Can You Spot Which Photo Is Fake? Most People Can’t

Sophie Nightingale / University of Warwick
Sophie Nightingale / University of Warwick

In a digital world, it’s easier than ever to fool people. Sophisticated Photoshop jobs, social media, and viral news cycles mislead readers into mistaking shots from a Lebanese music video for real scenes of destruction from Aleppo, thinking that Vladimir Putin was the center of attention at the G-20 summit, or believing that Elizabeth Taylor and Marilyn Monroe posed together for a photo shoot in the park.

While it would be nice to tell ourselves that we would never be duped by such fake images, the truth is, most people can’t distinguish between a manipulated photo and a real one. That’s the takeaway from a new study in Cognitive Research: Principle and Implications. As the team at Science reports, the participants were only able to pinpoint fake images two-thirds of the time.

First, psychologists from the University of Warwick asked more than 700 volunteers to look at real and fake images and identify the changes. The researchers used 10 color photographs sourced from Google searches, manipulating them through airbrushing, adding elements in, subtracting elements, and distorting shadows, and shearing trees. They applied each of these five manipulation techniques separately to a portion of the photos, eventually creating 30 manipulated photos and 10 real ones. All the participants saw one of each of the manipulation types in different photos.

Can you spot the differences between the manipulated image at the top of the page and the original version above?Sophie Nightingale / University of Warwick

The participants performed slightly above chance rates, identifying photos correctly as real only 58 percent of the time and spotting manipulations 66 percent of the time. Even when they did identify a manipulated photo, though, they didn’t necessarily know where it had been altered.

In a second study, the researchers did the same thing, but using photos study co-author Sophie J. Nightingale took with her Nikon camera, controlling for the fact that images found online could be manipulated before the researchers even downloaded them. They then had almost 660 people take an online survey testing their ability to spot fakes. They had to look at photos and label whether it was fake and if they could see where it was manipulated, whether it was fake but they didn’t know where it had been altered, or whether it was an original. At the end of the study, the subjects identified just 62 percent of the fake images correctly.

The first image is the original. The second was manipulated to add in a water spout, airbrush the woman's face, and make other slight changes.
Sophie Nightingale / University of Warwick

The results were the same regarding images that had been manipulated in both overtly unrealistic ways and photos that featured more plausible changes. One reason might be the way that our visual system simplifies information. As long as object geometries and shadows are roughly correct, our eyes accept them as accurate.

“It remains to be determined whether it is possible to train people to make use of physically implausible inconsistencies,” the researchers write. “Perhaps one possibility would entail ‘teaching' the visual system to make full use of physical properties of the world as opposed to automatically simplifying them.”

You can still take a 10-minute online survey for the project here and test your own manipulation awareness skills. (I had to take wild guesses on most of them.)

If this makes you weep for the future of the world, at least know that it’s a timeless problem. Manipulated, misleading images have been around since the earliest days of photography.

[h/t Science]

Friday’s Best Amazon Deals Include Digital Projectors, Ugly Christmas Sweaters, and Speakers

Amazon
Amazon
As a recurring feature, our team combs the web and shares some amazing Amazon deals we’ve turned up. Here’s what caught our eye today, December 4. Mental Floss has affiliate relationships with certain retailers, including Amazon, and may receive a small percentage of any sale. But we only get commission on items you buy and don’t return, so we’re only happy if you’re happy. Good luck deal hunting!

3D Map Shows the Milky Way Galaxy in Unprecedented Detail

ESA
ESA

It's our galactic home, but the Milky Way contains many mysteries scientists are working to unravel. Now, as The Guardian reports, astronomers at the European Space Agency have built a 3D map that provides the most detailed look at our galaxy yet.

The data displayed in the graphic below has been seven years in the making. In 2013, the ESA launched its Gaia observatory from Kourou in French Guiana. Since then, two high-powered telescopes aboard the spacecraft have been sweeping the skies, recording the locations, movements, and changes in brightness of more than a billion stars in the Milky Way and beyond.

Using Gaia's findings, astronomers put together a 3D map that allows scientists to study the galaxy in greater depth than ever before. The data has made it possible to measure the acceleration of the solar system. By comparing the solar system's movement to that of more remote celestial objects, researchers have determined that the solar system is slowly falling toward the center of the galaxy at an acceleration of 7 millimeters per second per year, The Guardian reports. Additionally, the map reveals how matter is distributed throughout the Milky Way. With this information, scientists should be able to get an estimate of the galaxy's mass.

Gaia's observations may also hold clues to the Milky Way's past and future. The data holds remnants of the 10-billion-year-old disc that made up the edge of the star system. By comparing it to the shape of the Milky Way today, astronomers have determined that the disc will continue to expand as new stars are created.

The Gaia observatory was launched with the mission of gathering an updated star census. The previous census was conducted in 1957, and Gaia's new data reaches four times farther and accounts for 100 times more stars.

[h/t The Guardian]