The city of Sacramento, along with the rest of the nation, watched in stunned silence on March 30 as officials released details of the death of Stephon Clark. Clark, a 22-year-old unarmed black male and father of two, was shot several times and killed in his grandmother’s backyard by two Sacramento police officers.

The two officers, guided by a sheriff’s helicopter unit, searched the neighborhood for a man in a hoodie allegedly breaking car windows with a “toolbar,” and identified Clark near the side of his grandmother’s house. When they came upon Clark, the officers shouted, “Hey, show me your hands! Stop!” For reasons that will never be known, Clark ran around to the back side of the house. Seventeen seconds later, the officers turned the corner and fired 20 rounds, hitting Clark seven or eight times. As he lay dying on his grandmother’s back patio, officers waited for six minutes after the last shot was fired before checking on Clark. Upon searching the scene and the body, the officers found only a mobile phone—no firearm, no “toolbar.”

Americans have endured this heartbreak far too frequently. While each shooting has its own unique set of facts, it feels as if we are watching the same film of police violence on repeat—violence that seems to disproportionately target black communities: Alton Sterling in Baton Rouge, Louisiana; Eric Garner in New York City; Daniel Shaver in Mesa, Arizona; Philando Castile in St. Anthony, Minnesota. The list continues to grow. So what explains the proliferation of tragic instances of civilian deaths at the hands of law enforcement?

One explanation comes to us from deterrence theory. It holds that, in situations where we deliberate over whether to break a law, we make simple, rational calculations. Laws that create fines or penalties, such as license suspensions for speeding violations, are designed to deter people from breaking the law. In John Locke’s terms, the threat of having our liberties stripped away or property seized forces us to consider the long-term consequences of our actions, which may prevent us from committing crimes. This theory has its roots in the works of Thomas Hobbes and other philosophers and is one of the cornerstones of the social contract.

In instances of police misconduct, deterrence theory may play out this way: An offending officer may learn that his fellow officers have not been disciplined for acting as he has. Perhaps he is aware that there are protections, whether legal or union contract-based, that eliminate most of the risk of consequences for certain behaviors, thereby cementing the notion that this style of policing is not wrong but rather is necessary to get the job done. He may consciously, or more likely subconsciously, ask himself why he should alter his behavior if there are no consequences.

So deterrence theory at least partly explains law-breaking behavior. But did the officer who shot and killed Stephon Clark break the law? To help answer this, we can look to the Supreme Court.

In Tennessee v. Garner (1985), the Court held that use of deadly force is not permitted against unarmed civilians fleeing the scene of a crime. However, Garner does establish an exception to this general rule in what is often referred to as the “fleeing felon doctrine”—a misnomer because the suspect does not need to be a felon. The Garner exception holds that deadly force may be used if “the officer has probable cause to believe that the suspect poses a significant threat of death or serious physical injury to the officers or others.” This is important because the recent independent autopsy administered by Dr. Bennet Omalu, the famed pathologist who shed light on the health consequences of concussions in the NFL, demonstrates that six of the eight bullets entered Clark’s back side. (The subsequent autopsy by the coroner’s office contradicted these findings somewhat.) The prohibition against shooting fleeing, unarmed civilians seems to be damning for the officers who shot Clark. But the “fleeing felon” exception makes it far less clear that the officers broke the law.

Such ambiguities help explain police misconduct through the lens of deterrence theory. According to the theory, when an individual is deciding whether or not to commit a wrongful act, three variables help tip the scales: the certainty of prosecution, the severity of punishment, and the swiftness of judgment. Vague exceptions to prohibitions of undue violence speak to the first variable: They suggest that there is little certainty of prosecution or punishment for officers who shoot civilians.

Whether or not the officers involved in the Clark shooting are guilty of breaking the law, the lack of deterrence seems to have encouraged an uptick in police violence in departments with long histories of controversial incidents. Very recent research suggests that police union privileges significantly increase the rate of officer misconduct. This dynamic is at play even in departments with a relatively uncheckered history, such as Sacramento’s. The clichéd “blue wall of silence” is as real there as it is in Chicago, and its role in preventing effective deterrence helps explain the rise in instances of police violence across the country.

Deterrence and Plato’s Ring

In order for deterrence to be effective, the negative consequences of an action must be greater than the benefits of going through with it. Punishment, after all, inflicts significant hardship on wrongdoers in part in order to communicate that society condemns these sorts of actions.

But what of cases where penalty and punishment don’t loom large, where one can preserve a semblance of respectability after wrongdoing? The “Ring of Gyges” story in Plato’s Republic argues that people are not willing followers of justice, and that morality is wholly conventional—a sucker’s game. Individuals calculate the maximum utility gained from breaking the law. Thus, when given certain protections or immunities, even the just man will behave unjustly, and police officers are more likely to act on their implicit biases if they have extra protections and recognize that they will likely not face consequences for errors or bad behavior.

We can apply the same deterrence theory framework to almost any environment governed by rules and regulations. In the workplace, for example, employees tend to take longer lunch breaks or feel at ease misbehaving at work if they know management will not impose certain, serious, or swift punishments, such as suspension or termination. From a manager’s perspective, such outcomes are undesirable. An institution cannot grant its employees such leniency without leading to micro-inefficiencies that damage the institution’s sustainability.

On balance, termination, discipline, and financial penalties are all appropriate consequences for bad behavior at the workplace, or even for mere poor performance. And this deterrence framework should be applied even more forcefully, it would seem, to occupations where misconduct or poor performance can lead to death or serious injury. Mistakes or negligence by a medical professional, for example, can harm or kill a patient. Doctors and nurses thus face the threat of losing not only their jobs and livelihoods (through revocation of their license), but their freedom as well, in cases of criminal negligence. Imagine if doctors had special protections against these kinds of punishments. Would it give you more or less faith in your doctors? And should others engaged in professions where mistakes can end lives have special job protections?

An Exemption in the Law

Many in the law enforcement community—more specifically, union leaders—argue that they should have such protections. They say the nature of the occupation warrants special treatment. And we cannot dismiss this perspective out of hand; at times officers do need to make split-second decisions that have life and death consequences for themselves and others.

But even if we concede that police should have some level of protection or forbearance for mistakes, the question remains: How many accusations of misconduct should it take to lead to a disciplinary measure? A data point from the Invisible Institute’s “Citizen Police Data Project” highlights a case of a Chicago police officer who received 61 civilian complaints, of which only two resulted in disciplinary measures. To be sure, there are limitations to extrapolating general rules from an extreme example; nevertheless, it’s hard to imagine a fast-food employee keeping a job after more than a handful complaints from customers, let alone complaints as serious as those leveled at the Chicago officer.

If there are so many protections and few serious or timely consequences for misconduct, then how exactly are we to deter officers from continuing to engage in misconduct? Under conditions in which repeated misconduct allegations are ignored, it should not be surprising when, as deterrence theory suggests, many officers—even ones with relatively fewer infractions—succumb to the temptation to commit wrongful acts.

Eric Garner’s death raises this very concern. Garner, an unarmed black male, was placed in a fatal chokehold by a white officer. The chokehold went beyond the types of force that officers are supposed to use—for precisely the reasons Garner’s death illustrates. The officer in Garner’s case had racked up a long history of complaints before the tragic incident—14 to be precise. The complaints involved a substantiated abusive stop-and-frisk incident, an incident in which the officer was accused of refusing to get a person medical attention, and another incident in which the officer allegedly slammed an individual against an inanimate object. Following Garner’s death, the officer was subsequently placed on desk duty and received a $20,000 pay raise in the year following the incident. Only 5 percent of New York Police Department officers have received more than seven complaints, which begs the question of what an officer needs to do to be seriously disciplined or terminated. These facts surely do not go unnoticed by the taxpayers of New York City, who just paid out $5.9 million to the Garner family for his wrongful death.

Like any profession, there will always be a few “bad apples.” There are probably no consequences swift, sure, and severe enough to deter some individuals who join the force with a twisted moral compass and harmful biases. But if the vast majority of officers who are amenable to such consequences feel that they are protected by a union or the “blue wall,” then they will break the rules from time to time, and civilians may be harmed or killed. Criminologist Samuel Walker argues that these sorts of police protections create a “culture of impunity” that, along with other systemic issues, undermines our policing institutions.

We should not forget that there needs to be active engagement between the community and its police forces. Police officers exist to shield the community from violence. Countless honorable police officers risk their lives daily to fight violent gangs and drug dealers. While the vast majority of officers are good, hard-working Americans, a minority are living embodiments of the systemic problems in America’s policing apparatus. These “bad apple” officers have a contagious influence on their fellow officers—one of the many reasons why misconduct and lack of accountability have become more widespread. After all, the full aphorism isn’t just “a few bad apples” but rather “a few bad apples spoil the barrel.” Thus the problem is about much more than just a handful of harmful officers; it’s about how those few officers corrupt the entire culture of policing in a community.

A Fairer Comparison

We should be cautious in comparing policing to private-sector occupations. There is no question that policing is a unique profession. Routinely throughout this country, officers are yelled at, assaulted and, tragically, killed. Lately, they have been scorned for adopting military equipment and tactics—what some call “militarization”—and accused of unjustified shootings and other instances of brutality. Many officers feel like they are at war with their job, the people they project, and the people they arrest.

Luckily, there is another, similar profession we can look to that is in many ways even more dangerous and more demanding on its practitioners: military service. Interestingly, the so-called blue wall is almost non-existent in the military, largely due to the cultural values that the military has adopted. In the armed forces, mission comes first. While there have been exceptions to this rule (allegations of sexual assault and cover-ups, for instance), in general, when someone in the military is caught doing something wrong, it leads to punishment. Being stripped of rank, dishonorably discharged, or worse, sent to Leavenworth Penitentiary, is not an unheard of proposition for those caught misbehaving in the military.

This is not the case for police officers, who receive more protections, from more sources (union contracts, state laws, the courts, and public opinion), that make it difficult to take action against them. Even when a police officer is fired for misconduct, all too often he finds himself back in blue in a neighboring jurisdiction. One report found that 451 officers fired from 37 of America’s largest cities were rehired or reinstated as police officers or hired by another department because of union intervention and legislative obstacles. One former police commissioner in both Philadelphia and Washington, DC, Charles Ramsey, summed up the process by noting that he had to rehire 80 fired officers, three of them twice. “It’s demoralizing” to the force, he says.

The lack of accountability among police ranks keeps departments from doing their jobs. So much energy is spent defending the department and, in many cases, bad officers, that the complaints of those whom they are sworn to protect are ignored. Police must acknowledge the trust gap that exists between them and many of the people they serve. The first step in closing that gap has to be engaging in real self-reflection, much as the military did in the 1970s in the wake of the Vietnam War.

In addition, police would benefit greatly from the emphasis that the military has placed on building a professional body. This is certainly not to say that police officers are not professionals, but rather that they should treat themselves more like professionals. For instance, most police forces have promotion systems that are linked to and resemble the civil service, as opposed to the military’s performance-based system. For service members, virtually every advancement to every major rank is associated with some type of professional development school, from E-5 (sergeant in the Army) all the way to O-6 officers (Army colonel or Navy captain). And those O-6 officers must attend the War College before they are considered for promotion to general or admiral. Many senior police leaders, by contrast, have a little more professional training than a rookie fresh out of the academy. While certainly there is some ad hoc professional development in police work, the majority of training is tactical—learning how to get the shot group tighter—not on developing the maturity to determine when deadly force is appropriate. Not enough time is dedicated to the profession of policing or on how to lead other officers.

It is important not to take the police-military comparison too far. Police departments have a different relationship to the communities in which they operate than does the military. Police departments’ efforts to model themselves after military units by acquiring their weapons and some aspects of their training have damaged police-community relations. Between 2006 and 2014, law enforcement agencies acquired more than 6,000 mine-resistant, ambush-protected vehicles (MRAPs), 79,288 assault rifles, 205 grenade launchers, 11,959 bayonets, 50 airplanes, 422 helicopters, and $3.6 million in camouflage from the Department of Defense’s 1033 Program. One report found a correlation between the militarization of law enforcement agencies and the number of citizens killed each year by police. There must be limits, then, on the extent to which police departments imitate the military. Still, the police and military are similar enough that there is much the police can learn from how the military overcame similar challenges.

The Limitations of Deterrence Theory

While deterrence theory is a useful tool in explaining the uptick in violent interactions between police and the communities they serve, it cannot explain everything. It would be a big mistake to overlook the human cognitive biases that undoubtedly impact officers’ decision-making in high-stress scenarios. Race, fear, and general human error inevitably factor into police-citizen violence.

Deterrence only works to the extent that a given individual is deterrable—which is to say that a person’s responsiveness to the threat of sanctions exists at some non-trivial level. Recent literature in the field of criminology suggests that individuals can be broken up into multiple levels of deterrability: acute conformists, who “comply with the law not because they fear formal sanctions, but because they believe that conformity is the right and moral thing to do;” incorrigible offenders, for whom “increases in the perceived certainty and/or severity of sanction threats mean little;” and, somewhere in the middle, deterrable offenders constituting “a group that includes anyone for whom a sanction threat is potentially influential.” Under this framing, deterrence works better when a person is relatively risk-sensitive and not naturally inclined to misconduct.

Environmental factors attending emotionally intense encounters clearly play a role in the effectiveness of deterrence. Of these, three seem most important: the role of race, the proliferation of firearms in the general population, and the power of fear—from whatever source—to drive human error.

The Role of Race: To acknowledge that race may have played a role in the shooting of Stephon Clark is not to charge the police with racism—or even these particular police officers. Research suggests that police misconduct and the excessive use-of-force are linked to cities with larger densities of black and Hispanic groups.

Although Sacramento is a racially diverse city (nearly 14 percent of its residents are Black and 26 percent are Hispanic), Clark was killed in an area that does not mirror the city’s overall racial composition. The neighborhood in which the incident took place, Meadowview, is nearly 20 percent black, 30 percent Hispanic, and 15 percent white. The surrounding area has a history of residential segregation and the forced displacement of minorities into Meadowview. In short, Meadowview is predominantly a nonwhite area and has experienced problems with crime and development in the past.

While the 911 caller was not explicit on the race of the alleged subject—“He had a hoodie on, I couldn’t tell”—the two experienced officers entering the area were surely aware of the neighborhood’s racial dynamics. This may have led them to presume that Clark was dangerous, despite the reports that someone with a “toolbar” was breaking windows. After confronting Clark and approaching him in his grandmother’s backyard, the officers saw a mobile phone in his hand. Given the context—the predominantly nonwhite neighborhood, the history of crime, the alleged behavior, and presence of a “toolbar” —they mistook the innocuous phone for a dangerous weapon.

This cognitive transformation was not accomplished through comprehensive police work. Rather, it was the manifestation of how the officers perceived the residents in the neighborhoods, in combination with their knowledge of, and preconceived notions about, the suspect. These notions were intimately tied with race. As Bryan Stevenson writes, “people of color in the United States, particularly young black men, are burdened with a presumption of guilt and dangerousness. . . . We are burdened by our history of racial injustice in ways that shape the way we think, act, and enforce the law.” In short, the context of the call and the officers’ likely presumption of guilt led them to assume that Clark was a dangerous individual.

Firearms in the General Population: The sheer volume of guns in America represents a significant environmental factor influencing the rational decision-making of police officers. Research has shown that the mere sight of a firearm induces greater aggression and suggests that gun prevalence can at least partially explain police misconduct. In the rational actor model of deterrence, individuals will comply with the law and avoid sanctions when they believe the pain of potential punishment will outweigh the benefits of breaking the law. But the presence of a gun takes an officer out of any mode of rational calculus and tends strongly to elicit violent self-preserving instincts. Still, in the virtual absence of deterrence, establishing credible threats of punishment for misconduct could do more to reduce police violence than perhaps any other policy solution.

Fear and Human Error: Some may argue that the Sacramento officers were driven to shoot first out of fear. Betty Shelby, the police officer who shot and killed Terence Crutcher in Tulsa, reflected this sentiment best when she stated flatly that she “would rather be judged by 12 than carried by six.” In other words, she would rather shoot and kill the suspect out of fear for her own life—even if it turned out the suspect was innocent.

It might be unpopular to say so, but police officers do not have the authority to kill civilians simply because they are afraid. Being a cop is a dangerous job, no doubt. But that job necessarily involves placing oneself in life-threatening situations and following procedure to make the right call, regardless of fear.

Of course, when officers are facing an individual with a deadly weapon, who is capable of causing them or others immediate harm, they do not have the luxury of running through de-escalation techniques; the use-of-force continuum quickly jumps to deadly force. And certainly, an officer should not be expected to sit idly by while someone takes a swing at them with a crowbar, which is certainly a deadly weapon. Indeed, officers have every right to go back to their families at the end of their shifts, just as Stephon Clark had the right to go back home to his family.

However, officers must serve all members of the community that they have sworn to protect and serve, including those suspected of wrongdoing. “Suspected” does not translate to guilty—as guaranteed by the Fifth Amendment: “No person shall . . . be deprived of life, liberty, or property, without due process of law.” In a democratic society, the policing apparatus is the enforcement agency upon which citizens principally rely for protection, not the one which plays judge, jury, and executioner against citizens who they are ostensibly obligated to protect.

Still, the fact that fear and the frequency of high stress/high stakes scenarios are a staple of policing implies that deterrence theory cannot fully explain police misconduct, and that, therefore, more effective deterrence may not be enough to reduce police violence. Put differently, even though officers should operate rationally, they will inevitably fail to make rational calculations in situations where their lives are on the line, at least from time to time. As previously noted, deterrence relies on individuals who are, at least to some extent, deterrable, but who is and who is not deterrable can vary according to circumstances.

Indeed, racial bias, variations in officers’ sensitivities to risk, the prevalence of civilian firearms, and other environmental factors will surely mitigate the effectiveness of the threat of sanctions for misconduct. But the argument here is not that stronger deterrence can solve the problem entirely, but that establishing clear and swift punishments for misbehavior can substantially reduce violent encounters between law enforcement and communities. Deterrence is not a silver bullet, but it represents perhaps the best opportunity for reform, in combination with better training in deescalation and implicit racial bias, hiring emotionally intelligent applicants, and offering robust educational and professional development opportunities for officers.

The “culture of impunity” and “blue wall” that make it difficult, if not impossible, to get bad cops out of the profession are the direct result of policymakers’ decisions to insulate public employees from scrutiny and accountability. To quote Radley Balko, author of Rise of the Warrior Cop, reforms are not “anti-cop;” they’re anti-politician:

The following reform proposals aim to rollback bad policy and establish effective deterrence, allowing the many good cops to thrive.

Rollback “Law Enforcement Officers’ Bill of Rights” (LEOBR) laws. Maryland became the pioneer state for LEOBRs when it passed the first such law in 1972 following police union lobbying efforts. LEOBRs add substantial due process protections for officers being investigated for abuse of power. Their exact provisions vary by state, but there are several common privileges. For instance, there are “cooling off” periods in which an officer can abstain from responding to any questions once a complaint has been lodged. Interrogations must be conducted at a “reasonable hour,” with a union representative present. There can be as many rest periods in interrogations as the accused officer wishes. And if an officer is threatened with disciplinary action during the investigation, whatever he says after that threat may not be used against him. Further, departments cannot publicly acknowledge that the officer is being investigated or disclose the nature of the complaint, and the officer cannot be questioned by “non-government agents,” a provision which precludes civilian review boards. While total transparency during these investigations may not be in either the officer or the public’s best interest, and while some layers of protection may be necessary to shield officers from frivolous complaints, reasonable people can agree that these protections are much more than is necessary to protect the rights of the accused. At the very least, civilian review boards should be allowed to operate in states with LEOBRs.

Limit police union contracts that erase misconduct records, require cities and taxpayers to pay for misconduct, and limit oversight and discipline of officers. Police unions exert tremendous influence over lawmakers’ decisions regarding policing policy in city and state governments across the country. One study analyzed police union contracts in 81 of America’s 100 largest cities and found that more than 40 allow for misconduct records to be erased after a certain period of time if the officer requests or is exonerated. Many cities’ contracts also require taxpayers to cover misconduct settlements and the full cost of legal defense, prevent civilian oversight boards from administering punishments, and a host of other cushy protections for officers accused or found guilty of misconduct. Such protections make it exceedingly expensive and difficult for cities and departments to purge violent officers from their ranks, or they allow “bad cops” to simply move on to a new department after dismissal. Unsurprisingly, researchers at the University of Chicago found that affording police unions collective-bargaining rights increases the risk of misconduct.

Remove caps on intelligence scores among applicants, and encourage hiring candidates with higher general and emotional intelligence scores. Deterrence works even better when the population in question is prone to good behavior. Police departments, then, can decrease the likelihood of violent relations with communities by hiring officers with higher “general” and “emotional” IQs. Some police departments set caps on acceptable general intelligence test scores to reduce turnover, fearing smarter candidates will become bored with day-to-day policing. This hiring practice also limits the “emotional intelligence” of the applicant pool—the applicants’ ability to perceive, use, manage, and understand their emotions and the emotions of those around them. Those with lower emotional intelligence scores may be more prone to violence, according to some research.

Increase opportunities for professionalization. Departments have generally not considered the need for more professional development for their officers. The consensus across many departments has been that “street smarts” are more important than “book smarts” to the job of policing. Indeed, in many cases officers look down on their more educated peers. Departments should adopt military performance-based systems, where professional development in a classroom setting factors into advancement through the ranks.

Encourage officers to pursue higher education. Finally, departments should capitalize on the prevalence of online training and education to encourage their officers to seek college degrees. A study conducted by Michigan State University found that college-educated officers had a lower propensity for violence, and a Florida report found that 75 percent of all disciplinary actions were leveled at officers who have only a high school diploma.

In The Republic, Plato makes the case that justice is a good in and of itself, and preferable to injustice on those terms alone. Yet Plato also teaches that one needs a noble soul to realize the truth of this. Unfortunately, the world we live in is not filled exclusively with people with noble souls. All people have biases, some more serious and dangerous than others. Consequently, wrongful acts will continue to tempt, and if it were possible to be assured of success in a wrongful act, the vast majority would indulge in them at least sometimes.

When it comes to police, who at times have the power over life and death, emphasizing protections at the expense of deterrence is especially dangerous. This is true because these protections have to, on some level, influence the rational calculations of some officers.

To what extent can a failure of deterrence explain the shooting in Sacramento? Did a bad apple ruin the bunch, did entrenched union protections subconsciously allow the officers to neglect de-escalation tactics, or were their motivations influenced by a combination of racial bias and the likelihood that they would not face swift punishment for their actions? The answer is not readily available. Nevertheless, it is reasonable to conclude that the absence of effective deterrence explains a substantial proportion of police misconduct in departments across the country, and in all likelihood it played a role in the Stephon Clark shooting as well.

While the lack of effective deterrence is not the only reason for police misconduct, it is certainly a major one. Humans are neither entirely good nor entirely bad, but they will weigh the possibility of having their liberties stripped against the benefits of breaking the law.

The immense power police officers wield, once mishandled and corrupted, poses one of the greatest threats to our institutions of law and order. Misuse of power has the potential to cause not only a breakdown of the police force but of our republic as well, which is why we must do more to ensure that the badge does not become a shield from injustice—a Ring of Gyges.