Written by Amanda Ripley for time.com

But the research on actual gunfights, the kind that happen not in a politician’s head but in fluorescent-lit stairwells and strip-mall restaurants around America, reveals something surprising. Winning a gunfight without shooting innocent people typically requires realistic, expensive training and a special kind of person, a fact that has been strangely absent in all the back-and-forth about assault-weapon bans and the Second Amendment.

In the New York City police department, for example, officers involved in gunfights typically hit their intended targets only 18% of the time, according to a Rand study. When they fired 16 times at an armed man outside the Empire State Building last summer, they hit nine bystanders and left 10 bullet holes in the suspect—a better-than-average hit ratio. In most cases, officers involved in shootings experience a kaleidoscope of sensory distortions including tunnel vision and a loss of hearing. Afterward, they are sometimes surprised to learn that they have fired their weapons at all.

“Real gun battles are not Call of Duty,” says Ryan Millbern, who responded to an active-shooter incident and an armed bank robbery among other calls during his decade as a police officer in Colorado. Millbern, a member of the National Rifle Association, believes there is value in trained citizens’ carrying weapons for defensive purposes. He understands what the NRA’s Wayne LaPierre meant when he said, “The only thing that stops a bad guy with a gun is a good guy with a gun.” But he knows from experience that in a life-or-death encounter, a gun is only as good as its user’s training.

Under sudden attack, the brain does not work the way we think it will. Millbern has seen grown men freeze under threat, like statues dropped onto the set of a horror movie. He has struggled to perform simple functions at shooting scenes, like unlocking a switch on a submachine gun while directing people to safety. “I have heard arguments that an armed teacher could and would respond to an active shooter in the same way a cop would. That they would hear gunshots, run toward the sound and then engage the shooter,” Millbern writes in an e-mail from Baghdad, where he now works as a bomb-detection K-9 handler. “I think this is very unrealistic.”

As lawmakers in at least seven states debate whether to allow teachers to carry firearms in school (something already allowed in Utah and Texas), it is worth considering: What happens in the human brain during a gunfight? And how much training would armed teachers or security guards need to prevail?

The Adrenaline Surge

At 3 p.m. one autumn day in 2004, Jim Glennon found himself being shot at without warning. He was a lieutenant, a third-generation cop who had decided on the spur of the moment to help out on a routine shoplifting call. The suspect, a white man in his mid-50s, had walked out of a liquor store with a bottle of vodka without paying for it, and the police had tracked his license plate to a condo complex in a suburb of Chicago.

The officers knocked on the door at the end of a long hallway and got no response. After a few minutes, Glennon started to suggest they come back with a warrant. That was when the man threw open the door and began firing a black snub-nosed revolver from three feet away.

Glennon was a police-academy trainer, unusually well schooled in survival skills. But from the moment he saw the revolver, his mind entered a state unlike anything he’d experienced before. “Oh s—! Gun!” he said, spinning his body hard to the left, missing a bullet by inches or less.

Without his conscious knowledge, the sight of the gun had sent a signal to his brain stem, passing a message to his amygdala—the primal, almond-shaped mass of nuclei that controls the fear response from deep within the brain’s temporal lobe. The amygdala, in turn, triggered a slew of changes throughout Glennon’s body. His blood vessels constricted so that he would bleed less if he got wounded. His heart rate shot up. A surge of hormones charged through his system, injecting power to his major muscle groups should he need to fight or flee.

His first actual thought was that the gun must have had only five or six rounds. He knew this because it reminded him of the revolver his grandfather gave his father years earlier. As he and a fellow officer turned and began racing down the hallway to take cover around the corner, he counted the number of shots he heard behind him, waiting for the suspect to run out of ammunition. Relying on his training, he pulled his .40-caliber Sig Sauer pistol out of his holster.

As happens for most people in life-or-death situations, his brain began to manipulate his perception of time, slowing down the motion as he fled down the corridor. “The hallway looked like one of those dreams where it is just really, really long,” he says. Later he would guess that it was 250 ft. long; it was really 79 ft.

But for each superpower his brain gave him, it took one away. In a flash, his brain reprioritized, shifting finite resources to the cause of survival. As he ran, rounds bursting behind him “like cannon shots,” he suddenly fell flat on his face in the carpeted hallway, tearing skin off his hands and knees.

“I was a 48-year-old guy wearing 20 lb. of equipment,” he remembers, “and I was running faster than I think my body was capable of handling.” In life-or-death situations, human beings often lose basic motor skills that we take for granted under normal conditions. (Attackers, not just those they’re shooting at, also experience such trade-offs, though they usually have the advantage of not being taken by surprise.)

Instantly, Glennon bounced back up and kept running to the corner, which seemed to get no closer with each step. Just then, his fellow officer fell down in front of him, screaming that he’d been shot. So Glennon’s brain reprioritized again. He grabbed the officer’s belt and heaved him the rest of the way around the corner. He remembers feeling pain in his back and thinking, Son of a bitch got me. It had taken seconds to get to the end of the hallway, but it felt like minutes.

Then, having finally taken cover, he turned and pointed back down the hallway toward the shooter. It was a chilling sensation to see his bare hand in front of him, pointing in the shape of a pistol like a boy on the playground. Where was his gun? “I looked at my hand. It wasn’t there. I looked in my holster. It wasn’t there.”

Without being aware of it, Glennon had dropped his gun in the hallway when he’d reached over to help the wounded officer. In moments of extreme stress, the brain does not allow for contemplation; it does not process new information the way it normally does. The more advanced parts of the brain that handle decision making go off-line, unable to intervene until the immediate fear has diminished.

Luckily, Glennon did not dwell on this mistake. Nor did he freeze or shut down entirely, as many people do in life-or-death situations. Instead he reached over and grabbed the gun out of the holster of the injured officer. When he looked back down the hallway, he saw the arm of the shooter pointing toward him—and, behind it, the arm of a third police officer pointing out from another doorway.

More than anything else, Glennon wanted to shoot back. He started to squeeze the trigger. Then from somewhere in the recesses of his brain, he reminded himself: You can’t shoot. If he did, he would risk hitting the third officer standing behind the gunman. His training kicked in just in time, overriding his instincts.

The third officer took two shots at the gunman from an awkward angle, missing both times. But seconds later, the suspect threw his gun into the hallway, surrendering. The officers handcuffed him, and a battery of backup officers arrived. Glennon’s deputy chief ripped off Glennon’s bulletproof vest to make sure he hadn’t been shot too; he was fine. The pain in his back was the pain that came from one middle-aged man lifting another. Only later, in the ambulance, did Glennon begin to shake, just as he’d read people tend to do in the aftermath of an adrenaline surge.

Beyond Target Practice

Today, Glennon runs Calibre Press, a law-enforcement training company based outside Chicago, and has trained tens of thousands of police officers nationwide. His primary message to his trainees is that they need better training than they typically get; real gunfights are nothing like the ones on TV. “Over half the police officers in the country are only required to go down once or twice a year and shoot holes in a paper target,” he says. Experts who study human performance in gunfights generally agree that people can train to perform better through highly realistic, dynamic simulation training. But that is expensive, especially compared with traditional target practice, and it doesn’t happen often enough.

In the aftermath of the Newtown shootings, as local governments contemplate allowing more firearms in schools, Glennon worries that communities might inadvertently undertrain civilians just as they have done with police officers. “Cops aren’t trained well enough, so what do you think they’re going to do with teachers?” he says. “It’s not enough just to carry a gun.”

When I asked police safety experts how much training would be ideal for teachers or, for that matter, police officers assigned to schools, they offered different estimates. In Arizona, Alexis Artwohl, co-author of the book Deadly Force Encounters and a veteran police psychologist and trainer, recommended a weeklong program with “a lot of practice” and a requirement that participants meet minimum performance standards in order to graduate. In Ohio, Bill DeWeese, a veteran police officer and head of the National Ranger Training Institute, recommended two to three times that much training, and he pointed out that the best training includes much more than firing a gun. “I’m an avid firearms person and always have been,” he says. “The one thing I’ve learned is that it’s not about possessing firearms. It’s about possessing the skills to read a situation—learning how to adapt and maneuver, to respond to an unexpected, fluid situation.”

But in DeWeese’s state of Ohio, 1,100 teachers have already signed up for the Armed Teacher Training Program, offered free by Buckeye Firearms Foundation. That class will last just three days. In other states, civilians can get concealed-carry permits with one day of training or less. About a third of all public schools in the U.S. already have armed security, including every high school in Chicago, and that number may increase after the Newtown shootings. To date, there is no clear evidence that such measures make schools safer. Some studies have found a decrease in violence in schools with in-house police officers, while others have found no relationship at all. Still others have found that armed security makes some students feel less safe—and may funnel more students than necessary into the criminal-justice system for small infractions.

Of course, it’s also possible that the mere presence of armed teachers or guards could deter a shooter from attacking altogether. There would be no need to perform well in a gunfight—because there would be no gunfight. (Likewise, over the course of a career, it is statistically unlikely that a New York City police officer will ever fire his or her weapon in the line of duty, but the silent presence of officers’ weapons surely influences the behavior of civilians around them.) Many gun-rights advocates worry that gun-free school zones actually attract shooters because they represent easy, vulnerable targets. It’s hard to know, though, if mass murderers apply such logic when choosing targets—or if they simply seek to create the most socially abhorrent crime scenes in order to breed maximum shock and grief. In the case of the 1999 Columbine High School shooting in Colorado, for example, the attacking students were aware that their school had an armed sheriff’s deputy in the school parking lot. (The deputy exchanged fire with one of them but missed.)

Of the mass shootings that are stopped by others, roughly two-thirds are brought to an end by civilians, according to Ron Borsch, a police officer and trainer in Bedford, Ohio, who has been keeping a database of such incidents since the Columbine shooting. That’s because they are typically the only ones in the immediate vicinity of the shooter. And most of those civilians are unarmed, Borsch has found. In the shooting of Arizona Representative Gabrielle Giffords and 18 others, which happened in just 15 seconds, civilians tackled gunman Jared Loughner, ripped the gun from his hands and confiscated his ammunition.

By then, though, it’s already too late for the victims. Dan Marcou, a former SWAT commander and police officer who was involved in three shootings in Wisconsin, argues that the public’s most important opportunity comes before any shooting starts. Most shooters belong to the communities they target and go through predictable phases before they kill anyone, from fantasizing about the murders to planning them. “We have to pay attention,” he says. “It doesn’t have to be a police officer who fires a shot; sometimes it’s a teacher who comes forward and says, ‘Hey, this guy is really dangerous.'”

By fixating on hypothetical school-yard gunfights, we are choosing to fight in the riskiest arena: the chances that an officer or armed educator will shoot a child by accident are high, as are the chances of arriving officers’ mistakenly shooting anyone seen with a weapon in the ensuing chaos.

With all this uncertainty, it is useful to remember that the odds of a U.S. student’s being killed at school are about 1 in 3 million, lower than the odds of being struck by lightning. Schools are safer now than they have been in 20 years. Kids do become victims of gun violence far too often in the U.S.—but almost always outside school, far from gun-free zones or teachers with pistols.