Researchers are starting to take a closer look at what people are posting on social media and why – and they are finding some interesting things.
Chris Danforth, a professor of mathematics at the University of Vermont, teamed up with Harvard researcher Andrew Reece on one such project to examine how the photos people post on Instagram reflect their mental health, specifically depression.
Danforth spoke to Vermont Edition on Thursday to elaborate on how this research was conducted and some of its findings.
"We used a popular service offered by Amazon, actually, where people can fill out surveys online," Danforth says regarding how they found adult participants for this study. "And we posted an advertisement to try and recruit people who were active users of social media, and Instagram in particular, and had also been given a diagnosis of depression at some point over the last few years."
Danforth explains the researchers didn't look at the participants' photos themselves, but instead had a computer analyze them for certain things, such as the spectrum of color and the number of faces in pictures.
"What we tried to bring to this project was some new methods in a field called 'machine learning' where you try and give a computer a whole bunch of data – maybe it's visual data; in the case of the pictures, it was colors and, you know, whether there were faces in the picture – and try and learn something about it," he says.
"What we tried to bring to this project was some new methods in a field called 'machine learning' where you try and give a computer a whole bunch of data ... and try and learn something about it." - Chris Danforth, UVM mathematics professor
Danforth discussed some of the findings based on the analysis of the Instagram photos that were looked at by the computer.
"People who have studied depression in the past have found that people who are depressed tend to prefer darker colors," Danforth explains. "And so, we weren't really sure what we would find with respect to that, but we did find – the computer algorithm found – that people who were depressed were posting pictures that tended to be bluer and grayer on this color spectrum, and darker.
"It found that they were a little bit less likely to use filters. But when they did [use] a filter, it was a black-and-white one. So there were some visual cues that the algorithm picked up on that were predictive of depression with respect to, you know, a control population of Instagrammers and their pictures, which tended to be brighter and more colorful."
"The algorithm did pick up on other features," Danforth adds. "It found that people who were depressed were more likely to post pictures with faces in them. But when they did, they were likely to have less faces in them, as almost an indication maybe that their social environment was diminished somehow with respect to the healthy participants."
"We did find – the computer algorithm found – that people who were depressed were posting pictures that tended to be bluer and grayer on this color spectrum, and darker." - Chris Danforth
Danforth also talked about some other findings of the study, such as how other humans rated the Instagram pictures, and also the value in the possibility of being able to screen one's phone activities with a computer algorithm for health purposes.
The study – which can be found here – has not yet been independently-reviewed, and Danforth acknowledges the relatively small and particular group of participants, adding that trying it with a larger group is something he'd be interested in going forward.
Danforth also touched on what he generally thinks the future may hold in this growing field of looking at social media for research material.
"I think now with all of the mobile technology and communication happening on the internet, the social sciences – political science and psychology – you know, we're going to learn a lot more about how people behave, how we interact, why we make the decisions we make ... it's an exciting time to be working on these types of problems. And I think there's a lot of promise for public health surveillance."
Listen to the full interview above.