A man was attempting to murder a toddler in San Diego, and Norm Stamper shot and killed him. The year was 1972 and Stamper, a police lieutenant in San Diego at the time, recalls that his heart pounded, his breath quickened and his vision narrowed into a tunnel. “I couldn’t have told you what was going on four-feet away, to the left or to the right,” he says. He pulled the trigger, the man fell and an official inquiry found that Stamper’s actions were justified.

Stamper went on to become chief of the Seattle Police Department in the 1990s and get a doctorate in human behavior research. He says stress helped him focus on the would-be murderer. But stress also leads to deadly mistakes, he explains in his new book, To Protect and Serve: How to Fix America’s Police. Mistakes have dominated the news with police shootings of Terence Crutcher in Tulsa, Oklahoma, Michael Brown in Ferguson, Mo., Walter Scott in North Charleston, S.C., Akai Gurley in Brooklyn, N.Y., and others. Brain scientists agree, and all say the tense situation is made even worse by the recent assassinations of police.

The trouble, research shows, is that the brain switches to a particular type of judgment system during stressful situations, relying on neural areas sculpted by evolution to make quick decisions. The fast responses, however, often ignore cues that could lead to a safer outcome. Police academies could take steps that give officers tools to avoid panic under fire, say several law enforcement officials and scientists. Yet such steps are rarely taken because training programs lack money and governmental overseers lack political will, critics charge.

“Put plainly, when cops mess up, the explanations offered tend to be ethical and political, when the more empirically solid explanations are much simpler than that—they are basic failures of human performance under stress,” says Jonathan Wender, a sociologist at the University of Washington and former police officer and sergeant. “We need evidence-based, human performance training that starts in the academy and continues across every career phase, so when you’re tired, scared or stressed, you still do the right thing.”

Stress has played an important role in our species’ evolution—it kept us alive. When faced with what we perceive as a life-threatening situation, we often go into autopilot, with all energy directed at escaping the threat—whether it be a tiger emerging from the forest or a man wielding a gun. Imaging scans show the brain shifts its activity (measured by blood flow and oxygenation, indicating which neurons are heavily used at a specific time) from the prefrontal executive control regions to subcortical reactive emotion areas. The prefrontal regions are responsible for analytic reasoning, impulse and desire inhibition, and cognitive flexibility based on information from many other brain regions. The reactive emotional areas that include the amygdala, hypothalamus and striatum are the realm of intuition, gut feelings and reflexes.

When relying on those latter regions, we respond quickly and automatically, without taking time to fully evaluate the situation. For example, something that appears to be a long, green, snakelike creature may trigger a quick motor response to run away when in fact the stimulus was just a bit of green cloth blowing in the wind. “That is fast processing based on beliefs, gut knowledge and not analyzing and looking for additional cues and stimuli,” says Rajita Sinha, director of the Yale University School of Medicine’s Stress Center.

These general findings about the brain under stress seem to apply particularly well to cops on the beat. In experiments performed with actual officers during a video simulation of a confrontation, a team from Vrije University Amsterdam found that shooting accuracy, arrest and self-defense skills and communication all decreased when stress levels are high, and that officers fired more often in a high-anxiety situation on suspects that had already surrendered.

Stamper says many cops operate in a continual state of hypervigilance, as though they are in imminent danger—a feeling that has only intensified in the wake of gun attacks on officers in Dallas, Baton Rouge, La., and New York City. Though vigilance is essential for safe police work, for a hypervigilant officer an action as innocuous as reaching for a wallet may activate the brain’s subcortical reactive emotion areas, which leap to the conclusion that the person is going for a weapon. A cop who is vigilant but calm may correctly recognize that the movement is harmless. “A stimulus that might have multiple interpretations for one person could very quickly push another person who is sensitized to threatening scenarios over the edge,” Sinha says.

Some of these tendencies may be innate, not learned. Sinha and her colleagues used functional magnetic resonance imaging to study brain activity in people exposed to stimuli ranging from highly stressful—images of mutilated bodies or someone pointing a gun—to neutral, such as a chair, table or lamp. They found that those with more flexibility—that is, shifting blood flow—in the brain’s ventromedial prefrontal cortex seem to have more emotional and behavioral control. Participants with such flexibility also self-reported they were less likely to argue and fight or to binge drink or be emotional eaters—all maladaptive ways of coping with stress.

There seems to be a way to counteract this problem. Remember those police officers in the Netherlands who, in simulations, fired at suspects who already surrendered? A second study showed that training in realistic environments, including practiced encounters with armed-opponent actors in buildings and streets, improved their shooting accuracy under stress Similarly, trials with 66 of the officers revealed that those who trained weekly on their own in a combat sport such as karate or kickboxing performed better under high-anxiety fights than those with no additional training, although both groups suffered performance decreases when conditions shifted from low to high anxiety.

This kind of training clearly helps, says human movement scientist Peter Renden from Vrije University. But to make it as relevant as possible to policing it should also include anxiety-inducing measures and tactical decision-making challenges. “We need to train for the whole situation, from beginning to end, including representative stress levels,” he says. In a follow-up study, he and his colleagues found that giving 11 officers additional self-defense and arrest training, and then subjecting them to high-pressure, realistic scenarios such as encounters with shoplifters or drunk drivers resulted in improved communication, alertness, assertiveness and resolution. Renden hopes to use these findings to develop an effective training program across the Netherlands’ police force.

Sinha also plans to explore behavioral training, such as mindfulness exercises that help people cope with stress as well as drugs like guanfacine, which reduces stress hormones in the prefrontal cortex. “This region shows a lot of dynamic activity, so there may be ways of intervening that really help,” she says.

Changing police training in the U.S. is not easy, however. There are nearly 18,000 local, state and national law enforcement academies, each of which has its own training requirements and standards. Although many departments say they do stress training, oftentimes it is simply a PowerPoint presentation, Wender says. In July the U.S. Department of Justice reported on a survey of entry-level officer training from 2011 to 2013. Of the 664 academies that responded, 48 percent did say their programs included a stress-based approach. But further specifics about those methods were not reported nor were any measures of their actual effectiveness. A spokesperson declined an interview request, citing a lack of data. Michael Becar, executive director of the International Association of Directors of Law Enforcement Standards and Training, says the only way to learn the details of stress-related training is to contact each training facility individually.

There are examples of training that emphasize practical, realistic skills that prepare officers to remain calm and in control when faced with a threat. Wender’s company, Polis Solutions, has trained hundreds of police officers and their supervisors using a measured stress approach that gradually amps up of pressure and complexity. A cop may first be asked to fire on a target—a simple drill—but then asked to fire while standing on one foot and being heckled by an actor. Eventually, the training builds up to mock scenarios involving multiple players and stressors. Wender’s earlier work with the Department of Defense found that intense rehearsal of stressful face-to-face interactions improves outcomes, and police officers who have undertaken his training report an increase in their ability to get hostile people to cooperate without using coercion or force. The results include a 17 percent improvement from a New York City metro area group, and a 10 percent improvement in a group of senior trainers and supervisors from North Carolina.

“The [Polis Solutions] training includes all the elements that will help officers reduce stress or at least recognize it and manage it,” says Geoffrey Alpert, a criminologist at the University of South Carolina with no ties to the company. “While a thorough evaluation the training has not been completed, anecdotal comments and other information suggest that officers with it will perform better than those who have not been trained in it.”

Other solutions are available, including VirTra V-300, a 300-degree immersive simulator—a sort of IMAX meets video game—that offers police-specific simulations of shootings, ambushes, traffic stops, disturbances and more. The company has approximately 200 U.S. law enforcement customers, although some agencies complain that they cannot afford the simulator’s $150,000 to $300,000 price tag.

The high cost of good training has a perverse effect, Wender says. It means police departments who least need the help—those in affluent low-crime areas—are most able to afford it. Departments that are struggling and are most in need of such training, on the other hand, cannot get it. This adds another layer of challenge to the problem of how to actually implement and sustain effective training.

Stamper would like to see the federal government put out requests for proposals and provide grants to develop best practices curriculum at the national level, although given the way our police forces are organized, he says, local agencies would still have to pick up the tab for actual training, which means fund appropriations from city or state politicians and taxpayers. “There’s an intellectual awareness of this,” Wender says. “But what’s really lacking is the political will to put the funding and resources behind dealing with this issue.”