[Originally posted in November 2006]
The recent controversial shooting of an unarmed black man in New York has generated terrible grief and perhaps justifiable anger. But if officers honestly believed the man was armed and intended to harm them, weren’t they justified in shooting?
Perhaps, but an important additional question is this: were they predisposed to believe he was armed simply because he was black? Consider this quick movie:
Click to play (QuickTime required)
It will flash two pictures. One man is armed, the other unarmed. Who do you shoot? I’ve primed you to think about race, so it’s not really a fair test. If you were a police officer who believed his life to be in danger, would you respond in the same way? (You can use the slider to see which man really was armed.)
In 2003, a team led by J. Correll flashed random photos of white and black faces, some superimposed with guns, others with harmless items such as cell phones and wallets. They asked college students to press one key indicating “shoot” the suspect, and another indicating “don’t shoot.” The students were more likely to mistakenly fire at black faces that were unarmed compared to unarmed white faces.
But what about police officers? With their special training and rules about when to fire, perhaps they will do better. A study by E. Ashby Plant and B. Michelle Peruche tested police officers on a similar task.
They created composite images by superimposing guns and harmless items on the faces of 9 white men and 9 black men. The faces were previously matched for attractiveness. These faces were randomly flashed on a computer screen; each face was shown both with a gun and with a harmless object.
48 mostly white police officers who volunteered to participate in the project were told only that the study was about “decisions to shoot.” They had 630 milliseconds after each face was displayed to press the “shoot” or “don’t shoot” button on the computer running the test. As the test began, results were the same as with college students: Police officers were more likely to mistakenly shoot black suspects holding harmless items than white suspects holding the same items. They made an average of 3.63 errors over 20 trials when the suspect was black, but only 2.65 errors when the suspect was white.
But an interesting result occurred as the test continued. After another 80 trials, about 40 of which included the key situation of a person holding a harmless object, the disparity between reactions to white and black suspects disappeared. During the second half of the experiment, the average number of errors for black suspects diminished to 2.60, statistically indistinguishable from the rate for white suspects.
Plant and Peruche believe this is a hopeful result, suggesting that police officers can be trained to eliminate bias in their work. They may be right, but I do have one potential objection: police are also “trained” by what they actually see every day as the patrol the streets. Won’t this training — for better or for worse — trump whatever they learn in a few hours in the training room?
The problem in the Queens shooting of Sean Bell was, like most police incidents, more complex than the scenario presented in this study. An informant may have told the police Bell was armed, and he had attempted to ram the police with his car more than once. Also, the question was less about whether shooting at all was justified, and more about whether so many shots should have been fired. Police may have assumed other officers were justified in firing, and so that’s why they fired themselves.
Malcolm Gladwell, in his book Blink, offers another, counterintuitive solution to the problem of accidental police shootings: Police should most often work alone. With the variable of other officers removed from the problem, individuals are more likely to think for themselves. Without the option of “backup,” they are less likely to get themselves into dangerous situations in the first place.
Ashby Plant, E., & Michelle Peruche, B. (2005). The Consequences of Race for Police Officers’ Responses to Criminal Suspects Psychological Science, 16 (3), 180-183 DOI: 10.1111/j.0956-7976.2005.00800.x