The Washington PostDemocracy Dies in Darkness

Your Instagram feed can tell us if you’re depressed, study suggests


If you’ve ever paused to reflect on the moody, filtered shots of a friend’s Instagram feed and wondered whether you should reach out, a new study may confirm your worries.

Researchers Andrew Reece of Harvard University and Chris Danforth of the University of Vermont collected survey information and 43,950 Instagram photos from 166 volunteers recruited through Amazon’s Mechanical Turk, which pays people small amounts to perform tasks.

Each participant was asked to complete a standardized clinical depression survey to assess depression level, answer demographic questions and share information about their use of social media, including their Instagram usernames. They also were asked to self-report their history of depression diagnosis.

The researchers then analyzed the photos for quantitative measures of colors, brightness and faces, as well as more-subjective assessments of happiness, sadness, likability and “interestingness.”

The study is the latest in a body of research that looks at how changes in your psychology are reflected in social media. For example, a team from Northwestern University’s Feinberg School of Medicine examined how three aspects of movement in time and space — aspects they dubbed  “circadian movement,” “normalized entropy” and “location variance” — appear to correlate with symptoms of depression. Another study, out of Sweden, found that frequent cellphone use was associated with stress, sleep disturbances and symptoms of depression among both young adult men and women.

How your cellphone knows if you’re depressed. It has to do with how you move through time and space.

The key finding in the new study has to do with the curious relationship between mood and color. It turned out increased hue, with decreased brightness and saturation, appeared to predict depression. Hue refers to the shade on a 360-degree color wheel. Brightness is defined as where on the spectrum a color lies between black to white. Saturation is what photographers call “colorfulness,” or the intensity of a color.

In other words, people who are depressed had pictures that were “bluer, grayer and darker.”

Below is what that looks like:

Depressed participants in the study were also more likely to post more frequently and to apply more and different types of Instagram filters.

The most popular one for people who are depressed? Inkwell, which turns color photos black and white. “Healthy” participants — people the study defined as not being depressed — tended to favor the Valencia filter, which, the study said, “lightens the tint of photos.”

This chart of filter usage by volunteers who are depressed vs. those who are not is striking.

One of the most fascinating findings has to do with faces. Researchers found that depressed users were more likely than others to post photos with faces. However, they had fewer faces in each photo.

“Fewer faces may be an oblique indicator that depressed users interact in smaller social settings, which would be in accordance with research linking depression to reduced social interactivity,” Reece and Danforth wrote.

The study also involved asking volunteers to rate the photos posted by other people in terms of happiness, sadness, likability and interestingness. As might be expected, the photos from depressed participants’ feeds were more likely to be sad and less happy.

One important thing to note is that the researchers were able to detect these “depressive signals,” as they called them, in posts made even before the date of first diagnosis.

The researchers wrote that these findings “suggest new avenues for early screening and detection of mental illness.” It's unclear how that would work. Would you one day give your clinician your Instagram username, or would bots scan millions of Instagram feeds and somehow report back to you that you might be depressed?

Reece and Danforth’s work was released on arXiv, an open-access service run by Cornell University that allows scientists to share research before it’s formally published, based on the idea that it can accelerate the pace at which others can build on the knowledge.

The findings have not been independently reviewed, so you should take them with a grain of salt. There’s also the issue that the volunteers were crowdsourced through Mechanical Turk, which tends to attract those with a lot of free time on their hands rather than, say, people who put in 60-hour workweeks.

(Amazon chief executive Jeffrey P. Bezos owns The Washington Post.)

Read more:

IN DEPTH: Sean Parker, Silicon Valley’s bad boy genius, wants to kick the *!$% out of cancer

$250 million, 300 scientists and 40 labs: Sean Parker’s revolutionary project to ‘solve’ cancer

Brain-zapping gadgets promise to make you a better you — smarter, stronger, even happier

Zika babies may have a spectrum of brain damage beyond microcephaly

Like our Health & Wellness page on Facebook for more science news about the ins and outs of the human body and mind, essays and advice. You can sign up here for our newsletter.