In February 2007, the Supreme Court heard a case that relied on approximately 15 minutes of video evidence from two police vehicles. dashboard cameras: Footage showed the front of a police car as it chased a driver in Georgia before crashing into the back of the driver’s car, which then crashed. The driver, Victor Harris, 19, was left paralyzed for life as a result of the incident.
Harris continued the officer who hit his car. He alleged that the officer, Timothy Scott, used excessive force. Before the Supreme Court considered the case, lower court judges had already reviewed the video footage and ruled in favor of Harris, with one writing that Harris posed little threat to the public despite his speeding.
But the Supreme Court disagreed and ruled 8-1 in favor of the officer. In the majority opinion, the justices determined that Harris posed a “real and imminent threat” to the public and wrote that “we are happy to allow the videotape to speak for itself.”
On supporting science journalism
If you enjoy this article, please consider supporting our award-winning journalism by subscribe. By purchasing a subscription, you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
The courts had examined the same video. But they came to radically different conclusions about what it showed. Indeed, the nine justices of the Supreme Court did not all agree on what they saw in the images; in a dissent, the late Justice John Paul Stevens wrote that the video “confirms, rather than contradicts” the lower courts’ ruling in favor of Harris. The episode begs the question: How can different people watch the same video and still see such different things?
“Seeing is not just what our eyes physically see,” says Sandra Ristovska, associate professor of media studies at the University of Colorado at Boulder, “but also the experiences and ideas that viewers bring to the images.”
In the years that followed Scott v. Harris, the affair became a typical example of this phenomenon and was studied by lawyers and psychologists. “Video is everywhere, from our phones to surveillance cameras on city streets. And it has become a vital form of evidence in court,” says Ristovska. And understanding how it can be interpreted differently depending on who sees it is essential to ensuring justice is applied fairly, she says.
How your brain changes what you see
One of the confounding factors in how someone watches a video is what psychologists have dubbed “slow motion bias.” In a 2016 studyresearchers showed that when viewers watched surveillance footage of a shooting in slow motion, they perceived the shooter as “more intentional.”
Likewise, if a video shakes, viewers can interpret events as more intense.
Then there’s “camera perspective bias,” says Neal Feigenson, a law professor at Quinnipiac University. In a series of studies at Ohio University, participants watched videos of people confessing to crimes. In some videos, the camera focused on a suspect’s face, while in others it focused on the interrogators. Viewers who saw videos focused on the suspects’ faces were more likely to perceive the confessions as “more voluntary.”
Eyewitness accounts can also be contaminated after the fact. If you and a friend witness a car accident, for example, and then talk about it, you might unintentionally adopt some of your friend’s memories as your own. This phenomenon is known as memory contamination, says Miko Wilford, an associate professor in the psychology department at Iowa State University.
A similar effect could occur if eyewitnesses were asked to recall an incident they also saw on video, she says.
“We’re just really bad at remembering where information comes from in our memory,” says Wilford.
When someone retrieves a memory, they’re “not reading a recording,” says Elizabeth Loftus, a psychology professor at the University of California, Irvine. On the contrary, “we construct” this memory, she says. In other words, the brain collects bits of information, sometimes from different places and times, and transforms them into memory. “Once that happens, it’s not easy to distinguish which piece came from where,” adds Loftus.
In 2016 Loftus and colleagues published an article arguing that police officers should write down their account of an incident before viewing body camera footage: If officers watch such video first, it could strengthen their memories of the details shown there, but weaken their ability to remember other information that was not captured in the recording.
Humans are particularly sensitive to visual information. More of the processing power in the brain’s prefrontal cortex is dedicated to visual information rather than audio information, Ristovska notes.
This partly explains why people generally trust video evidence, even when they know it is false. In a remarkable 2008 study of this effect, researchers asked the students to perform a gaming task on a computer. When students were falsely accused of cheating and shown a fake video of the alleged offense, the “vast majority” of students confessed “without resistance,” the researchers found.
“People intuitively tend to believe that video gives them the objective reality of what it depicts,” says Feigenson. “It is naive realism.”
How Bias Can Affect a Viewer
Cognitive biases can also affect our interpretations of a video. Take, for example, “selective attention”: If asked to focus on a specific aspect of a video, viewers may miss other important details, Ristovska says. People can also be prepared to see what others want. In 2024 research by Feigenson and colleagues showed that a lawyer’s description of a video could color jurors’ perception of the actual footage.
A person’s beliefs can also shape their visual perception. People who identify with law enforcement, for example, are more likely to perceive police officers as acting lawfully in video evidence than people who don’t identify with law enforcement, Ristovska says. A person’s opinions on other potentially controversial topics such as abortionthe army or the death penalty can also affect how they view video evidence.
In 2009, when researchers surveyed 1,350 Americans about the video released at the Scott v. Harris, most agree with the majority view of the Supreme Court. But researchers identified “stark differences in opinion” along cultural and ideological lines, including race, income and a person’s views on the societal hierarchy. In Ristovska, the research shows that “seeing is believing” ultimately depends on who sees.
Ideology might help explain why different people might view video evidence of the recent killing of Renée Good, a Minnesota woman, by an Immigration and Customs Enforcement (ICE) agent in such different ways, says Loftus, who has studied human memory and the law for decades. “People [have] “There are pre-existing biases that either ICE is good or ICE is bad,” she says, “and that can affect how they perceive the behavior they observe.”
Should we change the way we watch videos?
To help resolve problems that can arise when different people interpret video evidence differently, Ristovska says viewers should slow down and “interact more thoughtfully with this material.”
And Feigenson recommends that viewers recognize that “other reasonable people may reasonably see things differently,” adding that “this can help temper the overreliance on video evidence that naive realism tends to engender.”
Adding artificially generated videos into the mix only complicates things. In 2025, Loftus, in collaboration with colleagues at the Massachusetts Institute of Technology Media Lab, published an article who demonstrated how artificial intelligence can change people’s memory of an image.
Participants were shown different images, including a photograph of a man and a woman who were not smiling. Participants then saw the images again, except this time the images had been slightly doctored using AI. In the case of the one representing a man and a woman, the researchers modified the image to add smiles. When shown the original image with the woman’s face obscured, people then incorrectly remembered her smile.
The idea of false memories introduced by AI is “concerning,” says Pat Pataranutaporn, an assistant professor at the MIT Media Lab and co-author of the study. But he hopes these findings could also have positive implications. If people have traumatic memories, for example, “AI could help them remember in a more positive way,” he says.
