A pioneering study has highlighted our inability to detect fake images. In fact, the way that our mind is wired, we manipulate our memories to fit an image, even if that image is bogus.
How do you know that an image you’re looking at is a fake? Is it the suspicious cluster of pixels? Or the conveniently shaky photography?
Maybe it’s just that gut feeling you get when you see a photo of your friend and their new sports car that tells you that something just isn’t quite right.
Well, that gut instinct may not be as instinctive as you’d hope.
A study by the University of Warwick on our ability to detect manipulated images has found that people can detect a fake image 60% of the time. Even then, they can only tell what exactly is wrong with the image 45% of the time.
PhD student and lead author Sophie Nightingale says the findings show that “although people perform[ed] better than chance at detecting and locating image manipulations, they are far from perfect.”
Nightingale believes this has serious implications in our highly-connected world, with the high level of images – including possibly fake ones – that we see on a daily basis online and in the media.
Images have a powerful influence on our memories, so if people can’t differentiate between real and fake details in photos, manipulations could frequently alter what we believe and [what] we remember.
The researchers created an online test using a collection of 40 images they had created from 10 original images they had sourced from Google Images. Six of the images were subject to five different types of editing, including physically-implausible and physically-plausible manipulations, creating 30 edited images. Ten random images – including each of the five manipulation types and five original images – were shown to 707 participants.
Participants never saw a manipulation or original form of the same image twice.
Researchers found a mean of 60% of images were correctly identified as being manipulated, responding to the question of “do you think this photo has been digitally altered?”
This was just over the chance performance of 50%.
Out of those people that answered “yes,” only a mean of 45% of manipulations could be correctly located when a grid overlay was placed on the image and the participants were asked to select the areas where editing was present.
Study co-author Dr. Derrick Watson explains that the study found people were better at detecting physically-implausible manipulations, but not any better at locating them compared to the physically-plausible changes.
“Even though people are able to detect something is wrong, they can’t reliably identify what exactly is wrong with the image. Images have a powerful influence on our memories, so if people can’t differentiate between real and fake details in photos, manipulations could frequently alter what we believe and [what] we remember.”
659 people took part in a second experiment which tested their ability to locate manipulations regardless of whether or not they said there was one present.
Participants’ ability to detect something wrong with the image was similar to the first experiment, with a mean of 65% of the time, but manipulations were accurately located 39% more of the time than expected by chance.
This result suggests that people are better at the more direct task of locating manipulations than the more generic task of detecting if a photo has been manipulated or not.
“People’s poor ability to identify manipulated photos raises problems in the context of legal proceedings, where photos may be used as evidence,” explains co-author Dr. Kimberley Wade. “Jurors and members of the court assume these images to be real, though a manipulated image could go undetected with devastating consequences.”
“We need to work to find better ways to protect people from the negative effects of photo manipulation, and we’re now exploring a number of ways that might help people to better detect fakes.”
The findings were published in the open access journal Cognitive Research: Principles and Implications.