get,” Bakhshi said. “Posting too much decreases likes two times faster than comments.”

There are practical implications for the data, researchers say, as more academics further explore online user behaviour, and as other social media sites look to drive more consumers to their sites.

The team’s findings are to be presented April 26-May 1 in Toronto, at the ACM CHI Conference, which is an international conference on human-computer interaction.