How Neuroscience Explains Why People See Different Things in the Same Video

Why can people watch the same video footage and see different things? Neuroscience can help explain

In February 2007 the Supreme Court heard a case that hinged on about 15 minutes of video evidence from two police vehicles' dashboard cameras: footage showed the front end of a police car as it pursued a driver in Georgia before it rammed into the back of the driver’s car, which subsequently crashed. The driver, 19-year-old Victor Harris, was left permanently paralyzed by the incident.

Harris sued the officer who hit his car. He alleged that the officer, Timothy Scott, had used excessive force. Before the Supreme Court considered the case, lower court judges had already reviewed the video footage and found in Harris’s favor, with one writing that Harris had presented little threat to the public despite his speeding.

But the Supreme Court disagreed, issuing an 8–1 decision in favor of the police officer. In the majority opinion, the justices determined that Harris posed an “actual and imminent threat” to the public and wrote that “we are happy to allow the videotape to speak for itself.”

The courts had reviewed the same video. But they came to dramatically different conclusions about what it showed. Indeed, the Supreme Court’s nine justices were not all aligned on what they saw in the footage; in a dissent, the late justice John Paul Stevens wrote that the video “confirms, rather than contradicts” the lower courts’ decision in favor of Harris. The episode raises the question: How can different people watch the same video yet see such vastly different things?

“Seeing is not just what our eyes physically see,” says Sandra Ristovska, an associate professor of media studies at the University of Colorado Boulder, “but also the experiences and ideas that viewers bring to images.”

In the years since Scott v. Harris, the case has become a quintessential example of this phenomenon and has been studied by legal scholars and psychologists alike. “Video is everywhere, from our phones to surveillance cameras on city streets. And it has become a vital form of evidence in court,” Ristovska says. And understanding how it can be interpreted differently depending on who sees it is critical to ensuring justice is fairly applied, she says.

One confounding factor in how someone views a video is something psychologists have dubbed “slow-motion bias.” In one 2016 study, researchers showed that when viewers watched surveillance footage of a shooting in slow motion, they perceived the shooter as “more intentional.”

Similarly, if a video is shaky, viewers can interpret events as more intense.

Then there is the “camera perspective bias,” says Neal Feigenson, a law professor at Quinnipiac University. In a series of studies at Ohio University, participants viewed 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 centered on the suspects’ face 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 instance, and then talk about it, you might unwittingly adopt some of your friend’s memories as your own. This phenomenon is known as memory contamination, says Miko Wilford, an associate professor at the department of psychology at Iowa State University.

A similar effect could happen if eyewitnesses are asked to recollect an incident that they also saw on video, she says.

“We’re just very bad at recalling the origin of information in our memory,” Wilford says.

When someone retrieves a memory, they “aren’t playing a recording back,” explains Elizabeth Loftus, a psychology professor at the University of California, Irvine. Rather “we are constructing” that memory, she says. In other words, the brain collects bits and pieces of information, sometimes from different times and places, and forges them into a memory. “Once that happens, it’s not easy to separate out what piece came from where,” Loftus adds.

In 2016 Loftus and her colleagues published a paper arguing that police officers should write down their account of an incident before viewing body camera footage: if officers watch such a video first, that might strengthen their memories of the details shown in it—but weaken their ability to recall other information that was not captured in the recording.

Humans are especially attuned to visual information. More of the processing power of the brain’s prefrontal cortex is dedicated to visual information than audio information, Ristovska notes.

That helps explain why people typically trust video evidence—even when they know it is false. In a notable 2008 study on this effect, researchers asked students to perform a gambling task on a computer. When the students were falsely accused of cheating and shown a fake video of the alleged infraction, the “vast majority” of the students confessed “without resistance,” the researchers found.

“People intuitively tend to believe that video gives them the objective reality of what it depicts,” Feigenson says. “This is naive realism.”

Cognitive biases can also affect our interpretations of a video. Take, for example, “selective attention”: if prompted to focus on a specific aspect of a video, viewers might miss other important details, Ristovska says. People can also be primed to see what others want. In 2024 research by Feigenson and his colleagues showed that an attorney’s description of a video could color jurors’ perceptions of the actual footage.

A person’s beliefs can also shape their visual perception. People who identify with law enforcement, for instance, are more likely to perceive police officers as acting lawfully in video evidence than people who don’t identify with law enforcement are, Ristovska says. A person’s opinions on other potentially divisive topics such as abortion, the military or the death penalty can also affect how they view video evidence.

In 2009, when researchers polled 1,350 Americans about the video at the center of Scott v. Harris, most agreed with the Supreme Court’s majority view. But the researchers identified “sharp differences of opinion” along cultural and ideological lines, including race, income and a person’s views on societal hierarchy. To Ristovska, the research shows that “seeing is believing” ultimately depends on who is doing the seeing.

Ideology could help explain why different people might view the video evidence of the recent killing of Renée Good, a woman in Minnesota, by an Immigrations and Customs Enforcement (ICE) officer in such distinct ways, says Loftus, who has studied human memory and the law for decades. “People [have] preexisting biases that either ICE is good or ICE is bad,” she says, “and that may affect how they perceive the behavior that they’re looking at.”

To help address the problems that can arise when different people interpret video evidence differently, Ristovska says viewers should slow down and “engage with this material more thoughtfully.”

And Feigenson recommends viewers recognize that “other reasonable people may reasonably see things differently,” adding that “this can help temper the overconfidence in video evidence that naive realism tends to engender.”

Adding artificially generated videos into the mix only complicates things. In 2025 Loftus, in collaboration with her colleagues at the Massachusetts Institute of Technology Media Lab, published a paper that 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 weren’t smiling. The participants were then shown the images again, except this time the pictures had been slightly doctored using AI. In the case of the one depicting a man and a woman, the researchers tweaked the image to paste grins on them. When shown the original image with the woman’s face obscured, people subsequently falsely remembered her smiling.

The idea of AI-introduced false memories is “concerning,” says Pat Pataranutaporn, an assistant professor at the MIT Media Lab and a co-author on the study. But he hopes the findings could have positive implications, too. If people have traumatic memories, for example, “AI could help them misremember in a more positive way,” he says.