The shooting death by police of Gilbert Flores, a Latino man who bystanders claim had raised his hands in surrender, has San Antonio authorites rushing to equip their police with body cameras. Fortunately, this possible murder was captured by at least two observers with cell phones. As of this writing, it’s not clear which version of the incident the videos will support. This is only one of countless recent police-civilian encouters in which videotapes either made a difference or would have been helpful.

By a body camera, I’m referring to a small camera that clips onto an officer’s uniform or eyeglasses and records audio and video of the officer’s interactions with the public. Although I am generally opposed to warrantless surveillance, in this case the advantages outweigh the disadvantages. The American Civil Liberties Union (ACLU) refers to body cameras as a “win-win” because, if properly used, they can protect the public against abuse of power by the police while protecting the police from false accusations of brutality. Of course, there must be policies in place to prevent the police from selectively recording only certain interactions or parts of interactions, or from editing tapes after the fact. The ACLU recommends a set of policies to protect the public from this and other abuses of the technology.

However, there is a drawback of the use of this technology that so far seems to have escaped the notice of the mass media. A body camera records an interaction from a particular point of view—that of the police officer. There is evidence from social psychology that visual perspective can alter the salience of people and their actions, and this can affect the conclusions that people draw.

In a 1975 experiment, Shelley Taylor and Susan Fiske staged a conversation between two people, and placed observers at various locations around the room, as indicated in the diagram. Afterwards, the observers were asked to rate the amount of causal influence that each speaker exerted during the conversation. The results showed that the observers attributed greater causality to the person they were facing. Observers C and F saw Actor B as more influential, Observers D and A favored Actor A, and Observers B and E, who could see both actors equally well, tended to see them as equally important. This phenomenon is sometimes called illusory causation. People attribute greater causality to a person simply because he or she is more salient or noticeable than other people.

When an interaction is videotaped from different locations, the effect is called camera perspective bias. Daniel Lassiter and Audrey Irvine staged an interrogation in which a detective questioned a suspect, with the suspect eventually confessing to a crime. Three cameras simultaneously filmed the interaction, one looking over the detective’s shoulder, another over the suspect’s shoulder, and a third from the side with both the detective and the suspect equally visible. Observers were shown one of the three tapes and asked how voluntary the confession was. The confession was judged to be most voluntary—that is, caused by the suspect—when the camera was focused on the suspect and least voluntary—caused by the detective—when it was focused on the detective.

Lassiter and his colleagues have replicated this result several times, including under quite realistic conditions. In one study, they staged a mock trial and played jurors a videotaped confession filmed from one of the three perspectives. Not only was the confession seen as more voluntary when the focus was on the suspect, participants were more likely to find him guilty and recommend a longer sentence. Most police departments record confessions with the camera focused on the suspect.

These studies have implications for the police use of body cameras. The videotapes become important when there is an altercation between a police officer and a civilian suspect leading to some adverse outcome, such as the suspect being shot. Observers of the video must assign responsiblity under circumstances that may be quite ambiguous. When the camera is focued on the suspect, he or she will be more likely to be seen to have caused the bad outcome. Any aggressive behavior by the suspect is captured by the camera, while nonverbal behavior by the officer that is obnoxious or threatening can go unseen and become difficult to prove. The body camera is not a neutral observer of the interaction. It is biased in favor of the police officer.

A dashcam—a camera mounted on the dashboard of a patrol car—can  provide a more objective view of a police-civilian encounter, provided that both participants are visible. So too can a video taken by an observer with a cell phone. Pittsburgh Post-Gazette columnist Tony Norman recommends that everyone carry a cell phone and record every encounter they have with the police. He gives an example of a black motorist who used this tactic to embarrass a policeman who was harassing him. However, this approach is not without its risks, as police have been known to charge people who try to videotape them with a crime. If you wind up in jail or in a hospital, the fact that you had a legal right to record the interaction may provide little comfort.

I’m in favor of police body cameras. They’re a clear improvement over the status quo. However, camera perspective bias needs to be more widely publicized and better understood.