As U.S. law enforcement departments are accused of racist policing, one of the most common responses by the people in charge has been to have officers take “implicit bias” training.

The training usually consists of a seminar in the psychological theory that unconscious stereotypes can lead people to make dangerous snap judgments. For instance, unconscious associations of African Americans with crime might make cops quicker to see them as suspects.

After the 2014 Ferguson, Mo., protests, states rushed to require the training. Now a majority do, with New Jersey joining the list late last month.

But despite the boom in implicit bias training, there has been little real-life research into whether it actually changes what police officers do on the job.

“It’s like I’m offering you a pill to fix some disease, and I haven’t tested to see whether it actually works,” says Joshua Correll, an associate professor of psychology at the University of Colorado Boulder, where he studies racial bias. “Expecting that we can take people in and train them to reduce their implicit bias — I don’t think it’s been supported by the literature.”

That’s why Correll is excited about a new study at the New York Police Department that allowed researchers to track the effects of mandatory implicit bias training as it was implemented in 2018.