Philando Castile was shot by a police officer during a traffic stop. His girlfriend, Lavish Reynolds, broadcast the aftermath of the shooting live on Facebook. The video leaves a trail of questions about the sequence of events, and the actions that led to Castile’s death. It also reinforces a broader question: How can police better train and equip officers to serve and protect the public?

In search of some answers, I turned to some people who have deep expertise in these matters. Two of them are former police chiefs, and another is a leading authority on police training in the United States. Ronal Serpas, a former New Orleans police chief, is a professor at Loyola University New Orleans and the chair of Law Enforcement Leaders to Reduce Crime and Incarceration, which advocates for the end to widespread incarceration. Tracey Meares, a professor at Yale Law School and a member of the President’s Task Force on 21st Century Policing, has trained thousands of police officers around the country, including in Chicago and Oakland, via the Trust & Justice initiative. Donald Grady II, a retired police chief with over 30 years in uniform, served in multiple cities. (I also reached out to the Justice Department. A spokesperson declined to comment on police training in general, or on the Castile case specifically.)

I asked them about some of the most promising or widely discussed avenues for reform. In the days following officer-involved shootings, talk often turns to training, not as a way to assuage the pain caused by incidents for officers and civilians, but as a way to systematically address shortfalls in conduct, create procedures, and soothe some of the fears that persist. But is training really the best way to make a meaningful intervention? Meares suggested that the fatal shooting of five officers in Dallas illustrates its limitations; sometimes, conflicts can’t be de-escalated. “There’s a set of training that is designed to help officers deal with situations that become escalated because of emotions and because the person they’re dealing with might have mental challenges,” she said.

A lot of situations in which there’s a shooting, it’s time-pressure, there’s lots going on. We know how to train people to slow things down so that incidents don't happen. Now that was not this. It’s hard to see how something like de-escalation training, which is making a big difference, it's hard to see how it would have made a difference.

But Dallas was an exceptional incident. There are millions of encounters between police and civilians every year, and that, Serpas said, is where training can make an enormous difference. What’s happening in American policing is, as we embrace more and more of these concepts of justice and legitimacy and the delivery of services, we’re actually building good will along the way,” he said.

That’s what you really want to see, because when the public recognizes that the police is disconnected, and not interested, and not responsive to their issues; then when a crisis like this happens, there’s no reason for the public to expect the police is going to do anything different than they’ve always done. I think training should always approach these events from what can we learn, and how can we make sure that we don’t do this again? Then we recognize that in the latest data available, there were 63 million contacts between a 16-year-old or older person at least once with a police officer in the country in a single year. Forty-two percent of those, which is the biggest piece, were traffic stops. That’s where we are going to most likely talk to people, and that’s where we need to be the best prepared to practice these new skills and these new ideas of bias and how to deal with it and how to prepare yourself. We have a lot of room to grow and a lot of room to be better; and training is key to it.

But for Grady, another police chief, the focus on training serves to distract from broader problems. This is not a training issue,” he said.

The problem is we’ve been addressing the issues wrongly for years. We keep wanting to say it’s a training issue. It’s not a training issue. That’s just a convenient thing to say, which causes everyone to be disarmed, and we no longer continue with the issue. In 36 years of policing, I cannot suggest to you a single training course that I could give someone that would change their thinking when it came to making a decision to shoot or not shoot when there is absolutely no threat to their person. This is not a training issue. This is an issue of who it is that we’ve decided we would allow to police our country. This dates back to the beginning of policing, not to some recent phenomenon. Policing was never designed to take care of the people that it is being forced upon, generally speaking, the most vigorously ... I do not subscribe to the same theories as everyone else does when it comes to this issue of policing.

Recruitment is one mechanism by which large police departments with troubled histories have tried to implement system-wide changes. New York City has been very proud of its initiative to recruit college graduates into its police ranks, for example. And college career fairs often include a recruiting table from the local police department. Not unrelatedly, community colleges are responsible for educating the bulk of civil servants in police departments around the country. Is recruitment, then, a better place to focus reform efforts?

Grady is more optimistic about recruiting than training. “Who you hire to do the job makes a difference,” he said.

What are their predilections? We use the same criteria for hiring police all across this country. When I got to my last police department, the relationship between the police and the community was terrible. The police hated the public, and the public hated the police … I decided I needed to find the right people, so we changed the criteria for hiring instead of doing the things that the police typically do to find what they consider to be the best qualified candidates. There’s a certain amount of aggression that they look for in a person. If you’re too docile, I’ve known people to be rejected from police departments because they weren’t aggressive enough … Why are we hiring people to do policing because of their level of aggression?

I can teach you to be appropriately assertive. What I can’t do is pull unreasonably aggressive tendencies out of a person. We took a look at who was being used as a psychologist and the criteria they were using for bringing people in. Then we began to look for psychologists who understood that perhaps that’s not the best profile for a police officer.

We found a psychologist that was willing to look for a more cerebral, more sensitive, empathetic, rational person than who was being hired as police officers. We began to hire people based on that new criteria. We spent a great deal of time talking directly with the psychologist, working directly with him to hire different people … We’re screening in people that are too aggressive and less cerebral. We’re screening in people that don’t want to take the initiative to go to college to get the education so that they have some socialization beyond an adolescent socialization, but they actually have a socialization with people where they’re learning something.

We screen those people out, and we hire people that have the high school education or the GED because those people are more malleable, and we can get them to do exactly what we want them to do. Eighteen-year-olds typically use less reason than people who have gone to school and gotten a four-year degree so that they can go forward. But those are the people we’re encouraging to come in. Then we claim that it’s a training issue.

And on that point, Meares largely concurred. “I do think that being able to select different people could help,” she said.

We know a few things demographically, forget about attitudinally, which is what [Grady] is talking about. I think it’s a problem that we have so many men as police officers. I think it would be different if we had more women. It’s an issue that police officers can start this job when they’re very young. We know most evidence points pretty strongly that having more-highly educated and older people starting the job makes a difference, but then of course that impacts our ability to hire a more diverse workforce.

There is, the law professor Seth Stoughton has argued, a distinction between police who adopt the mindset of a guardian, and those who approach their job as warriors. In general terms, the former see their role as that of peacekeepers and protectors, while the latter see themselves more as enforcers and wielders of authority. Serpas, like Stoughton, urges police to see themselves as guardians. “I don’t think there is any question that if police are responding to what the community wants in a way that the community is going to support, then the officers have to be true to that by delivering the service in a way that the community can support,” he said. “Which means you don’t go in as an occupying force, as a warrior. You go in as a guardian. I think that when police officers enforce the law, particularly in neighborhoods that have a broken-windows component, or a quality-of-life component, they’ve got to really be in lockstep with the demands of the community.”

Grady argued that this sort of approach shouldn’t be seen as a new development, but as a return to the ethos of American policing. “We’ve gotten distracted over the years,” he said, “and if you talk to most police officers today, and ask them, ‘What do you do for a living?’ they will tell you, ‘I am a law-enforcement officer.’”

That’s a mischaracterization of what they were hired for. In the state of Illinois, the statutes don’t call them law-enforcement officers, they call them peace officers, but they don’t refer to themselves as peace officers. They refer to themselves as law-enforcement officers. They have given themselves a mission of law enforcement, but law enforcement was never the mission. Law enforcement is only one small part of what we should be doing in policing. Policing is community building. That means, I need to take every effort, every tool at my disposal to build a community in a way that allows it to be most responsive to the people that we have sworn to serve and protect. It cannot be law enforcement. If you take that one small part and make that your mission, what happens to the activity of policing overall?

The theory of broken-windows policing is not as acceptable to Meares, who believes there are procedures that police are trained to follow that may be contributing to escalating tensions in some areas. One of those factors is the way police officers are trained to police crime. “The police officer stopped [Philandro Castile] for a broken tail light. That suggests to me that there’s some version of what’s called ‘broken windows policing’ happening in that city,” she said.

One thing to say about that is that there isn’t actually a lot of evidence, or strong evidence, that pulling people over for broken tail lights actually translates to substantial reductions in crime. There is evidence strongly indicating that policing in that way creates distrust between members of the community who are often disproportionately stopped this way and the police, in a way that’s actually inimical to the goals of crime control. Our best evidence suggests that legitimacy, which is based on trust, leads to compliance.

Secondly, there is training to explain to police officers how policing in ways that are inconsistent with trust-building can make their job harder, can escalate situations that don’t need to be escalated, can undermine trust—all of these things which are inconsistent with commitment to compliance, cooperation, and engagement … I think the key is to say that when you stop someone for a broken tail light, someone could consider that in scare quotes, “broken-windows policing.”

Serpas said he thinks the research on broken-windows policing is mixed at best. “In one way, it makes perfect sense to people that if you see disorder, perhaps it’s going to be likely that more disorder will follow,” he said.

But as it relates to policing, this is where I think community policing and justice and legitimacy help us navigate those waters. I was a police chief for 13 years in three different states, and when you go to a community meeting, you are primarily responding to quality-of-life issues that the community is bringing to your attention. When you have a serious crime wave of some sort, you hear about those issues; but, overwhelmingly, people want the quality-of-life-things dealt with: vagrants, drunks, abandoned buildings, abandoned cars, overgrown lots, and those kinds of issues.

Communities may want police to deal with similar issues, but certain racial and ethnic groups often complain about being over-policed and yet under-protected. I asked Serpas if police departments should do more to integrate implicit bias training or other more racially aware training into their work, which the Justice Department recently adopted.Absolutely. That’s one of the things that is very important. I don’t think anyone can reasonably argue that people do not have inherent biases,” he said.

They do, and recognizing that makes it easier to talk about what that means. What does it mean as a police officer? We need to train officers to be aware of those inherent biases from the perspective of learning about their behavior and their perceptions—how they can bring those perceptions and behavior to reality in a way that’s non-biased. If you can recognize that the biases exist, then it becomes a positive way to deal with them.

As implicit bias training spreads across police departments, body cameras have been widely adopted to provide factual versions of police interactions with civilians from the officer’s vantage point. Serpas, like many other reformers, sees them as a significant advance. I am incredibly, and have been, in favor of body cameras, since 2012, when I was a chief in New Orleans and called for every officer to have one in the field,” he said.

Body cameras do give us everything before, during, and after from that camera’s point of view; but when you have multiple officers with cameras, you suddenly start seeing this panoramic view. You get to see so many more things, and that’s how you can better diagnose what might be a training deficiency, or a disciplinary deficiency, or [conclude] there was nothing that the officers could have done differently. Body cameras bring us there. There’s some research out of the University of South Florida that demonstrates that when officers were questioned about body cameras on several dynamics, the one that really was very powerful is that the officers themselves saw the use of body camera video as an excellent way for them to train to be better.

Meares is not nearly as enthusiastic about body cameras. “Do I think it’s going to make a difference in the particular incident?” she asked rhetorically. “No.”

In terms of is it going to deter it. I think body cameras can make a big difference in terms of training … We can and should be able to leverage actual footage of what cops do to be really effective training mechanisms. In the same way that doctors go over a surgery and talk about, “Oh, we could have done this,” kind of a sentinel review, you can imagine having a tape every day or every week during roll call, showing cops the tape, and saying, “Okay. What could have been differently? What could have been improved?” Literally analyzing what they do. It’s hard to do that without having footage of what’s happening right then and there. I think that has great potential.