Photo or fraud? New study says we can't spot fake pics as well as we think
Getting duped by digital alteration can have dangerous consequences, too.
You've been tricked by a photo before. Whether it's part of a spam ad underneath an article, or some bizarre story your uncle just posted on Facebook, there was at least a split second where you thought the image was real before realizing that you were duped by a digital alteration. You may think you're an expert at spotting fake news, but spotting a fake photo is far harder than you think because, as it turns out, we're not that good at it. A study recently published in Cognitive Research: Principles And Implications suggests not only that the human eye isn't great at spotting digital alterations of real life images, but that we're even worse when it comes to identifying exactly what's wrong.
To build the study, researchers started with 10 actual images of people in real-world scenes that were found on Google Images. Six of the 10 were given five different types of photo manipulation; airbrushing (like whitening teeth and removing blemishes), addition or subtraction (adding or erasing objects), geometrical inconsistencies (physically unlikely adjustments, like the angles of buildings), shadow inconsistencies (angling a shadow so it's incongruent with the rest of the image) and super-additive (meaning all of the different types of manipulation were used on one photo). This created a roster of 40 images, 10 of which were real and 30 of which were altered.
707 participants took the online test in which they were shown 10 of these images at random, half of which were altered (though they were never shown the same base image twice). When asked if they thought each image was digitally altered, participants answered correctly an average of 60% of the time. Of the participants who believed an image had been altered, only 45% of them could accurately locate the actual manipulation. A further experiment tested 659 participants with all altered images. These results showed that participants did noticeably better at identifying what was altered in the image, leading researchers to suggest that perhaps it's easier for participants to spot what is wrong when they already know an alteration took place.
Researchers also inferred that our inability to detect photo manipulations can have very significant real-life consequences. For example, in a courtroom, an undetected doctored image could be case-altering evidence. But the trickiness of these pictures is already causing complications in our everyday lives, especially with the instantaneous nature of social media. Photo manipulation technology is readily available to everyone and the lack of a keen public eye can have some dangerous results.
In 2015, a Canadian man found out via Twitter that someone had altered his selfies to make it appear as if he was wearing a suicide bomber vest and holding the Qur'an. The image was then used to falsely support the claim that he was one of the Paris suicide bombers. The image spread so quickly that by the time it could be debunked, it's almost too late. This issue also ties into the unavoidable reality of "fake news", wherein deceptive people and organizations can essentially throw a pseudo-bone that we, as the public, hurriedly chase after without questioning its authenticity. As these information wars make our newsfeeds their battlegrounds, we look to those social media companies to help us combat it. Facebook, for example, is introducing tools to help combat the spread of misinformation on their platform, making it easier for users to report fake news, which really only puts pressure on the public to be as discerning as possible.
There's no infallible way for us to spot fake photos, as technology always seems to outpace the human understanding. And more than a simple airbrush or filter, manipulated photos can alter our perception of reality. Hopefully, knowing our weakness is the first step to understanding how we can deal with this and not be so trusting of every eye-catching image that pops up on our feeds.
The images in this article were updated on October 17, 2018.