Jurisprudence

The Hidden Bias of Cameras

Body and dashboard-mounted cameras used by police don’t give an objective account of the facts. Here’s how we should reform them.

NYPD Body Camera
Body cameras can be just as biased as people. Here, New York Police Department Sergeant Joseph Freer demonstrates one during a press conference, December 3, 2014.

Photo by Andrew Burton/Getty Images

Excerpted from Unfair: The New Science of Criminal Injustice by Adam Benforado. Out Now from Crown.

Deputy Clinton Reynolds was sitting in his cruiser along Highway 34, southwest of Atlanta. It was a damp Thursday evening in late March. The cars would catch a glimpse of him and ease onto the brakes. He kept an eye on the radar gun. A Cadillac came up doing 73 mph in a 55, so he flicked on his blue lights. He was going to let it go—everyone speeds.

But the Cadillac didn’t slow as it passed. In fact, as soon as he turned onto the two-lane road, the driver accelerated.

Reynolds radioed in the license plate and reported that he was in pursuit. He didn’t request assistance, but another officer, Timothy Scott, was listening in and soon joined the chase.

The six-minute pursuit covered two counties, and finally, Scott had had enough. He radioed in for permission to “PIT” the vehicle. A Precision Intervention Technique maneuver involves hitting the back of the fleeing car in order to spin it to a stop. And though it’s known to be hazardous at high speeds, Scott got his answer, crackling across the radio: “Go ahead and take him out.” The cars had reached a narrow stretch of the road, with no shoulder, but Scott saw his chance and took it. He hit the gas and rammed the Cadillac’s rear bumper.

The result was dramatic: The Cadillac swerved and then swung to the right shoulder, careened down an embankment, and flipped over. Scott yelled into his radio, “It’s a 10-50 [police code for an accident]. It’s gonna be a bad 10-50. Bad.” White smoke poured from the wreckage. The officers ran down to the car and pulled desperately on the doors. Scott looked through the window and could tell that the driver “did not have his seat belt on. His head was beneath the brake pedal, his torso was across the seat, and his legs were bent over the back of the seat.”

Victor Harris, 19, beat the odds and survived. But the cost of the accident was severe: In the hospital, Victor learned that he was paralyzed from the neck down.

Taking stock of the events that led to his injury, Victor decided to sue Officer Scott under the Fourth Amendment of the Constitution, which prohibits unreasonable seizures. Victor’s argument was that just as the Constitution forbids a police officer from shooting a shoplifter in the back as he runs away, it also bars an officer from using a potentially lethal bumping technique to cause an individual’s vehicle to spin out when the initial offense was simply driving a few miles over the speed limit.

The government, for its part, contested the notion that Officer Scott’s actions were unreasonable, which raised some important factual questions. Perhaps chief among them were just how serious a risk the chase posed and who was to blame—Victor or the police—for the hazards that it created.

Had the accident taken place ten or twenty years earlier, those weighing Victor’s claim and determining what happened would have had to piece together a narrative from the testimony pro­vided by the parties, various witnesses, and experts. But in 2001, Coweta County, Georgia, was running dashboard cameras in its squad cars. To determine whether Victor created a significantly dangerous situation in fleeing from the police that justified Scott’s PIT maneuver, the chase did not have to be re-created; it could simply be watched.

The case eventually made its way to the Supreme Court. And it was clear from the oral argument that the justices were skeptical of Victor’s position. They had all seen the video, and it had left a powerful impression. As Justice Antonin Scalia explained, Victor “created the scariest chase I ever saw since The French Connection.”

The issue, though, was what a jury would think. Was it possible that a reasonable juror might watch the video and disagree with Justice Scalia? If so, the justices would have to send the case back for a full trial; if not, then Victor’s attempt to gain relief was over.

The Supreme Court majority found this an easy question to answer because the videotape provided such a clear picture of what actually happened. According to the court, after watching the tape, no reasonable juror could possibly believe that the chase was anything but extremely dangerous, and no reasonable juror could possibly imagine that the police were at fault for Victor’s paralysis. As Justice Scalia wrote in the opinion, Victor’s “version of events” was “utterly discredited” by the videotape. The court was so sure of itself that it took the unprecedented step of posting a link to the video online and inviting members of the public to watch.

We operate under the illusion that reality enters our brain through our senses unfiltered. But at any given moment, our race, gender, age, profession, politics, religion, and countless other identity-defining characteristics and affiliations are coloring what we see.

In a powerful demonstration of this phenomenon, a group of law professors decided to test the Supreme Court’s conclusion that “no reasonable juror” could watch the footage of the chase that left Victor Harris paralyzed and see Victor’s evasion of the police as anything but extremely dangerous and the cause of the eventual crash. The researchers asked a diverse group of 1,350 Americans to watch the video and then offer their impressions.

What they found were clear rifts in perception along ideological, cultural, and other lines concerning the key issues in the case. A less affluent, liberal, highly educated black woman with egalitarian and communitarian views was far more likely than a wealthy, conservative white man supportive of existing social hierarchies and individualism to see Officer Scott and the police as the primary culprits.

If different people with different backgrounds and identities can look at the same events and see very different facts, is it also possible that the same person can look at the same events and see very different facts depending on how information is presented? Over the last few decades, researchers have conducted a number of experiments showing that when we view events as if standing in the shoes of the person experiencing them, we are much more likely to attribute the actor’s behavior to forces and constraints in the surrounding environment than when we adopt the perspective of an outside observer, in which case we tend to make attributions that focus on the individual’s disposition and character.

Imagine that you are impaneled on a jury and have to decide whether the defendant’s confession was voluntary or coerced by the police. As luck would have it, the entire interrogation was recorded, and you are provided with a videotape from one of three cameras in the room: a camera directed at the interrogator, a camera directed at the defendant, or a camera positioned to the side, showing both parties. It would seem reasonable to assume that regardless of the footage you were shown, you would come to the same conclusion, since all three cameras captured the exact same scene. When scientists conducted a number of studies using such a setup, however, they found that perspective made a big difference. By simply shifting the point of view from the person being questioned to the interrogator, researchers were able to significantly reduce the number of people who thought the resulting confession was coerced. Watching the interrogator through the eyes of the suspect, it was a lot easier to see—and feel—the menace and pressure. Those who watched the videotape that showed both sides made assessments that fell in between the two conditions.

This bias seems to occur both for minor offenses like shoplifting and for more serious crimes like burglary, rape, and manslaughter. And it’s surprisingly sticky: Greater expertise (being a law enforcement officer or a judge), increased accountability, and judicial instructions aimed at encouraging people to be more mindful of perspective bias all appear to be largely ineffective.

Could camera-perspective bias have played a role in the outcome of Victor Harris’ case?

Justice Scalia thought he had rooted out any potential unfairness by noting that “there are no allegations or indications that this videotape was doctored or altered in any way,” but he and the rest of the majority were wrong.

The video camera mounted in Officer Scott’s cruiser seemed to offer an unfiltered lens on the core facts of the case. But it didn’t just record a high-speed chase; it recorded a chase from the perspective of the lead police officer involved in the pursuit. Having adopted the officer’s physical perspective, it becomes extremely easy to share his assessment of the situation and understand why he acted the way he did.

What if the Supreme Court had watched another, different videotape—say, a clip taken by a pedestrian standing at an intersection or footage from a news helicopter hovering above? What if there had been a recording taken from inside Harris’ car, looking back at the police cars hot in pursuit? Each of these tapes would be a true, accurate depiction of the police chase, but the research suggests that we would reach very different conclusions about the level of risk the chase created, whether deadly force was reasonable under the circumstances, and who was to blame for putting the public in danger.

Seeing events through Victor’s eyes would have put his actions in a whole new light. Viewers might have considered what it must feel like to be a 19-year-old black man in Georgia being pursued late at night by multiple police cars with lights flashing and sirens blaring. It might have triggered feelings of empathy and raised challenging questions: “How would I have responded under the circumstances, after making that initial mistake of not pulling over, knowing how officers tend to treat those who flee and seeing no witnesses to intervene if things got out of hand?” With such a video, it would have been easier to see Victor as a young man influenced by powerful situational pressures.

At the time of the accident, Victor was working at a temp agency to support himself as a full-time student at Griffin Technical College. He had left home at 4 in the morning and had been working all day. By 11 p.m. he was completely exhausted and was not paying attention as he passed the speed trap. When he suddenly saw the lights in his rearview mirror, he “panicked.” His license had been suspended for unpaid tickets, and he was afraid he’d go to jail. Running from the police is the last thing many Americans would ever do, but for young black men—for whom the threat and fear of harassment, capture, and incarceration is ever present—it can be a basic instinct, learned as a kid. There is no expectation that things will work out, that you’ll get a fair shake, that you can trust in the police and the system.

In a fateful instant, Victor pressed down on the gas. As he explained later, “It wasn’t my intention to put anyone’s life in danger or to scare anyone. … I couldn’t believe this was actually happening to me. … The last thing I wanted to do was hurt myself or anyone else. I was 19 years old, and I was scared.”

Every year, more squad cars are outfitted with dashboard cameras, more jurisdictions require interrogations to be taped, and more officers are wearing recording devices on their bodies as standard procedure. Prompted by a string of highly publicized incidents of police brutality, the trend has only increased in recent months. But amid the fanfare proclaiming the camera the savior of the public and bane of injustice, Victor’s case should give us pause. Most of these recordings will show events from the perspective of the police, and each one will hold the potential to sway those who watch it later, whether that is a prosecutor deciding whether to charge a suspect with resisting arrest, an internal-affairs detective reviewing the conduct of an officer, a juror trying to figure out who the aggressor was, or an appellate judge reviewing a case.

But the solution isn’t to eliminate cameras altogether. Used in the correct manner, cameras may offer ways to make our system fairer.

In 2012, a program was introduced in Rialto, California, to place small cameras on the bodies of police officers (in this case, on their sunglasses). Over the next year, both the number of complaints filed against officers and the number of incidents where officers used force decreased by well over 50 percent. Knowing that they were being watched seemed to encourage cops and members of the public to behave more civilly.

The proliferation of video footage may also help us to avoid certain biases (for example, by eliminating the need to rely on faulty eyewitness memory to identify a suspect). And once we understand how footage can influence perception, we can change how we use cameras to address that distortion. Since camera angles that offer third-party perspectives can prevent biased assessments of whether confession were coerced, those should be the first choice. Since it’s often not possible to employ dashboard cameras—or those attached to an officer’s glasses—in a neutral manner, such video footage shouldn’t be used in a conclusive fashion, as it was by the Supreme Court. We should be particularly careful when footage doesn’t show all parties in the frame or when it captures only some of the key events. Alternatively, we might allow it to be entered into evidence only with the consent of the defendant.

Victor Harris made a terrible error in judgment that rainy night, but he did not deserve to be paralyzed, and he did not deserve to have his case taken away from a jury. The system failed him because of the psychological limitations of the people who operate it—limitations we all share. We spot bias easily in others, while ignoring how our own backgrounds and experiences shape how we see the facts of a case. We blindly trust in the objectivity of our technologies because they seem to deliver reality, when, in truth, they may distort our perspectives.

Adapted from Unfair: The New Science of Criminal Injustice by Adam Benforado. Copyright © 2015 by Adam Benforado. Adapted by permission of Crown Publishers, an imprint of the Crown Publishing Group, a division of Penguin Random House, LLC. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.